Are you still marketing to idiots?

It’s done, we are set. We have a marketing strategy, budgets allocated and are ready to push the button to “market the hell” out of our customers so that they will come running to us, love our brand and buy our products.

But it wasn’t easy to get to this point…

By now we spent a considerable amount of time crafting the perfect marketing program. We created buyer personas, came up with a content strategy, one that will get our audience more than just excited, and we evaluated customer journeys in so much detail that even the word “journey” itself feels embarrassed.

Now, we are ready, we finally push the button to drive traffic through our online channels, we capture our audience in the moment of pure intent and we watch our users convert one step at a time, deeper and deeper down the funnel.

Exactly as we planned it… until everything went totally out of hand and our audience went full-on zombie. Every user started behaving completely random and instead of following our step-by-step plan (clicking the banner, going to the landing page, ordering online) they follow a completely irrational pattern.

What happened?

The truth is that many digital marketing strategies assume every user behaves like a complete idiot.

We assume that everyone will follow the same path with little deviation, but the path a user will take, from initial contact (awareness) until purchase (conversion), is rather random by nature.

Here is an example of the journey I took, when I was recently shopping for a new laptop bag:

I was carrying a Timbuk2 bag for 12 years, my first choice was to go and check out their website. I typed in Timbuk2, landed on Google clicked the first ad (which was their own), checked out some of their new bags. Went back to Google searched “best MacBook laptop bags 2016” read some articles. Researched some of the brands that were mentioned in the article. Landed on and looked at some bags. Got retargeted (not retarded) on Facebook in the coming week with sponsored posts from Amazon. Went back online a week later and showed some of the bags I liked to a friend. Got an email after two weeks from Timbuk2 with some offers. Bought one of their bags…

While the touchpoints I interacted with are quite common, two users will never have “exactly” the same journey, it will always change slightly based on the sequence, timing, steps taken or content they will see.

In order to be able to analyze this problem better digital marketers have created the notion of micro and macro-conversions. Micro-conversions are multiple events that eventually lead to a [macro] conversion. Micro-conversions are often considered to be actions such as liking a Facebook post, signing up for a newsletter, or downloading a report. Macro-conversions are the events that impact the bottom line i.e. a contact request or purchase.

Micro-conversions are not sequential and may occur in any random order.

Since their order is completely random it is difficult for marketers to know which micro-conversion will lead to another. We also have to keep in mind that each of these micro-conversions requires their own marketing effort. For example, in order to get someone to like your Facebook page, you will need to create some great content and get it noticed by your target audience. The “like” you get from this effort may lead the user to see another post in the future, which then may take him to the next action such as visiting your website, signing up for a newsletter etc.

In order to know which micro-conversion performs best, we need a way to measure their performance. This is where attribution modeling comes into play. What it means is to attribute a value to each micro-conversion that leads to the final macro-conversion. When assigning the value we can use different models such as last interaction before conversion counts, first interaction that lead to conversion counts or to use a multi-touch attribution where each micro-conversion receives some credit. A time decay multiplier can be added for even more accuracy, which means that micro-conversions in the past receive less credit, however, this would lead to the debate that if the first ever micro-conversion never happened, the sale may not have happened as well.

If the first ever micro-conversion would not have happened, would the sale still have happened?

Simplified Example:

[Facebook like] > [Banner Click] > [Purchase]
[Banner Click] > [Facebook like] > [Purchase]

In the above example and depending on which attribution model you use the attribution would look like this:

Last interaction
[Facebook like] > [Banner Click 100%] > [Purchase]
[Banner Click] > [Facebook like 100%] > [Purchase]

First interaction
[Facebook like 100%] > [Banner Click] > [Purchase]
[Banner Click 100%] > [Facebook like] > [Purchase]

[Facebook like 50%] > [Banner Click 50%] > [Purchase]
[Banner Click 50%] > [Facebook like 50%] > [Purchase]

Based on this data we can find out which channels as well as which micro-conversions have the biggest impact on the final sale.

What are the common challenges when using this approach?

Of course, there are a couple of challenges such as that in the real world things are just a little too chaotic at times. Customers are interacting offline, they are doing things we can’t track and they may have a past history with your brand that you are not aware of.

Brand awareness, word-of-mouth, special offers are just a few more influencing factors that may have an impact on your conversion rates. Again, your customers are not idiots and just because the last thing they did before making a purchase was liking your Facebook page doesn’t mean everyone who will like your page will buy your product too.

Trust, affinity and top-of-mind are a few more factors that can influence buying behavior and so are pricing, seasonality or the economy itself.

Predictive analytics try and solve these challenges and through big data, machine learning and statistics we can try to predict user behavior. This means that by applying these techniques we can optimize a marketing program and increase conversion rates by identifying the set of customers that are most likely to buy your product, or finding out which marketing messages will resonate the most with a certain customer segment.

Digital marketing and the vast amount of data we can collect allows us to make our efforts as scientific and complex as we wish. The truth is that most marketing programs are still very basic and most of the effort should go into optimizing these basic programs.

Applying machine learning to your marketing data may not make sense at this stage, but going beyond the assumed customer journey of a Lemming may help you to measure marketing return on investment.

Simple analytics, such as attributing conversion credits to your various online channels, can be done easily using free tools such as Google Analytics.

The most important thing to remember is that customers have access to more information and choices than ever before, which makes them smarter and more powerful than ever before.