Want a more effective marketing strategy? Use scienceWant a more effective marketing strategy? Use science

Want a more effective marketing strategy? Use science

Written by Axonn on 21st Jan 2015

We all claim to be data-driven…

We know we need a strategy for marketing, we know we need to A/B test, we know we need more than a passing glance at our Twitter analytics to know if what we’re doing is on track. But do we? Can we all hand-on-heart claim that our marketing processes are purely backed by science?

As marketers, most of us are confident in our creative skills. If we are to believe that there is a spectrum from scientifically to artistically-minded, the majority of us will fall pretty far on to the creative side. But we know that particularly in marketing, art and science are not mutually exclusive. In order to be successful at marketing we must straddle the line between the two.

But how many of us are completely honest about the role of science in our marketing? Axonn managing director Alan Boyce recently presented a webinar on the topic of science and numbers in digital marketing, citing that “time and time again we see data being using like drunks use lamp posts – for support rather than illumination.”

As Alan said in his webinar – data itself doesn’t lie, but it can be misused, misattributed, and misinterpreted. We can skew data – intentionally or not – to produce the results we desperately want it to show. And let’s face it, we’re all suckers for anything that claims to be “scientifically proven” – but how accurate is that concept?

Why we need science in marketing

Marketing by numbers_Page_11.jpg

In scientific experiments, particularly on humans, there is usually a placebo or a control group. These groups are monitored in the same way as the groups involved in the experiment and ensure that behaviour or results exhibited are not naturally occurring, and also act as a comparison.

The marketing equivalent of this is A/B testing, where two groups receive the same data or information with one slight change (to be measured). With all other variables identical, it is easier to see the results of the change, and to see that other elements have not influenced the results.

But even in a situation with such isolated variables, the marketer needs to be wary of jumping to conclusions regarding cause and effect – scientific integrity must be employed to identify which results are causal and which are simply correlation or coincidence.

Theoretical physicist Richard Feynmann said of experimentation: “You should report everything that you think might make [your experiment] invalid – not only what you think is right about it.”

Often we are so keen to demonstrate the scientific success of our experiment that we ignore significant flaws in our methodology.

As Feynmann states: “The first rule is that you must not fool yourself – and you are the easiest person to fool.”

A marketing experiment is really no different to a scientific experiment – in both instances experimentation is about isolating the smallest set of variables, testing a hypothesis and repeating the process to see if the results can be replicated under the same conditions.

Alan adds that this is why you should be suspicious of any marketer who cannot tell you about a spectacular fail they have been involved in – they can’t possibly have tested enough alternatives to know if their success is just a lucky coincidence.

Love data? Check out our infographic on social media use and how to get your strategy right.

How to employ more science in your marketing

Marketing may not be theoretical physics, but that doesn’t mean it can’t apply the same scientific integrity, attention to detail and analysis of all results.

Perhaps you were expecting your A/B test to result in a spike in traffic, but instead it increased the ratio of males coming to your site over females. This might not be the result you were expecting, but that is still a result that needs noting and analysing.

Analyse the data your analytics present to you and look at potential cause and effect, eliminating anything coincidental. Check your analytics daily and compare to previous days, months or even years if you have that data available. Look for trends and experiment. Try new analytical tools. Test every possible variable, and don’t be that marketer who has never failed. Failing shows you are trying and testing every possible outcome.

Be honest with your results. And in the words of Alan Boyce…

“If we apply a little scientific integrity to what we’re doing and claiming, and think a little harder about why our experiments might be loaded or our results might be coincidental, then I think we can make marketing more predictive, more reliable and – dare I say it – a little more respectable.”


Read More