Jeremy Smith Feb 19, 2019 11:25:50 AM 15 min read

The Science of Conversion Optimization

If you’ve read any of my writings about conversion optimization, you’ve seen me refer again and again to the need for a scientific mindset or an understanding of science to truly understand this field and be able to effect true website optimization.

On my conversion optimization services page, I state:

True, valid and valuable conversion optimization requires an understanding of how to merge different data sets as well as the fundamentals of the complex worlds of science, psychology, statistics and data analysis.

But I’ve never really explained what makes this a scientific undertaking. It sounds kind of heavy-handed. Oh, it’s science; you wouldn’t understand.

And while it really does require science and you may not understand, maybe I can explain why we say this.

I’m not alone. Others also state but don’t readily explain how science figures into conversion optimization.

These fellows seem to be pretty sure conversion optimization is a matter of science. Hell, they even  claim to be conversion scientists.

And while even I am not so sure how broadly we can stretch the term and take up the mantle of “scientist,” I agree that conversion optimization requires some measure of scientific understanding. But what do we mean?

Don’t Be Blinded by ‘Science’

Here’s how WiderFunnel describes the role of science in conversion optimization:

(Y)ou cannot simply “guess and hope” which changes to your web pages will achieve a higher conversion rate. Instead, ... experts develop and test valid hypotheses, run controlled tests, and evaluate results to lock in the improvement.

In fact, the whole process of conversion optimization, if approached as it should be, aligns with what’s known as the “scientific method.”

Music video still of Magnus Pyke yelling 'Science!'The scientific method is described as "a method or procedure that has characterized natural science since the 17th century, consisting in systematic observation, measurement and experiment, and the formulation, testing and modification of hypotheses."

Magnus Pyke (at right) could have explained. (Image source)

Measurement or Data/Statistical Analysis

Conversion optimization, in the broadest picture, comes down to numbers. The point is to increase the number of conversions, which – it is hoped – will lead to better revenue numbers. And much of the work requires analyzing the statistics relating to website visitors, your web analytics data, to see what they tell you.

Is statistics a science? For many business majors, “statistics” is a required class they hated and struggled through. But the American Statistical Association quotes Marie Davidian and Thomas A. Louis and says, yes, it is a science:

Statistics is the science of learning from data, and of measuring, controlling, and communicating uncertainty; and it thereby provides the navigation essential for controlling the course of scientific and societal advances.

Do increased conversion rates equal societal advances? Well … let’s say some yes, some not so much. But that’s not the point here and now.

graphic description of how science worksThe point here is that you must understand what the statistics provided by your analytics tool mean and how to communicate their meaning for them to do any good. Or as a USC-Berkeley “Understanding Science” primer (at left) says, “Data become evidence only when they have been interpreted in a way that reflects on the accuracy or inaccuracy of a scientific idea.”

And understanding statistics, as with any discipline, scientific or not, requires some combination of training and experience – usually, the more the better.

Understanding statistics is also crucial to testing your theories in a scientific manner.

Testing With Valid Results

Marieke van de Rakt, project manager at Yoast and a former criminologist who studied criminal behavior in large-scale data sets, has expertise in statistics and research design. She explains that “without some proper knowledge about research and statistics, the pitfall of interpreting conversion rate tests results incorrectly is large.”

She also explains, as I have maintained, that the various tools available online to conduct tests for you are there to make the job easier but cannot substitute for knowledgeable guidance. A hammer helps you drive a nail, as she says, but it cannot decide, for instance, whether the nail is too large and will likely split the board you’re about to put it in.

Testing, typically A/B split testing, is integral to conversion optimization. But valid testing requires a specific knowledge about how to set up and run tests, recognize and weigh variables, and ensure statistical confidence in your results. And, as already explained, someone has to know how to interpret and communicate the results, even when ostensibly you’re simply testing one iteration against another.

Psychology and User Behavior

We all recognize that psychology is the science of understanding the workings of the human mind and more particularly how it affects behavior. I know it’s a mistake to say we “all” recognize anything, but if you’re in marketing, I think I’m pretty safe in saying you believe in the science of psychology.

It’s a great interest of mine, and I’ve written extensively about web psychology and the multiple ways that understanding the working of the human mind applies to your conversion optimization efforts. But, again, let’s find someone to back me up.

Authors Khalid Saleh and Ayat Shukairy laid out their “conversion framework” in Conversion Optimization: The Art and Science of Converting Prospects to Customers. The framework’s eight principles deal chiefly with psychology:

  • Understanding your website visitors through persona creation
  • Creating confidence and trust (in users’ minds)
  • Understanding the impact of the buying stages
  • Dealing with FUDs (fears, uncertainties and doubts)
  • Using incentives (the psychology of customer motivation)
  • Engaging users.

The seventh principle is that you will test any changes you make – “By applying the science of online testing you will measure the impact of any change you make on your bottom line.” And the eighth is making a long-term commitment to conversion optimization, “or the willingness to iterate,” which is what I mean when I say conversion should be an ongoing process.

Saleh and Shukairy’s first principle refers to developing customer personas, which is necessary to understand, essentially, who it is you are optimizing for. This work encompasses the principles of psychographics, or the study of your customer’s interests, activities and opinions (or IAO variables).

This goes back to the observation aspect of the scientific method. You learn about your customers’ IOA variables by interviewing them, conducting surveys and focus groups, digging into social media to see what they’re talking about, and digging into analytics data, which portrays what they do on your website.

More narrowly, the psychological aspect of conversion optimization is often referred to as neuromarketing, the application of psychology to advertising and marketing. When we talk about psychology and marketing, we can traipse far into the weeds while examining topics like cognitive biases, risk theory, the use of competitive matrices, font choice, and on and on.

Indeed, marketing psychology is the stuff of books. I don’t need to change any minds here.

Where Statistics and Psychology Intersect

What’s spooky about all this is that science has worked out how to use data analysis to accurately predict human behavior. A discussion of the rising use of big data to predict behavior in the Deloitte Review points out that documentation of how data is superior to humans’ prediction of behavior dates to 1954 and the publication of Clinical Versus Statistical Prediction by the psychologist Paul Meehl.

A more recent example is Michael Lewis’ Moneyball, which recounts the use of statistical analysis to determine how well baseball players would behave on the field and showed that all the professional scouts overlooked what the data proved.

“Meehl’s lesson – routinely echoed in case studies ranging from baseball scouting to evidence-based medicine to university admissions – is that in virtually any domain, statistical analysis can be used to drive better expert decisions,” the study authors write.

“The reason has nothing to do with data volume [i.e., big data is not necessary] and everything to do with human psychology.”

But analysis of big data, which is information about people’s behavior and interactions instead of their beliefs, is taking the capabilities of analytics much, much farther.


I hope I’ve persuaded any doubters or the confused that conversion optimization is a process that relies at least in part on understanding and applying scientific principles. As the sources I cite also say, there’s an art to it, as well.

But please don’t make me explain “art.” You’ll know it when you see it.


Jeremy Smith

Digital marketer with a penchant for dance; helping clients see the light through the jungle of tweets since before Twitter was cool.