5 practical and simple tools for digital marketing professionals

My previous posts on smart tools for digital marketing professionals are among the most read in the blog. Since you seem to enjoy it, I’ve listed five more smart digital marketing tools for you to try.

Tailwind – Smarter Pinterest marketing

Tailwind is a nifty tool for visual marketing, in other words: Pinterest. A platform quickly becoming increasingly relevant for marketers.

With Facebook losing its position other platforms are gaining importance for content creators and influencers. Pinterest is one of those platforms. Over the next year, we will probably see more types of businesses on Pinterest, using it to drive traffic or sell products directly. Pinterest is no longer a platform only for food photos, pretty shoes and interior design (although, categories like those still consists of a large share).

But Pinterest hasn’t been a priority for most social media tools. And if you need to do everything manually; naturally, you won’t prioritise it. Therefore, using Pinterest to create value for your business has been hard. Tailwind is bridging this gap.

The tool has four main features:

  1. Tailwind makes it possible to schedule pins and spread out what you usually pin all at once over days or weeks
  2. It can create loops of evergreen content, taking the oldest pins on your board and pinning them back on top
  3. Additionally, it lets you participate in and create tribes, where users collaborate with each other to get their content more exposure on the platform
  4. It gives you easy access to analysing your pins, making sure you learn and improve over time

Tailwind sure ain’t the prettiest tool I’ve seen. But it’s functional, and it’s a great way to step up (or start) a Pinterest presence.

PixelMe – Smarter UTM-tagging

PixelMe is a URL-shortener that comes both with simple built-in UTM-tagging but most of all, with retargeting possibilities added to every single link you create. The short link collects enough info about the person who clicks for you to be able to retarget them on any platform. (You should make sure you use this tool in a way that goes with GDPR though.)

Additionally, they’ve realised that the process of UTM-tagging is a hassle for many marketers today, and after tagging, you often want to shorten the links to look nice. Pixelme has created a way to make this process much more straightforward and reduced the risk of beginner errors.

Hemingway App – Improve your writing

Most people spend a lot of time writing online today. Hemingway App is a neat little writing app that highlights lengthy, complex sentences and errors in your text.

Hemingway App will help you find words you can swap with simpler ones, or sentences that are too complex and need more straightforward language. Additionally, it will help you find both weak adverbs and passive voice, to help improve your text.

You can paste in something you’re working on and edit away or compose something from scratch. There’s also a desktop app if you fall deeply in love.

Hunter.io – Find the e-mail address you need

Have you ever wanted to contact someone by e-mail instead of via LinkedIn without knowing the correct address? Hunter is a neat (and perhaps scary) little tool that collects e-mail addresses and makes them searchable. So, if you know the correct e-mail address domain (for instance amazon.com or wholefoods.com) you can most of the time find the full address with Hunter.io’s search engine.

Sure, it might feel like an invasion of privacy. But every single email address Hunter.io collect and distribute in their Domain Search have a public source that they disclaim, along with its discovery dates. So, they’ve just done the stalking for you.

And well, it comes in handy very often.

Front – A smarter inbox for all your messages

Front is a smart inbox for teams that let you collaborate with your colleagues. But it’s not only e-mail. Front makes it possible to take shared responsibility for Facebook Page messages, Twitter messages, website chats and forms, Intercom support messages and so much more. You can even build your own integrations if you want to.

With Front, you’ll have all your messages, and all your teammates, in one place. Someone will always be there to reply – when UPS have lost yet another package or your servers are taking a break – without having to jump from one tool to the next. You can assign messages to the right people, collaborate on drafts and loop in reinforcement when you have to. It’s also possible to create advanced rules and canned responses for automatically taking care of your most common types of emails.

It might sound like a small win. But I don’t even have a team to collaborate with yet, and I still benefit from getting everything in one place. And as soon as you start to add a couple of active Facebook pages with messages turned on, and a shared e-mail address or two the number of places to keep track of new messages increases drastically and so does the number of messages that you miss.

If you can relate, try it.

Looking for more tools for digital marketing professionals?

Here are some of my other posts on the topic:

How great data make great stories

Thousands of years ago, when we lived on the savannah, we were all data analysts. A lion passed by, we analysed the situation and killed the lion or tried to run very fast. We found berries, scrutinised them before eating (and died or survived).

The need to be analytical is not new. It’s always been important, even long before the digital big data era.

Being able to analyse data is an important skill set because we create our models of the world based on the results we see when we experiment. If we can adequately examine our data, our models will better represent the real world.

But data is not enough.

To be an excellent berry investigator, data analyst, or astronomist, you need two other skills: 1. You need to communicate your findings to others (“Last night I found a planet I’ve never seen before, you should look at it too”), and 2. You need to convince them that your conclusions are correct (“Finding a new planet means that our solar system consists of at least twelve planets”).

These two steps might be familiar, maybe you know them as storytelling…

Storytelling is a tool so powerful that we sometimes get moved by something entirely fictive or fake. Because storytellers have no duty to base its stories on facts. Instead, it often becomes the consumers’ role to find and analyse the data, to see the errors in the original analysis and adjust their worlds accordingly.

But, the relationship between storytelling and data is most often a fruitful one, one where they make each other better. The result is usually best when both get to shine.

The balancing act between data analysis and storytelling is continuous. It impacts every slideshow ever made, every ad ever cut. Too much data makes the story halt. To little data makes the story lose its power.

Very few data analysts masters the art of presenting data. Also, storytellers and analysts are not known for their vivid collaborations. But when we start to see our data reports as storytelling, our perspective shifts.

As someone with a somewhat rational logical brain, I sometimes have a hard time with things that removes focus from the fact I want to communicate. Hence, I present the point I want to make and most often, nothing more.

Data is boring

But turning my data analysis into storytelling is not for me – it’s for you, the recipient. Because data is boring ( – and beautiful if you’re anything like me, but most people aren’t). So, I sometimes lose the people who’ve got a hard time putting the data I display into context. And if I want to make sure people listen to what I have to say, I need to move it into a narrative or framework that will help them understand what I’m trying to say.

So, what’s the secret to moving from boring data presentations into something more engaging? Creating great stories based on data is a little bit like building a house. You start with the foundation, add on the framework, continue with the roof, and keep going until you have some final steps of painting and decorating.

Here’s how I’m trying to go about it:

Step 1 – Find you why

I write down my own “why”. Sometimes this is: “I believe this client is better than it’s competitors”. Or, “I believe I can learn something interesting by looking at data from racist Facebook groups”. Most often it’s more in line with “I need to figure out how we’re doing so that I can make the necessary adjustments.”

Step 2 – Collect your data

I collect my data points in a very unstructured form. A lot of my data analysis is exploratory, and I believe that’s fine. I don’t want to already have a storyline in mind when looking at what’s in front of me. I just collect them, often as bullet points and with visualisations from Tableau in a Dropbox folder.

Step 3 – Organise your data

I organise my data points in a straight line. The findings that support or are close to each other should be presented either together or following each other. I also want to find a natural way to introduce the story and a natural way to end it. I often start with broader strokes, move further into detail later, and move back out again to summarise.

Step 4 – Map out a storyline

Starting to get a storyline in place, I translate my findings into a narrative. Here I’m making sure to use “everyday language” setting the context to the real world.

Finding a story and making a coherent argument is as relevant for an audience with no domain or data knowledge as it is when presenting for experts. However, in a domain specific context, they will want to know more about particular tests, definitions of variables and other things that I leave out for most audiences.

This step is often where I make that differentiation. Putting an analysis into context means two very different things for a group of domain experts and a group of school children. Here’s a great example of how to change your story based on your audience. So, knowing your audience is always crucial. So, knowing your audience is always crucial.

Step 5 – Start to think about visualisations

I decide what visualisations that are needed to communicate my argument well. Sometimes it’s with a chart, but more often, it’s not. Most people I work with are not data literate, and if I show them a data visualisation, they tend to get very stressed.

If I decide to use a chart, there are some things I have in mind:

  • I. The visualisation should be able to stand on its own entirely. So, it needs a title, and both the axis and the included data need to have names.
  • II. Categories should be simple to understand and make sense to anyone
  • III. The colours used should make sense and be coherent. Not in a “pink for women, blue for men” type of way, but keep your colours throughout your story and make sure they are easy to separate and remember. Remember that a lot of people are colour blind, so green and red is a bad combo.
  • IV. Make sure the chart is possible to decode in the format you will present it. Make sure text is big enough to read for instance in a video or on a screen

Step 6 – Make your story come to life

When I have a storyline in its raw form, I try to find ways to make it come to life. This is by far the hardest step for me. How can I invite the audience to feel, or to use their imagination so that they are well primed for the argument I’m going to make? What examples would make people care?

This is also the part where I most often bounce ideas around and ask for help. I’m slowly starting to realise that using a story to introduce and end an argument help people care. It’s my responsibility to make them care.

Using cases is also an effective way to make a story come to life. It’s so much easier to understand the problems of a small business owner when they are presented as the daily struggles for a 55-year-old widow owning a petrol station instead of 15 statistics. Building a narrative around one or more data points are not tampering with the results, it’s a smart way to make the result stick.

Be careful with metaphors. Most people are misusing them, forcing a metaphor to work when it’s not even close to telling the same story as you’re trying to communicate. When they work, it’s a fantastic storytelling tool. But you really need to think twice before you use them.

Step 7 – Connect back to where you started

End by connecting back to where you started. Either by tying it to the initial story or by summarising it all neatly. Zoom out and put your argument into a broader context.

Step 8 – Test!

Test it on both your audience and on people who know the underlying data. Data analysis and storytelling is not an either-or relationship, they need to coexist. If your storytelling is making your data analysis incorrect, (for instance by using metaphors that don’t capture your argument well), you need to go back and change it. So you’ll need to double check that the human-friendly version is perfectly aligned with the data analysis you started out with.

In short

This was a very long guide to how I combine data and storytelling. Sometimes I start with the data, sometimes I begin with the message. You can attack as well from either end.

And while it often sounds like data and storytelling are far apart – and we usually give people responsibility for one but not the other – we should try to push them together. Sometimes, we just need to educate ourselves a bit before it feels comfortable to work closely with tools that you’ve not used before.

We need both. Most stories need data, and most data need storytelling to get comprehensible by a larger mass. And even if you decide not to learn both skills yourself, you’ll at least need to learn how to collaborate with someone who knows what you don’t know.

How to make more people care about what you do by hacking the brain

It’s late afternoon and the warmest day of July. The air is stuffy and standing completely still. A woman in her fifties is walking past her favourite gelateria and decides to buy two big scoops of strawberry ice cream. Slowly she realises something is wrong – they only have vanilla.

After walking over to the store next door, vanilla is all she can find in the ice cream section. And further down the street the nice man with the ice cream cart happily tells her vanilla is the only flavour he’s selling now. Is this a bad dream or the rest of her life?


Imagine if you were only allowed vanilla ice cream for the rest of your life. Forget about chocolate, pistachio and raspberries. You can get how many scoops you like – every single one of them will taste vanilla. Your interest in ice cream would probably decrease. Sure, vanilla lovers would be in heaven, but the majority would be moderately interested in ice cream after a while.

Vanilla is Index 100 in the ice cream world. Always available as a reference value for all other flavours, a standard. Being the ice cream norm makes it unengaging. We are not very fond of the neutral – the neutral is not exciting. Few would argue that their absolute dream car is a Honda. Or think Billy’s bookshelf is the best-looking piece of furniture available.

What makes us engaged?

Feelings have always been the secret sauce of engagement – we need to feel to care. Most marketers and click bate news outlets are talking about sensations and emotions in every second sentence. But the truth is: engagement doesn’t have to come from conscious feelings. We can game a large part of how someone feels about something by understanding the brain.

While part of how we think about the world is a conscious or semi-conscious thought process – “she was nice” or “cute dog” would probably both fall into this category – a lot of how the mental processes about how we perceive the world is unconscious. Also, our brain is sluggish. We are taking every single shortcut we can find, saving lots of energy over time. 

One of the shortcuts we are using to understand the world is associations. 

Concepts with shared attributes activate partly the same parts of our brains, and therefore we unconsciously link them together. This concept is called “networks of association”. Since I’m not a neuroscientist (and we don’t have all day), I won’t try to explain this further. Instead, you should read some of Drew Westen’s work in political psychology to get more insight.

The important things to understand before you keep reading are:

  1. Our brain works partly by associating things with one another
  2. This is an unconscious process

Why are networks of associations relevant in marketing?

Getting people to feel used to be about making them upset or happy or sad. Now you might start to realise there’s another very effective way. You can make your audience feel in line with your goals without them even knowing – only because of how you package your message. If you create your message with the brain in mind, you can piggyback on neuroscience to get people engaged.

Packaging a message is more than just the wrapping paper (even though everyone knows that the outside is at least as important as the inside when it comes to gifts) on a pair of socks or the big white box that arrives from Net-A-Porter. It includes everything from the colour on your dress at an important meeting to product names. Packaging is everything that makes you perceive something a certain way.

  • Compare how you feel about the environment with your reactions towards the air we breathe and the water we drink. Any difference?
  • Or compare your thoughts about the unemployed with your thoughts about the people who’ve lost their jobs through no fault of their own. Any difference?

The two cases above are good examples of how the packaging of “the same” message or idea can differ significantly in how it makes us feel.

To get the outcome you want, you need to actively control the associations you activate in the brains of your audience. And this isn’t just about pairing things with known positives. Some things that you might think are neutral objects have positive associations, and others have negative.

The nine most common associations to “immigrants”, when tested among Americans*, consists of five positive and four negative connections:

  • + Opportunity
  • + Better Life
  • + Nation of Immigrants
  • + Hard Working
  • + American Dream
  • – Law Breakers
  • – Government Benefits
  • – Don’t speak English
  • – Don’t pay taxes

*Westen & Greenberg, Quinlan, Rosner, 2008 Handbook for Progressive Messaging

Some cases where the network of association matter

The fact that someone named “network neutrality” and tried to get people excited is, therefore, a little bit sad. Few people will be enthusiastic by the neutral. When we talk about network neutrality, the patterns activated in our neural networks are very similar to when we are talking about vanilla ice cream or a boring car. It’s the same with something like gay rights, no one gets excited by “rights”.

Instead, some brilliant packing of activism is the concept of “pro-life” created by the American conservatives. The term “pro-life” activates the same, or similar, neural networks in our brains as when we think about positive stuff. This activation has nothing to do with our feelings towards abortion (yet), but because the term “pro-life” creates positive associations in our brains. Few sane human beings want to be “anti-life”. (Therefore, the opponents have started to rebrand this group as “anti-choice” or “anti-free choice”).

The packaging above becomes even more interesting because the groups that are pro-life are also the groups advocating weapons – a tool commonly used to end lives. Instead, they are using freedom as their main argument. Because who wants to be against freedom?

Being pro-life and pro-gun are positions are rhetorically incompatible, something American conservatives ignore. The naming is purely communicative, and most Americans don’t give this a single thought.

Republicans are brilliant at packing politics in a way that makes people’s minds instinctively feel the way they want. Did you know that Obama Care and The Affordable Care Act are the same? The first name is the packaging by the Republicans (to make sure people don’t fancy it), and the other one is the “correct” name. Interestingly, politicians talk very little about what this initiative could be about: “a family doctor for everyone”.

There are two lessons to learn in this – one about net-neutrality and one about engagement.

Lesson 1. When we talk about network neutrality, we must stop talking about the importance of a neutral internet, and start talking about the importance of the free internet. Because that is the core idea in net-neutrality. (Your internet service provider should not be allowed to choose which traffic is prioritised, or charge different rates on traffic to different websites.)

Lesson 2. What we call things do matter. This statement is not just about good copywriting – it’s about good brainwashing. You impact someone’s feelings towards an object or idea purely by how you name it you will undoubtedly have an advantage when people then start to think about what you are saying.

If you want to get people excited, don’t talk about it with vanilla ice-cream language. If your goal is to make people angry, name it and talk about it in words that are close to anger in the neural networks of the audience. The same goes for happiness, fear and every other feeling.

How do you find out what the networks of your audience look like?

I usually start by looking at social media behaviour. Tweets, Instagram hashtags, Reddit content and Facebook groups are a variety of sources to mine. If I’m lucky, I can add focus groups to the mix, having the opportunity to listen to people talk about a topic uncovers a lot of the underlying emotions and associations.

The next step is trial and error. Always spend time analysing the results from your messaging. Did people react as you intended? What was the response? Do this over and over again, and you will become a messaging machine.

And once again. This is about so much more than good copywriting. Although a copywriter with analytical skill will probably have an advantage.

Finally

Don’t serve vanilla. And if vanilla is the only ice-cream you’ve got, at least add some topping.

13 common data mistakes you should learn to avoid

You might have heard most of these terms before. Maybe you nodded in assent when someone claimed one of them must be involved, even though you had no clue what it meat. If that’s the case, this text is for you.

I’ve collected 13 common data mistakes or data fallacies that we all make at times. I will explain both the background and I’ll give you some real-life examples. Because I know you are working hard on getting more comfortable working with data. And you shouldn’t have to make these data mistakes yourself or get fooled by someone else making them – if you don’t want to.

1. Regression towards the mean

Regression towards the mean happens when luck plays a role. But all luck and unluck, will even out with time and most of the events end up close to average over time.

An example of Regression towards the mean

Think of long jumping. Sometimes a strong wind against the athlete will lead to even the best jumper showing poor results. Similarly, a strong wind can “help” a mediocre athlete creating a remarkable (but temporary) bump in her results. Both these effects will disappear when the conditions change, and the results will move back to “normal”. So with more jumps, the results will tend to move towards an average.

Therefore, if you run an experiment over and over again, and there’s a component in luck (and trust me, that’s always the case) most of your results will be “average” over time.

2. Cherry Picking

“Cherry picking” is when you choose to communicate only the results that confirm a particular position and exclude the ones that don’t. This behaviour is well-spread and happens when you point to data or individual cases that suits your goals while ignoring a large number of related cases or data that go against that position.

An example of Cherry Picking

Remember the last time you made a CV? You probably didn’t include everything thing you’ve ever done but picked the parts you thought was most likely to give you the job, to fulfil your goal. You did, however, leave lots of things out, things that you felt was irrelevant. Still, the data you left out could have painted a very different picture about who you are. (This is probably why a CV is not the only way data collected by potential employers).

Sure, cherry picking data for your CV, or online dating profile, is standard practice. But in many situations, it’s not. If you have employees presenting a small fraction of result data, it might be a good idea to ask about the rest…

3. Sampling Bias

Sampling bias happens when you draw conclusions from a dataset that isn’t representative of the population you’re trying to understand. It’s a systematic error that can appear if you don’t have a random sample. Sampling bias is the same as the “selection effect”.

An example of Sampling Bias

What do Americans think of Donald Trumps presidency? Most of us have an instinctive answer. But most people use Facebook as their primary information source when answering the question, and what they see there isn’t a public opinion – it’s their friends’ opinion. This is classic selection bias. You use data that is easy-to-access, but it only captures a particular, unrepresentative subset of the whole population.

Another case of sampling bias is rape and crime statistics. These datasets only contain the known and reported cases, but are missing a lot of cases who never see the light. The statistics over rape is therefore never showing the rates of actual rapes but of reported ones.

4. The Observer Effect

The Observer Effect is when people modify aspects of their behaviour because they know they’re being observed. So, when you monitor someone, you might not get findings that are representative for situations outside the controlled setting. The Observer Effect is sometimes called the Hawthorne effect.

An example of the Observer Effect

I remember when I, as a child, was supposed to estimate how much time I spent on brushing my teeth as part of my homework. I decided to set the timer, and then I brushed and brushed and brushed (I was pretty ambitious as a kid). This was of course very far from the actual amount of tooth brushing I usually did. Since I knew someone would look at my data point, it became unrepresentative. This is often through for self-reported data as well.

5. False Causality

False Causality is when two events appear at the same time we sometimes falsely assume that one must have caused the other. But sometimes this is just coincidence, or it’s something else creating both events.

An example of False Causality

So, if I eat an apple before I take a big test, and do really well… the apple must have caused, or at least had a significant impact on the result, right? Or, if the number of crimes goes up about as much as the number of ice creams sold, surely eating ice cream must make people more criminal? You realise that both these two are wrong, but sometimes this fallacy is not as obvious as this.

(It’s also good to learn the difference between causation and correlation.)

6. The Cobra Effect

The Cobra Effect is when you create an incentive that accidentally produces the opposite result to the one you intended. (Oops!).

An example of the Cobra Effect

The Cobra Effect got its name from when the British government wanted to reduce the number of Cobras in India during colonial times. Therefore, they offered a bounty for every dead cobra. Initially, this was a successful tactic, but soon enterprising people began to breed cobras for the income. The Brits quickly terminated the program, the cobra breeders set all the bred snakes free, and the end result was an increased wild cobra population. So the intended a solution made the problem even worse.

The Cobra effect is sometimes called “Perverse Incentive”.

7. Gerrymandering

Gerrymandering is when you manipulate geographical boundaries that group data because you want to change the result in an election. In practice, it’s often about drawing district lines to give a specific political party, minority, or other interest groups a disadvantage in an election.

An example of Gerrymandering

I wish that Gerrymandering was mostly a hypothetical fallacy, or at least that was not common practice to redraw election districts to give certain groups a disadvantage. I’m sad to say it’s not.

8. The Monte Carlo fallacy

The Monte Carlo fallacy is the mistaken belief that, if something occurs more frequently than usual during a given period, it will happen less often in the future (or vice versa). You might know this as Gamblers Fallacy or “fallacy of the maturity of chances”.

An example of the Monte Carlo fallacy

You know that friend with 4 children, all of whom were daughters? It’s straightforward to assume that when the fifth child is on the way, it must be a son. Still, the probability is still the same as always.

The previous turn-out rarely has anything to do with the results of upcoming events created by chance.

9. The Danger of Summary metrics

The danger of summary metrics appears when you only look at summary metrics. But, the summary is only part of the story, since there might be a lot of variation in a dataset that a summary won’t tell you. So, you can easily miss interesting or significant differences in a dataset by doing this.

The danger of summary metrics is also why you need to know the difference between mean, median and mode by heart.

An example of the Danger of Summary metrics

Say you are the CEO of a large fishing industry. To be able to sell your fish, each one needs to weigh about to 500 grams. Every week you get updates about the weight of the average fish in a sample and the number of total fish. It looks good, the average fish weighs about 500 grams, and you feel confident and base your profit calculations on these values.

When it’s time to sell the fish, and it’s captured and prepared. About half of the fish weighs 200 grams, and the other half weighs 800 grams. Neither of these is fish you can sell to full price. So, by looking at only the average [(200 + 800)/2 = 500], you got fooled into thinking everything was perfect, while it most certainly was not.

10. Data Fishing

Data fishing is when you misuse data analysis to search for patterns in data that reach statistical significance when there is no real underlying effect. By repeatedly testing new hypotheses against your data, you forget that most correlations will be the result of chance. Instead, you keep going until you find some significant effect to communicate. Data fishing is the same as data dredging, data snooping, data butchery, and p-hacking.

An example of Data Fishing

Imagine you take a huge sample of people winning the lottery. You don’t know why some people are more likely to succeed than others, but you know there must be some pattern you can find if you look closely enough. So, you start testing hypotheses: length, food consumption, number of siblings, and you keep going until you see a pattern that shows significance. Finally! Of course, you should have known that people born in August are more likely to win the lottery.

Well, they’re not. You just found a pattern that appeared in the dataset by chance.

11. Survivorship Bias

Survivorship bias is a logical error of focusing on the people, things or events that have “survived” some selection criteria. Overlooking those that did not, typically because they are not visible. This error can lead to false conclusions and is a form of selection bias. Sometimes called “Survival bias”.

An example of Survivorship Bias

You might have heard about the damaged US airforce planes in World War II? The returning bomber planes were filled with bullet holes, and the US armed forces realised they needed to reinforce them with armour. They started to think about where to put the reinforcement and plotted out the damages on some planes. The wholes were spread out but concentrated around the planes wings, tail and body.

But Abraham Wald, a statistician, made an interesting observation. He claimed that reinforcing the plane in these areas would be a tremendous mistake. When looking at the bullet holes, the army had only looked at the aircraft they had in front of them and had missed to factor in the damage on those who didn’t make it back.

Some planes didn’t make it back because their bullet holes weren’t in the same areas as the ones in the sample of returned ones. Most of these planes were hit in the engine, a part that – compared to the tail, wings and body – was extremely vulnerable. A bullet in the engine made the plane crash, so it didn’t return back home to be part of the sample.

12. McNamara fallacy

The McNamara fallacy is when a decision is based exclusively on quantitative observations (i.e., metrics, hard data, statistics) while ignoring all qualitative factors. You, therefore, lose sight of the bigger picture.

An example of the McNamara fallacy

One good example of the McNamara fallacy happens daily in classrooms – learning and performance. While it’s easy to measure performance, measuring learning is hard. We tend to focus on the thing that is easy to measure, using it as a proxy for learning through test results. We are then using our test result data to improve standards.

But with too much focus on test results, we might get worse teaching or even cheating, because the focus moved away from making sure students learn, to a focus only on scoring specific numbers on a test. So, when we focus on something that is easy to measure, it’s a significant risk that everything else is considered unimportant. Instead, we should be looking to improve the amount of learning and let this improvement drive up test results.

13. Overfitting

Overfitting is a modelling error. It happens when a function fits a limited data set to firmly because you’ve created a too complicated model to explain dataset you study in detail. This makes your model overly tailored to the data you have and, therefore, not representative of the general trend.

An example of Overfitting

“Oh no! Lisa is leaving the marketing department. How will we ever find a good replacement?”
Wanted: 36-year-old female with degrees in marketing and political science from Stockholm School of Economics. Needs to have a boring husband and two kids (4 and 6). You should spend your weekends hiking with your Golden Retriever. You should be 67 inches tall with blonde hair and freckles, and loudly curse people who eat fish in the microwave.

In this case, the employer is unable to differentiate relevant and irrelevant characteristics. The asked for qualifications are probably only met by the person who they know is right for the job, because she uses to have it. The problem is that she no longer wants it.


The difference between an offline and online audience

The one thing I get most requests about these days is helping clients when their online audience is not working. And I realise I see the same problems over and over. Naturally, that’s when a blog post is born.

Most people working with marketing audiences today started long before online marketing was the norm. They still create audiences for the online world like they did (or still do) for the offline world. But the two are very different, and naturally, audiences don’t translate very well between the two contexts.

The difference between online and offline audiences is how you decide if someone is part of your audience or not.

Your audience hypothesis

Before you create an audience for your ads, you usually have an idea of who you want the reach with your product. If you’re going to market a contraception app, spending your marketing budget on women between the age of 23-45 seems fair, but if you try to sell fancy cheese, your audience is somewhat different and probably should consist of cheese lovers with enough income to spend on cheese.

The big problem with audiences today can be summarised in this sentence: People are trying to target rich people through their interests in sailboats and expensive watches instead of targeting them based on income. (Because, yes we can!)

Often you have to translate your business audience into an advertising audience. Maybe because you want to personalise your ads based on preferences, or age; Or, because you have a small budget and want to make sure you spend it on those who are most likely to consume it. So, we are taking a hypothetical business audience and turning it into a concrete advertising audience. 

Sidenote: Some people work with personas in ther margeting strategy work to get to know their audiences. I try to avoid working with personas since I find it limiting. I will save my take on personas for a separate post. But its safe to say that personas create a lot of trouble when people are trying to reach their “personas” with online ads.

The limit with offline audiences

In the offline world, you have very little information about people. You often know the average income level in a zip code or a magazines rate of female readers, but you don’t have rich profiles or detailed information about a single person.

So, when you advertise offline, you do it in a zip code where the average income level is similar to what you think your audience earn. Or you choose a magazine with mostly female readers in a relevant age span. But you will never know if they like cheese or are trying to get pregnant.

What is a “proxy”?

In statistics (yes, building audiences is statistics), a proxy variable is a variable that is not in itself directly relevant, but that serves in place of an unobservable or immeasurable variable (Wikipedia).

The zip code is a proxy for income, and you can use a women’s magazine as a proxy for gender and maybe also specific interests if the magazine focuses on a particular type of content. Marketers try to find good proxies to make their advertising do better – but it is hard for some products and services.

Say you run a house cleaning service. You believe you should try to reach women because they feel more responsible for house cleaning (bleh). But you also want them to make enough money to afford your service. But how can you narrow it down further? Say you want to reach women with demanding and high paying jobs. Then you can advertise in magazines that people in this category are likely to read.

This magazine sure seems like a good proxy for your audience, but you won’t know how many of the magazine readers who are relevant to you. Some women who read the magazine might have demanding and high-paying jobs, but they have husbands that do all the housework. And others might not have a high paying job yet but wish to have it one day, so they are reading the magazine as inspiration.

Offline audience spill-over

So among those you target with an offline audience, only some people are the ones you’re trying to reach. You will have “spill over” to other groups that you’re not looking to reach. It’s the same with “out of home” ads, you can put them up on the bus stops in an area where the income level is high, but you will never know if the lion share of people walking past them have high incomes.

The benefit of online audiences

When you create audiences online, you have much more data about people. Either, you have your own customer data (“first-party data”), or you use data from Facebook or Google (“secondary data”), or you buy data from someone else.

Facebook and Google collect data not only on their platforms but all around the web and in apps through different scripts such as Facebook like buttons and analytics scripts. This data collection is why they know so much about their users, and this is why their advertising solutions are thriving online.

When you have data directly about each person, you don’t need proxies. You can target users on Facebook directly on income level. Or cheese interest, or their interest in contraceptive apps. These targeting possibilities are why Facebook ads can become so relevant. We don’t need to guess if those interested in cleaning services are also reading a specific type of magazine, or are more likely to have a gym membership. We can target directly towards people interested in cleaning services.

Why your Facebook Audiences don’t work

One small but super common mistake I see is people using AND instead of OR when building the rules for an audience. The audience people who: “live in London AND like trains” (small and neat), are very different from the audience people who “live in London OR like trains” (huge and messy). So this is a simple one to change.

Also, almost everyone that contacts me, about poorly performing online audiences, have built them incorrectly going from a rich but hypothetical idea of people who they want to reach, translating that into something unuseful. This is even more common when they are trying to create audiences based on fancy marketing personas they’ve got in a slideshow somewhere.

Hence, they are using proxies instead of pinpointing the behaviours or interests they are trying to target. This set-up makes their ads taking a massive detour in deciding who is relevant and who isn’t.

Instead, try this: 

  1. Start with defining what you actually think is the key characteristic in who you want to target and why. Is it age? Income? Location? Is it a combination of them?
  2. Stop trying to guess what your audience is interested in. You will most likely be wrong.
  3. Only add interests if it’s extremely relevant to your product. For instance, if you sell books, target on books
  4. Make sure to use the AND rule and not the OR rule if you combine interests
  5. Have correct optimisation goals and let the algorithm do the job in finding the best matches for your ad among the people you’ve selected above

And the use of proxy-based audiences for social media is widespread. I’m continually meeting both media agencies and social media marketers that are doing it daily. But it is just MUCH MORE WORK that will give you a WORSE RESULT. So it’s pretty easy for me to recommend you to stop.

5 simple skills you need to become more data-driven instantly

With marketing moving towards being more of a data trade than a creative profession, we all need to become more data-driven if we want to stay on top. Since not everyone loves math and analysis as much as I do, I’ve listed five simple skills I think you should learn to get started.

(And no, you don’t need to be a math geek to become more data-driven, if you know some basic math you’ll have more than enough foundation to get started.)


1. Learn how to convert files from .csv to a format readable by humans, like .xlsx

“There’s something wrong with the file”. This comment is by far the single most common I get from clients and colleagues who start to work more with data.

However, there’s nothing wrong with CSV-files. The format is used to store lots and lots of data without ending up with huge files. A CSV-file is a text file where you have one observation or object per row, and list all the related values in a specific order. You separate the data points by a known symbol, most often a semicolon or a comma.

When you want to work with data stored as a CSV-file in spreadsheet software like Excel you’ll have to convert it from a text format into spreadsheet format. Here are two simple ways to do that:

Option 1: When you start with a clean Excel document

  1. Open a new Excel document and navigate to the Data tab
  2. Click “From Text” close to the top left corner
  3. Choose to the CSV file you wish to open from your computer and click “Import”
  4. In the window that opens up, choose “Delimited” and click “Next”
  5. Check one of the boxes next to the different delimiter suggestions – most CSV-files uses either a semicolon or a comma to separate the values
  6. Click “Finish”
  7. Tada!

Option 2: When you’ve already opened your file

  1. Open your CSV file in Excel and select column A in your document
  2. Navigate to the Data tab
  3. Click on “Text to Columns” somewhat in the middle
  4. In the window that opens up, choose “Delimited” and click “Next”
  5. Check one of the boxes next to the different delimiter suggestions – most CSV-files uses either a semicolon or a comma to separate the values
  6. Click “Next” and then “Finish”
  7. Tada!

2. Learn the difference between average, median and mode

When you have a dataset with multiple numerical values, you sometimes need a single number to represent the dataset. A simple method to use is to summarise all your data points into a “typical” data point that represent the “centre” of the dataset.

However, when you calculate the “center” of a numerical dataset you have three different measures to choose from: mean, median, and mode. They each summarise your dataset with a single number, but they are not the same.

Mean 

The mean is the “average” number in a dataset.  You calculate it by adding all your data points together and dividing this sum by the number of data points.

Example: The mean of 37, 10, and 67 is (37+10+67)/3 = 114/3 = 38

Median

The median is the middle number in a dataset. To get the median you order all data points from smallest to largest and pick out the number in the middle. If you have an even number of data points and there are two middle numbers, you take the mean (see above) of those two numbers.

Example: The median of 37, 10, and 67 is 37 because when you organise the numbers from smallest to largest (10, 37, 67), the number 37 is in the middle.

Mode

In any dataset, the mode is the most frequent value – the value that occurs most often among all values

Example: The mode of [2, 4, 3, 3, 3, 1, 1, 2, 2, 2, 4, 1] is “2” because it occurs four times, and all the other numbers occur fewer times than this.

Why should you care? – Mean, median and mode in different datasets

In a normal distribution, the mean, mode and median measures are equal. However, if a dataset is right or left skewed, they are different from each other. Using these three metrics are therefore an excellent way to learn about the distribution of the data points in your data set, and that is often an essential part of analysing a dataset.

Three different data distributions and how the mean, mode and median are impacted byt the distribution

3. Learn what axis is the x-axis and what is the y-axis in a two-dimensional (Cartesian) coordinate system

The X-axis and Y-axis in a Cartesian coordinate system
The X-axis and Y-axis in a Cartesian coordinate system 

The X-axis and the Y-axis when looking at only the first quartile of a Cartesian coordinate system
The X-axis is also called the horizontal axis, and the Y-axis is called the vertical axis

Even though you don’t work with coordinate systems, any data-driven marketer should know which axis is which when someone says “on the x-axis, you can see the time of day” and “on the y-axis, you can see the number of likes”.

Most of the time in marketing analytics the correct terminology to use is “the horizontal axis” and “the vertical axis” since the graphs displayed are not coordinate systems. However, even if this makes more sense to you, some people will use x and y instead, and you’ll have to know what they mean.


4. Learn the difference between correlation and causality

Being data-driven will always include working with data analysis. Analytical skill is something you pick up over time, but there are two key concepts that you need to keep apart from the start: Correlation and Causation. These two concepts are important because if you don’t get them right you won’t get the rest of your analysis right. Additionally, it’s very easy to divide the data-driven people from the non data-driven people based on these two concept.

Correlation = describes the size and direction of a relationship between two or more variables.
Causation = indicates that one event causes the other – i.e. there is a causal relationship between the two events.

One classic causation vs correlation example is that smoking correlates with alcoholism, but it doesn’t cause alcoholism. However, smoking causes an increase in the risk of developing lung cancer.


5. Know how to distinguish the dependent variable from the independent variable

This skill is a little bit more advanced than the others, but if you want to become more data-driven and know what an analyst or data scientist are saying about your A/B-tests, it’s a good thing to remember which is which.

When you test a hypothesis with an experiment, the two main variables are the independent and dependent variable. An independent variable is a variable that you change or control in your experiment to test if this has effects on the dependent variable.

Independent variable = the variable that you manipulate (i.e. the time you post on social media)
Dependent variable = the variable that you hope to impact from the manipulation (i.e. the number of impressions for your social media post) 


My tools for Digital Marketing – Data Processing and Data Analysis

I often get questions about tools for digital marketing. Great topic for a blog post I thought, but when I started to write it became so massive that I realised I need to make a blog series out of this topic. In this post, I’ll walk you through the tools I use for data processing and data analysis.

Do you need technology to do marketing? Well, not long ago you could get by pretty well without it, but marketing is becoming more and more about technology. Sure, the underlying idea of understanding people is mostly the same. However, you will soon be left behind if you don’t add technology to your toolkit.

Finding tools for digital marketing

With a background as a tech journalist, I have this weird interest in new technology. Whenever I find something new I can try out, I get excited. I love spending time on ProductHunt to see what new products that might help me at work. Maybe not the most normal thing to do on a Sunday night.

I’m always on a hunt for two types of tools: 1. technology that can simplify or even automate part of my current job or 2. technology that can give me possibilities that I don’t have today.

1. technology that can simplify or even automate part of my current job.  or 2. technology that can give me possibilities that I don’t have today.

However, I started like most people: writing reports, making presentations and sometimes using spreadsheets to do calculations. I guess most of my time is still spent putting thoughts on paper, and my every day “martech stack” is still pretty basic.

Everyone can learn data analysis

Since I’m going to talk about tools for data, I want to say that my background is not in statistics or engineering. I do have some university credits in basic statistics, and I once knew how to perform a significance test in SPSS. However, I’ve learned most of my marketing data skills by doing. I started testing with small side projects, spending time with both MOOCs and tutorials online has been a pretty good way for me to learn.

But I would also say that part of why I learned it all was that no one else around me knew how to do data analysis, so if I didn’t try to figure it out on my own, I wouldn’t have any quantitative insight at all. Not using the data I had access to in some way felt more stupid than to try to do some data analysis on my own.

My tools for marketing analytics and data analysis

I regularly walk clients over slides with data analysis nowadays. So I do everything from the first export to the visualisation on my own. I think I have three types of reoccurring projects where I need my data skill set:

  1. Auditing – Looking at historical data, in a delimited and pre-defined context, to find how something performed
  2. Monitoring and Measuring – Visualising data in real-time, creating opportunities for better decision-making
  3. Research – Looking at trends and decoding information in an unknown context

The tools I use are either for processing, analysis, or visualising and presenting data. However, it wasn’t long ago I didn’t know how to split up a CSV-file without help from Google; Hence, you shouldn’t be intimidated or feel like it’s something you cannot do yourself.


Tools for Data Processing

Google Sheets

Google Sheets Screen Capture
Google Spreadsheets is a free tool for basic data tasks

The free cloud-based tool Google Sheets is a great way to work with certain types of data, especially medium-sized data structured in rows and columns. My first real relationship with spreadsheets started here, and its interface made me feel safe(r than other spreadsheet tools). I also realised I could find the answers to most of my Google Spreadsheet related questions online.

The first time I used Google’s spreadsheets was for personal budgets and other types of simple calculations. However, I like tracking stuff, so most of my courses at Uni had a spreadsheet with all the tasks and deadlines and suggested readings, and I kept track of my progress using colours and conditional formatting. I used it more to create well-structured files than to process data or calculate anything. I later started to do “Content Calendars” for clients in Google Spreadsheets, since it was easy to get an overview.

My introduction to using Google Sheets

My first real use of Google Sheets for work was to create monthly performance reports. I exported standard data files from Facebook, Twitter, Pinterest and the other sources I needed. In my Spreadsheet I had one “master sheet” calculating all my KPIs, referring to data from sheets named per data source. When I wanted to update my KPIs, I just overwrote the data in my import sheets, and all calculations in my master sheet refreshed automatically – as long as all the columns in my export were in the same place, but usually, they were.

I later started to write Google Script to make more automatic data imports, not having to download and upload files. Using scripts to get access to data from an API makes it possible to have automatically updated data in your sheets. This method makes it possible to use Google Sheets as a simple dashboard solution if you need to. I also use tools like Supermetrics to automatically import data from different APIs through a Google Sheets plugin (more on that further down).


Microsoft Excel

Microsoft Excel Screen Capture
Microsoft Excel is the standard software for data analysis using spreadsheets

While more advanced users might disagree, I would say that Google Sheets and Excel are very similar. You can do most things on a fundamental level in both tools. Some things are more comfortable to do in one than the other. I guess this is why I continue to use them both.

I tried to stay as far away as possible from Excel for very long. It felt like Excel was something only dull people needed, and my biggest fear in life was to seem boring. So, I stayed far away. For a very long time, I opened my .xlsx files in Google Sheets.

When is Excel better than Google Sheets?

The more I work with data, the more time I spend cleaning and prepping data files for data analysis. Additionally, my data files got heavier (and still do). So, the first feature I needed in Excel for was to use it without an internet connection. I know you can work with Google Sheets in offline mode too, but it doesn’t feel as safe.

Quite quickly I also learned that Excel was a bit more stable than Google Sheets when I was working with larger files. (It’s not like Excel is perfectly durable though, I’m very often looking at the spinning rainbow ball when I work with Excel). One elegant Excel feature is that you can turn off automatic calculation and make all your changes to the document before it calculates what’s in your cell. This feature might give you some extra processor power when you need it.

Today I often use more advanced functions in Excel than I did in the beginning. I could probably do the same thing in Google Sheets with a plugin, but it’s neat that they are already part of Excel. However, the type of data imports I do in Google Sheets, using Google Script is not something I do in Excel, even though it’s possible. Also, as soon as I need to share my documents with someone else, working in Google Sheets is often much more accessible.


Atom

Atom Text Editor Screen Capture
Atom is a text editor, where I get a clean formatted view of my data

Atom is a text editor (like TextEdit on a Mac or Notes on a PC). So why do you need a text editor to work with data you might think, it doesn’t make any sense? Well, at times when I work with data processing, I get across data stored in JSON or XML-files. When you try to read what’s in there, it looks like gibberish. A good text editor is very helpful when decoding these files.

However, any text editor won’t help you; you need a good enough one (like Atom or Sublime). A good text editor formats the text in your file to make it more readable. This process is often called “prettifying”, unfolding code that is hard to read into a structure that is friendly and easy to understand.

For example, Facebook Ads API display the targeting data per ad as a JSON-object. If you try to read it in a spreadsheet column, you will struggle. Atom formats JSON beautifully if you activate one of the “prettyfiers” that comes with the tool. You create a JSON-file, paste your FB-targeting data into it, and save it to your hard drive. On the save it will magically become formatted and (almost) readable. Also, the same is true for many other file formats that you might need to decode.


Tools for Data Analysis

Tableau

Tableau as a tool for digital marketing
Tableau is a Business Intelligence Software perfect for data analysis in larger data sets

I felt intimidated when I first found Tableau. Partly because I didn’t know about the tool, nor that there’s a field called Business Intelligence (with people doing data analysis for a living). Also, because it felt like I needed an exam to have the right to use it.

The primary usage for Business Intelligence software is data analysis. Naturally, this makes them pretty good at this. So I decided not to care about my lack of previous knowledge and downloaded a demo version. I quickly found myself with two problems: 1. It’s hard to get started with Tableau as a beginner, 2. Tableau is super expensive.

When should you use Tableau?

I use Tableau to extract information from my data. I look for learnings or insights that I could never see without help from software. For instance, are we spending our advertising money on the content our clients engage with or the material we think is best? How are different segments of our users behaving when they interact with our product? I ask questions to my data through Tableau. Most of the time it shows me that my first gut feeling is incorrect. At the same time, Tableau often shows me things that I had no clue about and would have never found through empirical studies.

Tableau is hard. This is because it’s different from most other software you’ve used before. It’s hard to translate prior knowledge into Tableau. Your data becomes dimensions and metrics, and in the beginning, nothing makes sense.

Teaching myself Tableau is probably the best thing I’ve done – it made me understand data on a deeper level. I’ve always known about the difference between boolean variables, strings and integers from my background in programming. However, it became much more tangible when I started to look at different data sets, trying to extract information from them.

Tableau has excellent tutorials online, so you are not alone in this process of not having a clue. There is also a great forum where you can read and post. Users always help each other out which is nice. But Tableau is not (yet) as well documented online as Google Sheets and Excel. When you have a problem, you might not find the answer in your first search. Stack Overflow is another place if you need help.


R

Last but not least, R is a tool that I don’t have to use that often. It is a useful backup tool for analysing data. R can solve some problems that are not possible with tools I’ve talked about earlier in this post. If you have a coding background, you will have fun learning R. If you don’t, R will seem pretty hardcore since you are interacting with the program through writing code and not through nice visual interfaces. It is similar to the terminal on your computer, in many ways.

I only use R when I have extensive data files, or if I need to calculate relationships between data points or data sets. Neither of these needs appears very often in my everyday life in marketing. However, I did a network analysis once, that wasn’t possible to do in any of the other tools. One key feature is that you can run your calculations on a server in the cloud if your computer cannot handle the size of your data sets or calculations. I’m not saying you should start doing that, but it’s good to know that it’s possible.

Overlaps between data sets are also hard for the other tools to handle, but R nicely calculates differences and draws Venn diagrams. R can do a lot of graphs and visualisation, but they are not the most visually appealing, so I don’t recommend to use it only for that.

How to learn R

If you think you’d like to learn R, multiple MOOCs can help you get started. I’d recommend taking one of them. However, if I were you, I’d start with one of the other tools; I only use a fraction of the functionalities in R. But sure, I plan on getting better at it, I just have some other things in life that I might prioritise before that…


Can marketing personalisation be unethical?

Personalisation is an ongoing marketing trend: far from new, far from over. Being specific and relevant to every single customer is a powerful marketing tactic, often appreciated by those receiving your message. But it has some downsides worth knowing.

I don’t know how many times a week I talk about the magic triad of marketing. Providing customers with 1. the right message, at 2. the right time, in 3. the right place. Personalised experiences give you tools to nail the first part, creating more relevant messaging.

We can personalise experiences in many ways. Personalisation online includes everything from adding customer names to welcome phrases in emails, to changing website content and design for every single visitor with help from artificial intelligence. The goal is often to increase customer engagement and conversions through improved relevance.

In this post, I talk about both targeting and personalisation. And why it’s not the same thing, doing one without the other is hard. If you have personalised content, you need to target the right person. If you target a specific group, you often (but not always) do it to become more relevant to this group. This will turn into personalisation if you narrow your audience enough and change the content to suit them.

The current state of personalisation

Few companies use full-scale personalisation today, but some display content on websites or in emails based on a customer’s previous behaviour. Others are creating content for different segments of customers, although there might be more than one person in each. This marketing tactic is underlying when customers with a birthday in November will get the same email, or those who bought notebooks recently will get the same ad.

Netflix or Amazon are both using advanced personalisation. They select every item on display specifically for you, and it sometimes feels like they know you better than you know yourself. Looking at my Netflix recommendations is like looking directly into my brain, it reveals every little quirk I’m not talking about in public.

Why personalisation needs careful thought

While personalisation is a powerful marketing tactic, it is sometimes perceived as “creepy” by customers. In the best of worlds, people like their personalised experiences. At times they might “only” get a bit uncomfortable. But personalised content can also be very unsuitable or even unethical.

There will be a constant battle between personalisation and privacy, and it is essential for marketers to know a bit about the risks. I will discuss two types of content personalisation and online targeting situations that are more problematic than we might think at first – but there are of course many more.

Many (most?) digital marketers use these methods today without knowing it is problematic. In this world of continuous consumer data collection, we need to discuss marketing tactics and marketing ethics – but we don’t. This blog post is far from a complete guide, but it might get your thoughts started.

Targeting based on health data

We give away a lot of health information online. Googling symptoms, looking for home cures, worrying about constant headaches or trying to break our bad habits. Our online behaviour additionally gives away many cues about our mental health – how we interact with social media, for instance.

But just because we can target an obese person with diet tips or a depressed person with advertisements for self-help books or therapists, doesn’t necessarily mean we should. And it just becomes even worse if you start targeting cancer patients or their relatives (for instance, people who have visited the cancer wing at your local hospital) with ads about funeral services.

Health issues are one of these things people don’t want others to know about, and we will find that targeted ads violate our integrity or are intrusive. While it is probably okay to communicate with parents around the pains of having sick kids in February, we should carefully make sure we don’t fall over.

Retargeting users with health-related content

Another issue that often appears is retargeting based on earlier shopping behaviours. But what if we spend some time looking at health advice, comparing medicine or googling back problems? If you have browsed around for self-help books, and all of a sudden get suggestions for all the other self-help books you should buy too. Or if you look for specific medical treatments or drugs and it follows you around the web for weeks, that is not very good marketing.

Personalised or retargeted content based on someone’s health, is often a terrible idea. It will give people the feeling that you know things should not know about them, and the ads likely will not perform since customers find this intrusive. Pick another marketing tactic.

Political opinion as targeting

Using someone’s opinions for targeting is an ancient (well, sure) marketing tactic. It is not only about how someone is voting; political views include so much more.

Many niche opinions that might seem harmless to you is controversial or even illegal in parts of the world. People engaged in gay rights, pro-drugs or anti-abortion movements – might end up in unpleasant or even dangerous situations if their opinions get out. Maybe not in your context, but in theirs.

With Facebook’s interest targeting, it was previously possible to target people with interest “Jew hater” or “How to Burn Jews”. And while Facebook took away these particular targeting options, many others are still there. You can target people based on ideology, such as liberal or conservative, or fans of a specific political party.

There are cases where targeted ads seem to have put people at risk, or in uncomfortable situations, both in online and offline contexts. For example, people do lose their jobs when their employers find their opinions inappropriate.

Sure, if you live in a democracy and everyone is free to speak their mind openly, the risk might be small. But it was not long ago that people had to hide their political views from friends and family in parts of Europe – and people still get into trouble because of what they think politically, all over the world.

Calculating the risk

Giving someone’s political views away by showing them personalised content can lead to consequences we might not consider, or even understand when we do the targeting. These risks vary from market to market, but you are responsible for deciding what is okay and what is not.

Often, when you do use someones political opinion for targeting, you are using it as a proxy to categorise people with a specific set of values, behaviours or traits. Using these traits directly is often more ethical, and it will make your marketing much more relevant to your customers.

(A side note: Using consumers political residence for targeting or personalised content does not mean you cannot promote political content. But who gets your political content and why needs to be an active and careful decision based on fair grounds.)

The need for marketing ethics will increase

Marketing ethics will become a big deal over the next few years. All marketers will need to know more about ethics than we do today, or we will get in trouble. Many companies will get lost, and they can end up in some real trouble putting their brands at risk.

Technology, both specific to marketing and not, will continue to give us marketing opportunities that are inappropriate and unethical. New tools help us target and personalise messaging, and they are continually getting better and more popular. But most marketers using these new tools have never had to think about what is okay and what is not. Instead, we do things just because we can, because that’s what we’ve previously been doing.

To follow the law will never be enough

Customers don’t mind to barter with their data if they get access to free valuable services in exchange. But they are picky about how brands use their data. As soon as ads feel intrusive or inappropriate, customers will not engage. This behaviour creates a paradox where marketers will have to balance between extremely relevant to each customer, and not to make customers feel uncomfortable with why they get what they get.

Policy and legislation will soon create more boundaries for digital marketers. But it won’t ever catch up with technology. Law development being slow is a problem for computer scientists and programmers as well. So, we all need to brush up our ethics.

I’m not saying that business has ever been entirely ethical. Too much money is often at stake. But marketing is easy to review since it’s on display, so I think we’ll have to get in line pretty quickly – or we will get in trouble.

Do you have any thoughts on this? Please leave a comment!

Why you should take a break from Facebook marketing

Marketing is at times a bit like finance. You place a bet on what you think will perform, and then you wait and see. Luck plays a part in the result, but often experience adds to the success rate.

But just like an investor is not placing bets by guessing, neither are marketers. A variety of information and some intuition goes into every bet. We often base our bets on a combination of historical data, the calculated risk and the potential return.

Both marketers and investors are looking for the best return on their investments. For every dollar we spend, we want as much back as possible. For an investor, the gain is often in dollars, but for a marketer, it might be in sales growth or recognition. And while it might be easier to calculate ROI for a financial investment, the underlying mechanism is very similar.

Betting on the right thing

Marketers, just like investors, need to know how their surroundings are changing to make sure they place their bet strategically. Today, the surroundings for marketers primarily consists of algorithms: Facebook’s news feed, Google search, YouTube’s video recommendations, and some others.

When algorithms change, a bet that once was lucrative won’t pay off anymore. And while investors are quickly moving from an investment that is no longer paying off, marketers seem to keep going for a long time before they realise they should change something.

Why is that? Why do so many marketers continue with efforts that do not pay off? Is it because we are stupid? Or lazy? Is it because marketers don’t follow what’s happening in their field as closely as investors read the financial press? I don’t have an answer, but it fascinates me greatly.

Facebook marketing and ROI

Last summer the debate intensified around “fake news” and Facebook’s part in the outcome of the US presidential election. Quickly after this, the referral traffic from Facebook started to dive. The Facebook algorithm changes from time to time, but they didn’t say anything about changes at this time.

Facebook has made changes before as well. The platform has moved from being a source of free engagement and referral traffic, to become an advertising platform where you buy your impressions. These changes make a lot of sense; Facebook is a for-profit company.

As you see, betting on Facebook for marketing results comes with a high risk. If they change things on their platform, you will lose. Even if you’ve invested all your money on the Facebook platform for years now, you still might end up with nothing at the end of the day. Since you only “borrow” the relationship with your fans from Facebook, there is no real value in a Facebook page even if it has hundred thousand followers.

During 2018 we have continued to see multiple indicators suggesting the ROI on Facebook marketing is declining. When you, as a brand, publish content on Facebook today almost nothing happens. Some features, like groups and live streams, are ways for brands to still be relevant for users on the platform, but it’s very hard to get a good return on your investments.

Maybe, it’s time to take a break? At least, if you don’t have an advertising budget and if you spend a lot of time on your Facebook content and still get small results.

What to do instead of Facebook marketing

A lot of people think that digital marketing equals Facebook marketing. Some might throw Instagram and Twitter into the mix too. And while Social Media marketing is an essential part of digital marketing, one skill you should have in your toolbox, it’s not everything.

You need to move over to platforms where you have more control of your success. Brands doing content marketing should most likely focus on other platforms, and increase their focus on SEO and e-mail.

E-mail marketing and blogging

E-mail is an alternative to Facebook because your success is very closely correlated with factors you can impact. Search is another (now) stable platform, although not entirely without risk since it’s still an algorithm behind it.

Business blogging have for long been a bit uncool, “why should anyone wanna read a business blog”, but I think it is will soon have a revival. It’s safer than Facebook if you want to get back what you invest. A bonus is that more formats and topics will work in a blog setting than on Facebook. To be successful, you don’t have to do 45-second videos with text in the frame.

Podcasts and YouTube

Podcasts and YouTube will also become more popular when Facebook investments are no longer creating great returns. YouTube is apparently the second largest search-engine in the world, after Google, so there’s potentially an even more significant return to make from this platform. Podcasts are unique because users can use multiple technical solutions to consume your content, something that almost makes it similar to e-mail and reduces the investment risk.

Not that many marketers are very good at working with podcasts or YouTube today unless it is one-offs or sponsorships. Much of the best material is instead generated by “Podcasters” or “Youtubers”, with the content as their core product. But both these platforms now feels more stable than many other content platforms (as long as you are not trying to use it to make a living and don’t have to care about the compensations models), and the potential audiences are large.

How I try to place my bets

Personally, I began to refocus my Facebook marketing initiatives for most clients or projects last summer. More than a year ago (maybe earlier) I moved away from Twitter and Snapchat and recommended others to do it too. The platforms weren’t delivering enough results, and the risk of them just disappearing felt too big to for the small ROI they produced.

Instead, I’ve started this blog, and I have an e-mail newsletter (you can subscribe in the sidebar to the right). I’m also looking at starting a pod or a YouTube channel when I have the time. Sure, this tiny blog is not Spotify, but I would probably recommend them too to move away from Facebook at the moment.

My focus is on consistency and not quantity. It’s often wise not to hurry when it comes to scaling content initiatives; it takes time to build an audience.


Dark Posts in Social Media, what is it?

One buzzword currently circling marketing departments is “Dark Post”. It sounds like something dangerous and illegal, probably because you have heard about the “darknet” or the “dark web”. However, the two have nothing to do with each other.

Dark posts are nothing unlawful or dangerous; they show up daily in our Facebook feeds. Moreover, the only negative thing is that they are sometimes hard to track.

If you are active on Facebook now and then, you know that what you see isn’t only posts from other personal users. Pages – representing everything from your local flower shop, an extremist political party, or an initiative to teach kids to code – can also create posts that show up in your feed.

However, a lot of the posts that show up in your feed are ads. When a page creates a post on Facebook, it appears both on the page wall and in some followers feeds. The page can if it wants to, decide to boost the post to reach more people. Then they pay Facebook to show the post for Facebook users they want to reach. When this happens, the post technically becomes an ad*. These posts are often called boosted posts.

*If you do not boost your post it is referred to as an organic post.

You see dark posts all day long

Facebook advertisers want all their ads in your feed to look like any Facebook post. But the also want to send different messages to different users based on what they think they’ll like. To become relevant, or spend their money as effectively as possible, advertisers often create multiple versions of a message. They have different versions of copy, links and images, to find the one combination that performs best. This is a common advertising method called A/B testing.

But brands on Facebook usually don’t want their Facebook page wall to show the same post over and over with slightly different combinations of text, links and images. So creating an organic post and boosting it is not a very good option. Instead, they create a large variety of ad posts in Facebook Ads Manager and Facebook shows you the ad that it is most likely that you act on.

There’s a difference between a boosted post and an ad post created with Ads Manager. Boosted posts show up on the page wall, but ads don’t show up on the wall (if you don’t want them to). Dark posts are ad posts that don’t show up on a brands page wall. Dark posts live “undercover” or “in the dark” and no one except the ads targeted audience knows about the post. Another name for dark posts is “unpublished posts”.

You can create dark posts on both Facebook, Instagram, Pinterest, Snapchat and Twitter. On Snapchat and Instagram, all promoted posts are dark posts. The basic idea is the same to all these platforms, but this text will continue to talk about Facebook.

What a dark post looks like

On Facebook, boosted posts and post ads created in Ads manager look identical. You can see the  “Sponsored”-mark directly under the page name when they show up in your feed.

You can only see a dark post if you are in the target audience for the particular ad and it happens to show up in your feed or if you know its direct URL. (You find the URL for a Facebook post by clicking on its timestamp).

Who sees a dark post?

Sponsored posts do not show up randomly. We all have different ads in our feeds, and the same ad does not show up for everyone. When a page creates an ad, they decide whom they want to show it to, and then it only shows up in their feeds. Targeting is the advertising term for choosing who to reach.

An ad shows up in your feed because you fall into a group the advertiser wants to reach. Either because of your gender, age, or where you live. However, it can also be because you have behaved a certain way online. Facebook use user data for targeting ads on its platform. Sometimes the pages you like or the pages your friends follow is motive for ad placement. But it is also possible to base targeting on different interests Facebook believe you have, most likely because you have clicked links or viewed videos about a subject.

However, even if the ads are showing up on Facebook they collect your online behaviours all around the internet. Facebook collects data from an enormous amount of sites online. Pages can re-target their web page visitors on Facebook if they have installed a Facebook script called a “Facebook Pixel”.

Dark post as a tool to reach niche audiences

Dark posts target particular niche audiences. The main idea is to create an ad and promote content towards someone likely to enjoy the content. The targeted user is often a potential customer, but dark posts can also be used to persuade voters in a presidential election or increase streams on a specific Netflix show among existing Netflix users.

For a brand publishing content on Facebook, it is sometimes hard to create content that talks to all your potential customers or fans at once. People are much more likely to engage with your brand if they feel like you are relevant to them, but how can you be of interest to a varied group of potential customers? This dilemma is why dark posts are often part of a successful content strategy.

Since an advertiser is in almost complete control over who sees an unpublished post, they can talk to multiple audiences at once, with different voices and propositions to each of them. The people who get the content in their feed do not know they are part of a bucket; they are just happy their Facebook ads are somewhat relevant**.

**Facebook also wants to show people relevant ads. They assign all ads on their platform a score from 1-10 based on their relevance to the target audience. If your ad shows a low relevance score, you can tweak your ad, or it’s targeting, to improve the effectiveness of the ad.

Dark posts in Social Media are not good or bad

Dark posts are in some circuits starting to become somewhat mythical, a tool used to manipulate people without their knowledge. The name for it is probably not helping though, especially not when it is few that know what it is.

The thing that makes dark posts somewhat criticised, or at least met with scepticism in the debate, is mainly two things: 1. Fake (junk) news sites, companies and political campaigns have used dark posts unethically in different ways, 2. It is very hard to track dark posts from an ad account that you do not own yourself.

First, the ethical aspects of dark posts

The fact that we can target people in extreme detail while being somewhat non-transparent makes it easy for an advertiser to enter into semi-legal or at least unethical activities either by mistake or knowingly. The regulations are often vague, and no custom exist around these issues.

Facebook is a profit-driven platform and seems happy as long as they get paid by advertisers. They have changed some policies related to fake news and ad posts after the storm of criticism that followed both the US presidential election and Brexit. With the new GDPR regulation from EU, arriving in 2018, this will probably change.

Second, transparency

Keeping track of what others are saying might seem like a vague argument against dark posts. Targeting ads have always been part of marketing and communication. But the vast amount of ads and the difficulty to track them makes it complicated. This grey zone is not a big deal when it comes to shoe sales, but if it is false political arguments getting spread around in the dark, it is somewhat problematic when they cannot be met or debunked.

It is also likely that we will see more tools for tracking dark posts popping up due to increasing demand from marketers to keep track of competitors. We can probably also expect greater transparency on Facebook, as well as on other digital platforms, both when it comes to both dark posts, and different types of online targeting. Facebook is trying to become an infrastructure more than just a social platform, but for that to happen, society needs essential insight into the platforms citizens use daily.

Should we be worried?

As with all tools, dark posts in the wrong hands can create some damage. However, it is not a dangerous tool in itself. If we keep spreading the knowledge about how dark posts work and how we can check if the messages we get online are really true, we don’t need to be afraid.