top of page
  • Writer's pictureSally Khallash

Online Behaviour Insight #1 - Beauty

This is the first article in a series on how humans make online decisions - and how it affects your customer journeys and your pricing.

In this article we'll cover

  1. The effect of beauty on our first impressions

  2. How to measure beauty (we crunched 2.400.000 page views)

  3. Segmentation of 'beauty' across things like gender and culture

  4. Pricing Power implications of beauty.

  5. An algorithm that can measure the beauty of your website across customer segments.

  6. Offer to participate in our pilot programme on building Price-Beauty algorithms

Reading time: ... who cares! If the article is good don't you want it to be long? (2800 words).

First: why care about beauty?

Because: most customers spend 15 seconds or less on a site before leaving (according to Chartbeat). So your average visitor isn’t deciding whether or not she likes your website.

She simply reacts to it.

It's old news in decision theory that humans tend to put an extraordinary amount of weight on the first piece of information they get (that's why anchoring and priming works). And the result is that your customers trade dept for speed in online decisions.

Screens: making us either dumber or smarter.

The theory of the online world is that with fast access to information, your customers have an easy time comparing prices and products and buy the best product at the right price.

This, however, is wrong.

Or rather: it is right only with sophisticated buyers who:

1) have access to good information and comparison tools and

2) buys this good/service frequently and

3) care deeply about getting a good deal.

Miss just one of the three and you are not a sophisticated buyer.

For everyone else the information online becomes an overload factor and makes our online decision making significantly worse. In the digital world, it is much easier to slide into an instinctive way of thinking, relying and acting on our superficial first impressions.

A few examples:

  • If your clients have a 73-86% chance of doing a quiz on pen and paper, they only have a 54% success rate online.

  • When making decisions about financial wellbeing, your customers have a 57% chance of making the right decisions offline and only 45% if they use their smartphone.

So if you have highly competent, dedicated buyers with a lot of experience (like purchasing officers or single moms hunting for discounts on diapers) then first impressions will matter less.

But if your buyer is less sophisticated - like a CEO shopping for a brand-new technology or an engineer looking for someone to handle his pension - then first impressions have a huge impact (in both cases because both the CEO and the engineer lacks direct experience in this purchase).

So the first Key Takeaway:

Key takeaway #1: screens make sophisticated customers MORE sophisticated - and everyone else a whole lot dumber.

Fun fact: the smaller the screen the larger this effect - people make a lot worse decisions on their smartphones than on desktops (tablets are predictably in between).

SO: what makes a first impression?

This raises the question: If first impression forms customers’ online judgments, what goes into these first impressions?

The answer is rather intuitive to most of us: it’s beauty.

When customers rate a website in just 15 seconds, there isn’t much time for immersion. Instead, they make many of their decisions based on the aesthetic look of the site.

In computer science, it is called "visceral beauty" - a clumsy way of saying that they react to how the appearance of something makes them feel.

It will be a difficult and perhaps confusing task to run around asking people what they find beautiful - and fortunately, we don’t need to.

Or rather: we ask 40.000 participants from 43 countries to look through 430 websites for a total of 2.4 million page views.

Quantifying Online Beauty

When you think about beautiful visual impression, what springs to mind? Pretty pictures? Nice colours? Anything else?

The most promising research on this (according to us here at Behavioural Strategy) was done at Harvard Computer Sciences and University of Washington where research teams reduced the bulk of online aesthetics to just two variables: “colourfulness” and “visual complexity”.


Let’s start with colour: numerous studies have shown that customers behave differently depending on the colours they are primed to on online shopping sites.

Leaning on that extensive body of research, the researchers at Harvard and Washington University measured colour in two ways: first colour intensity by measuring the amount of non-white pixels on the screen; afterwards range of colours was measured by analysing the percentage of pixels that were close to the sixteen standardized colours that are part of the HTML 4.01 specification.


Secondly, the researchers looked at visual complexity, which basically measures amount of information and how the information is presented.

It was measured by a space-based decomposition method that automatically identified each text and image area. Once the analysis was completed, the website was reduced to a number of distinct "content areas" that allowed the algorithm to figure out how systematic or tight the page actually was.

The algorithms automatically analysed web sites in terms of colourfulness and complexity to see how they fit into customer preferences.

2.400.000 ratings of visual appeal across 430 websites from nearly 40.000 participants spread across 43+ countries later, the results were frankly stunning:

Key Take Away:

Key Take Away #2: Colour and complexity explain 48% of the variation in customers preferences for a website.

What does this mean?

That is, half of our immediate response to a website, what we like and don’t like about it, can be explained by just looking at colour choices and complexity applied on the site.

How does it work?

Basically, a customer is exposed to a number of websites every few seconds - remember that the average customer stays on a site for 15 seconds or less - and then rates whether or not the site is attractive enough to browse or leave.

Then we calculate Colourfulness and Visual Complexity for each site, on a scale of 1-10.

The figure above shows how the variables are measured. It shows that the relationship between visual appeal and both complexity and colourfulness is an inverted U, and that visual appeal for complexity peaks at 4.23 and for colourfulness at 6.10.

So how do you interpret this?

First: most customers do not like extremely simple and colourless websites (I'm looking at you, MVP start-up with your 1-2-3-click-and-you’re-a-customer)

Second: they don't like extremely complex and text-rich websites either (financial institution, you know who you are) or the very colourful ones.

Similar to Goldilocks, customers like the visual appeal of websites to be just right.

Most of us working in corporate life have some experience hiring designers and creative agencies to develop advanced and beautiful website with the ultimate goal to sell product, service or message. And some of the design inspirations are quite creative, for instance a client of ours and Director at a conservative finance institute who disclosed in dismay that their website was originally inspired by a Berlin-meets-Goth theme.

But in light of the Harvard/Washington study, perhaps a lot of the stuff that designers think are important based on their intuition and gut feeling – and charge you a pricy fee to draft up – aren’t in fact that important at all.

It is this misunderstood confidence in irrelevant design features that might explain why the websites nominated for the prestigious 'Webby Awards', on average, were no more attractive than a randomly selected website.

Because the designers focused on less relevant design details, they ended up with poorer performing websites.

But – as you would interrupt by now – do everyone really like the same thing? Isn't the role of the designer to understand the particular end user and then model the aesthetics to that target group?

Yes. Or you could just run the statistics again.

Modelling for different customer demographics

Demographics strongly influences our aesthetic preferences. In fact, demographics is so influential that even rudimentary demographic information can dramatically improve the performance of your website.

For instance customer age. It turns out that customers older than forty years prefer more visually complex sites, in some cases with older customers rating complex sites as 60% more appealing than customers below 20 years. On average, your younger cliental prefers saturated colours and larger images, while the older cliental likes more text-heavy sites featuring many distinct text groups, but with less saturated colours.

When it comes to gender, the main difference is what colour males and females prefer. For men, visual appeal peaks at a colourfulness level of 5.8, while women give higher ratings to sites with a colourfulness level of 6.3. Men like sites with primary colours on grey or white background, while women favour sited that use more homogenous colour schemes and pastel shades.

What does this look like? Here are a few examples:

Level of education also has a say, especially when it came to colourfulness, as people with longer educations prefer web sites with little colour.

Finally, location and nationality also exerts a strong influence on which web sites customers prefer. For instance, customers from Mexico and Chile prefer web sites rated almost twice as complex as customers from Russia.

Denmark is located in the lower end when it comes to complexity and colourfulness, the US is in the middle of the pack and brits in the UK prefer much higher complexity and colourfulness levels.

What is more interesting, is that general cultural trends follow a predictable geographic pattern, so that countries that resemble each other share similar preferences. For instance the Scandinavian countries are all in the top 11 and Finland and Russia are closest to each other.

Here is each country's graph:

There are also indications of rural-urban preferences, with customers living in big cities preferring more modern web site designs with less colour and more streamlined layouts.

So far so good.

However, aside from the obvious implications of website visual appeal for your business, there are also a number of unintended sequelae that come in the wake of your customers’ aesthetic experience. It turns out, customers who like your site also tend to find it more user-friendly and trustworthy.

Beauty affects Usability and Trust - and pricing power.

Yes, aesthetics and visual presentation affect a wide range of unrelated factors, the most important being experienced usability and trust towards your brand.

Turning first to usability, it turns out that aesthetics greatly affect your customer's experience of user-friendliness. The more attractive your customer finds your website, the more likely is she to rate it as user-friendly - even when your website is functionally on par with uglier sites.

So while we once held a firm belief that actual usability qualities is what fuels online activities – for example, customers don't like to make lots of clicks to get from A to B – it turns out that a major part of your customers’ experience of user-friendliness is shaped by how attractive they find your site.

So next time you decide to funnel a truck load of cash into your customer journeys and UX, try the quick fix of improving the visual appeal of your site.

Key takeaway #3: Instead of focusing your online UX on removing hassle and friction - try adding beauty. Make your customer journey 'the scenic route'.

The next question is whether customers find pretty websites more trustworthy? There is no logical reason why they should, but the human brain is filled with weird and irrational quirks.

Key takeaway #4: If your customers find your site visually appealing, they are also more likely to think it worthy of their trust.

It’s the "halo effect" at play, leading them quickly and unconsciously to conclude that because they find a website appealing, the site is also credible.

An interesting fact is that the level of trust is completely disconnected from readily available online ratings objectively assessing the credibility of websites.

Case study - Danish Financial institutions

In a world filled with screens, the issue of trust is perhaps most relevant for the financial sector. In Denmark, almost 20 percent of all customers say have hardly any trust in their financial institutions. In the US, it’s double up to 40 percent of all customers.

While some of this mistrust is certainly deserved, it is also unfortunate - where there is mistrust, everybody loses. Suspicion and lack of trust lead to longer processing time, more resources gone into monitoring, more bureaucracy, less flexibility – with services becoming more expensive, frictional and less satisfactory for all parties involved.

We analysed the 10 largest financial institutions in Denmark across a blend of landing pages, running the algorithm of Complexity and Colourfullness.

When we analysed the landing pages of the 10 largest B2C financial institutions in Denmark (banks and insurance companies ranked by balance sheet) our results were outright depressing.

The sites had a “average colourfulness” of 5.11, far outside the optimal 6.10. This indicated that the Website was too colourless for almost every demographic group.

But this wasn’t their only problem: The site was way too complicated with a rating of 4.94, scoring way outside the 4.23 optimal level.

And you could see why. The sites were completely cluttered with words and text boxes and contained an average of 29 links per landing page compared to the optimal 14 links!

This is not unique to Denmark: An US-based colleague of ours analysed Bank of America’s site and came up with even more depressing results: Colourfulness of 4.09, complexity of 4.80, and with 68 links before you even started scrolling down.

Given that excessive complexity reduces visual appeal much faster than extreme simplicity, this is a tangible challenge for financial institutions.

Of course, it’s easy going at it at the financial sector when in reality these are quite common errors. The average B2C site is too colourless (5.17 instead of 6.10) and too complicated (4.80 rather than 6.10). Instead of using blocks of colour to help users navigate, too many rely on words and text boxes.

Pricing Power of Beauty

And now to the money: how are customers' willingness to pay affected by beauty?

If you ask 515.000 customers to rate their perception of a selection of online brands - some of which they were familiar with and some they saw for the first time - on a scale of 1 (Bad) to 7 (good) and also ask them to rate their willingness to pay the price the brand is charging for its goods/services, then you get this:

Now remember that we earlier established that this perception could be 48% attributed to the two visual parameters of Complexity and Colour - many of the 4300 websites researched were well-known online brands, so the 'new and old' selection matches at least roughly across the two data-sets.

This implies that just about half of the above pricing power effect from brand perception can be attained by visual engineering.

So if we take the purely visual impact of brand perception merging colour and complexity into one graph and then cross it with the data on willingness to pay, we get this picture:

So the implication here is that you are missing out on at least 15.79% of pricing power if you have a 'medium appeal' site compared to one that hits your target demographic spot on.

If your website is ugly (according to the algorithm) you could be missing out on much more.

And just to drive the point home: mis-match of pricing makes acquisition more costly, increases churn, reduces customer satisfaction, creates un-wanted customer behaviour (like cheating) and a million other issues besides the main one: that poor pricing means you make less money than you could doing exactly the same as you do now.

Now: There are still some big caveats in how these conclusions are reached. Less so in the direction and overall conclusion that beauty has pricing power, but considerably so in precisely how much and how easy it is to manipulate with just the factors of Colour and Complexity (se more on our pilot project just below).

Nevertheless: Key take away #5:

Key Takeaway #5: Beauty is a HARD item in your pricing strategy - not a soft one you should leave to 'creatives'.


You can do it like us: compile all the studies from Harvard, MIT and Washington and build the algorithm yourself (it's open source) and start improving it with your own key insights (for us that was pricing, which the original research doesn't touch).

  1. Study with the results here.

  2. Paper explaining the algorithm here.

It is easier than it sounds (and we're happy to offer guidance).

OR.... drumroll...

Best Deal in Town - Pilot Programme

You can also sign up to be a part of our pilot programme: as an extension of the above work we are working on improving the visual algorithms for pricing purposes and are looking for 5-10 pilot-project partners.

We want to run test groups of people looking at your website evaluating their perceptions of beauty and willingness to pay - across a variety of levels of complexity and color schemes.

You qualify if your business checks all these boxes:

  1. Danish company - or a Danish branch with a DK project anchor.

  2. Annual revenue of 50mio DKK or higher.

  3. You have a website.

  4. You display products/services on your website - including at least some pricing.

  5. You are willing to have (most of) the results published.

We are making the study not-for-profit, but not pro bono. The value for you is that the project will result in some in-depth insights into your customers willingness to pay and - perhaps - some validated recommendations for targeted price increases.

Plus, of course, a roadmap for optimising beauty on your online platform.

We always aim to offer full-scale service even in our pilot programmes - they are an integrated part of our offering and the place where we develop our strongest tools.

We will start the pilot programme ONLY if we get at least 5 participating companies.

Deadline is May 31st.

If you are interested then write me a personal email at : and I will send you the full information regarding the pilot programme.

I'm also of course happy to discuss details, answer questions, meet for coffee etc.


bottom of page