Skip to main content


Consumer research and testing environments for website optimisation


Guest Blog: Dr David Lewis, best-selling author, award-winning broadcaster and Chairman of Mindlab International speaks about the difficulties of getting ‘pure’ readings in research, and how to get closer to understanding consumers’ true motivations.

Dr David Lewis, psychologist and neuroscientist known as the father of the billion-dollar Neuromarketing industry, was the first to use EEG (Electroencephalography) to study brain patterns to help determine sub-conscious motivations with commercial applications. This built on work by Krugman in the US who conducted the first experiment using electrodes to decipher the subconscious thoughts and feelings that occur below conscious awareness.

Dr Lewis has since developed other research techniques for retail situations to help uncover subconscious behaviours and the difference between what people say and what thy actually do. He then relates this to ways in which websites can improve sales by creating greater cognitive fluency and reducing cognitive dissonance.

AWA optimisers got the chance to quiz Dr Lewis on his research at one of our regular quarterly meetings.

Behavioural contrasts and surprises

Q. Dr. Lewis, I wanted to share some of the research that we found, also that colleagues of ours have found, which is we use a split-testing tool called Optimizely, which you may well have come across. The founder of that business is an American businessman based on the West Coast.

He was Director of Analytics and Optimization for the 2008 Obama campaign where they used a range of different images and a range of different calls to action. They tested different ones against each other, and they found out that ultimately a picture of Obama with his family and the call to action saying “Learn More” rather than “Donate” or “Sign Up to be a Volunteer” was the most effective combination. They got far more donations and offers of volunteer help as a result of their optimization process.

What’s interesting about that, I think for me, is the contrast between research and what people say. You make an important point that sometimes they’re not even aware of what they know. When you test it in a rigorous scientific measured environment, you can often get different reactions between what people say and what they do.

I wondered whether in your level of research, when tested, are there ever surprises? You think, “How did we discover this in research?” But that conclusion is confounded or refuted when we actually test in a scientific experiment.

A. Yes. I think that’s an interesting point. I do think context in which the message is received is hugely important. There’s a world of difference between a laboratory based and a real life based study. This is why we undertake ambulatory work, that is people actually walking around a supermarket or mall, despite the technical problems of getting clean readings. People are moving around, they are experiencing stimuli in the actual environment where, for example purchasing decisions are made.

Apart from artifacts, the other problem with ambulatory readings is if you get a spike of attention on the EEG, you can’t always be certain of what produced it. The shopper is examining a new product and they show an increase in attention as measured by the EEG trace but what caused their interest to peak?

There are so many confounding variables. Sounds and sights and maybe something that caught their eye out of the corner of their eye that they looked at. You don’t really know what’s going on in their surroundings. Certainly if the spike was consistently present across a number of participants that would be more significant in terms of the product, package design and so on. We also have them wear camera glasses, which provide a continuous video, and audio recording of the shopping trip together with a time code we can use in conjunction with the EEG trace that helps use identify the most likely source of their increased interest.

In the lab, you can reduce the confounding variables. You can make a very much more sterile environment, but how realistic is that sterile environment in contrast to the actual messy, noisy, busy, buzzy environment in which most people shop and I guess go online, as well, if they’re working in a busy office or something like that?

There are a lot of things happening where their senses are being aware of which they’re not necessarily being so aware of in a laboratory, which for most people are very unfamiliar and perhaps slightly daunting environment. I think to an even greater extent, if you’re using an fMRI scanner, which is why we don’t think they’re of any use at all in consumer research. A lot of people would argue against that.

I don’t know if you’ve had a brain scan or a whole body scan, but you’re lying inside basically a large dustbin, which is very noisy. Somebody describes it as like lying inside a dustbin and being kicked around the yard by people who are wearing hobnail boots. It’s a pretty aversive environment, even if you’re not claustrophobic.

You have to lie absolutely still or you’ll blur the picture. You’re seeing perhaps a television commercial or there’s a stimulus on the small monitor directly above your head, as you’re lying flat on your back. Come on, people don’t shop like that.

I think there is a conflict between actually getting the real environment with everything going on and then not really being very easily able to untangle all the confounding variables and doing a more sterile environment where you’re putting in things, which are not truly relevant to the actual thing, which is going on.

So I think you have to take a lot of different looks at different ways of assessing this data, and it’s very difficult. I think a lot of companies manage to sell a lot of clients on the idea that it is actually basically just easy. Stick some electrodes on somebody’s head, and you can tell what they’re thinking, and that is just absolutely so wrong. A better approach is to mimic the environment in which real decisions happen as closely as possible. 

Contextual validity

Q. Would you agree that split testing on a website, you are reducing the factors because they are likely to be looking at a screen? If it’s a desktop, they might be sitting down, they’re in a controlled environment, and they’re not outside, and so on. The results, you would think, are contextually valid.

A. I think this is very likely. When you’re looking at things which happening online, the environment is very similar to the testing environment. I would think that is a huge plus to really understanding how to create the kind of message and the various subcomponents of the message.

We did a study for a charity and created a mock charitable site where we displayed various endorsements.

Some were delivered by celebrity (or apparently by celebrities) and others were by friends on Facebook and things like that to see would a big name celebrity persuade you more than somebody you met on LinkedIn or Facebook and this kind of thing. What they were interested in was which works best. In those studies, it is really nice because it was a nice, clean study and we were much more confident of the data we got.

People have a very special relationship with computers and programs. They anthropomorphise them. Let me give you an example from the early days of computer. We’re going back some forty years to when computers were primitive compared to today.

ELIZA was a computer program that was supposed to act as a therapist. It would pick answers from a menu. You would go switch on the computer and ELIZA would ask you to type in your name: Let’s say, David Lewis. It would then print out: “Hello, David. What’s your problem?” I reply, “My father doesn’t love me.” Then it would ask, “Why do you think your father doesn’t love you?” I might type. “He always forgets my birthday.” To which ELIZA could respond, “Why does he forget your birthday?”

As I say, it was very simple-minded. It mimicked what was called Rogerian therapy, which was giving you unconditional positive regard and essentially reflecting back to you what you just said. The point of it is people got immensely involved with this program. They actually believed that the program understood them and was responding in a positive way to their problems, and they felt it actually had helped them.

On one occasion, Weizenbaum, the professor who compiled the program went into his office and his secretary said: “Excuse me, professor. Do you mind going out a minute? I’m just talking to ELIZA.” She really felt she was talking to a real therapist.

I think all the research suggests the people will respond very well to computer programs. The thing about a computer program is it never takes no for an answer, and it can be made to go personal in regard to your likes and dislikes. Also, it can get the timing right.

It doesn’t mean that we’re all going to be taken over by artificial life, but I do think it’s got a lot of benefits, which maybe in our conceit about carbon-based life forms being superior to silicon-based life forms, we sometimes don’t acknowledge.

I conducted a study while researching my latest book Fat Planet: The Obesity Trap and How We Can Escape It in which I counted the amount of popcorn eaten by a cinema audience as they watched a film. People will always pick the popcorn from the box with their dominant hand, the right hand in most cases. I then prevented this by having them wear an oven glove on their dominant hand, typically the right hand, obliging them to eat with their non-dominant hand. Because they had to think about what they were doing the amount of popcorn consumer was cut by about two-thirds.

It comes back to my point about the ease of any kind of transaction. Anything which interrupts the cognitive flow where you actually have to think about it, where you’re not doing it almost in a mindless sense, is going to make it harder for that message to get across and that message to be acted upon.

Research from eyetracking studies

Q. One question I had, which relates to eye tracking. In your book, you show how you use eye tracking to look at I think it was male and female bodies. We use a tool called EyeQuant. I don’t know if you’ve come across it.

A. Certainly when we’ve done studies of commercials and where people are looking where to put the price point, not just on a commercial but maybe on a webpage as well.

Eye tracking suggests that you need to be very careful where that goes because people are not very often going to look at that. They’re going to look at the other things of greater interest, which will be people’s faces, people’s actions, and things like that.

We’ve done studies of labeling of food, for example, and found that very few people actually ever look at the nutritional content of processed food. We did a study looking at various kinds of processed meats, and the people looked at the picture and they looked at the type of food and this kind of thing. They very rarely looked at the nutritional information, which was there in clear sight but yet not particularly noticed.

Just because something is not particularly noticed, bizarrely, doesn’t necessarily mean that it’s not going to be acted upon. There was an interesting study in Germany where people went into a butcher shop and were offered samples of different kinds of cooked meats.

What the researchers did was they put a small poster on the glass door of the butcher’s shop that was advertising the recipe for a low calorie meal. Although most people didn’t look at the poster (at least they didn’t pay attention to it, it seemed) when it was present people consumed less meat. They ate fewer calories.

There’s another very interesting study in America, which again, you can read about it in ‘Psychological Science’, where they showed that the presence of the American flag, even very unobtrusively positioned, actually affected people’s voting habits.

Again, in another study they found that a fast food logo placed in the laboratory, influenced participants in a study that had nothing to do with fast food. No attention was directed to the logo, it was just there in the environment, yet people were found to be less patient in its presence.

The ability of things we never consciously notice to influence our behavior is a subject of great interest to me and one I am actively researching at the moment.

The right environment for desired behaviour

Q. In your book The BRAIN SELL: When Science Meets Shopping you discuss experiments you’ve done with people with their arms bent and arms straight and how they consumed more of the drink when their arms are bent than with their arms straight.

Do you think that that may have any bearing on how people behave when they’re buying on a computer, whether their arms are going to be more straight, versus buying on a mobile, where you tend to have your arms bent because you’re holding it a bit close to your face. 

A. I don’t know. We haven’t done that work. It’d be very interesting to see because a lot of work has been done by us and by other people looking at how people respond when they’re shopping with a trolley or a basket. 

Work has been done where people have been working with weights and expressing preferences for things. When they’ve been pumping weights where they’re pushing the weight above the body, their arm extending, they show less liking than when they’re actually pulling the weights towards them, doing a curl, for example, rather than a bench press.

Another very interesting study found that when people were holding a hot mug, they actually liked people better. They had a scenario in an elevator where the woman or the man would ask somebody to hold this cup of coffee for them for a moment while they did something. They found that just holding the warm cup briefly increased liking for them.

Why should this be?

It arises due to our survival needs, as infants, for warmth. During the early weeks of life these needs are met by clinging to our mother. By the age of five the association between feeling warm and secure have become so tightly intertwined that we no longer distinguish between emotions and physical sensations.

As adults we use a lot of caloric phrases to describe things we like and dislike: “I’m warming towards you”; “She’s hot stuff” and “He’s warm hearted,” for example. Whereas when we use phrases such as: “He’s a cold fish”; “I gave her the cold shoulder” and “The atmosphere was icy” to emphasis our dislike for someone or some situation. We also tend to like people better at first meeting if the encounter takes place in a warm rather than a cold room.

When monitoring shoppers in places such as supermarkets and malls we find customers respond much more positively to products viewed in a Goldilocks environment. That is neither too hot nor too cold, too moist nor too dry.

Light levels too are important. Under bright lights emotions are felt more intensely. In the brighter room participants wanted spicier chicken wing sauce, thought the fictional character was more aggressive, found members of the opposite sex more attractive and felt better about positive words and worse about negative ones.

When you enter a supermarket or mall you first go through what is called a decompression zone. When you come off the street or off the car park into the shop, you don’t immediately get into the shop. You go through an area where they’re actually not trying to sell anything, where there may be some charity displays or a Photo-Me machine or toys for kids to play on, and then you get into the shop.

That’s the decompression zone. It allows customers to adjust the speed of their walking from a fast pace across a car park or off the street to a more ambling place to go around and graze in the shopping center. Their bodies are adjusting to changes in humidity and temperature. Their eyes are adjusting to changes in lighting. These decompressions zones, I think, are very important. Just an indication of how subtle some of the influences are.

The same thing, I think, applies online, obviously not in terms of temperature and things like that. The online retailer can’t control that, obviously. Nonetheless, I think in terms of coming back to this idea of cognitive fluency. The smoother the message, the smoother the pathway through the site, the more likely it is to result in a sale. Anything which interrupts that, where you have to pause and think or get aggravated because something doesn’t work properly, they turnoff, abandon the trolley, go on to another site.

Cognitive fluency and dissonance in long lifecycle product purchases

Q. Thinking about that cognitive fluency, how do you deal with products that have a longer sales cycle? Do you have any information about how people process motivation and urgency and things like that?

A. I take your point entirely. What I’ve been talking about so far, I think, are pretty much in a sense impulse purchases. I think when you need to take more time and consideration, probably I would say there are three things which our research suggests are very important.

Firstly is the authenticity. We found authenticity to be a hugely important component of people’s ability to trust a product or to trust a producer of a product or producer of a service. Do they seem to be authentic? Do they actually seem to live the work they’re doing, live the service, live the product?

Is the product in some way different, unique, and special? People like to have this idea of something that is special. They feel they’re belonging to a special club. Apple, I think, would be a very good example of this where we’ve done research with looking at people queuing through the night to buy the latest Apple product.

They’re investing an awful lot of time and effort into this notion of cognitive dissonance. Are you familiar with cognitive dissonance? If you put enough effort into something, you’ve got to value it. Otherwise, why are you wasting your time?

The more invested in a product, the more likely they are to think that that money is going to be well spent and that it must be a superior and a more authentic product. Otherwise, they have to see themselves as, in a sense, somewhat foolish to invest that amount of time in it.

Authenticity and clarity are very important. They help engender feelings of trust in the consumer. We’re a very mistrustful population now. Again, I think this comes back not just to the actual service itself but also to after-sales service.

Amazon, in my experience, is very good. If you have a problem, they will sort it out, eBay similarly. Customers with complaints are extremely valuable because they’re providing some good feedback and some honest feedback in most cases. Also, if you can win back a consumer who in some way is disgruntled with you, that’s likely to be a very loyal consumer from now on.

We know that as consumers we’re all just a series of digits on a computer somewhere deep in the heart of Texas, but nonetheless we’d like to believe that we’re seen by our suppliers as unique human beings and very special people. When customers believe, or can be persuaded to believe, that’s how you see them then they are much more likely to buy from you.

Keep up-to-date

People from Facebook, FarFetch and RS Components receive our newsletter. You can too. Subscribe now.

Interested in turning experimentation and testing into an advantage for your entire business?