Data-Driven Web Design

Could anything sound more dull?

Yet, I think this could be a 3rd and important theme of this blog and key to my attempts to improve my business.

Before I move on to what it is and what it solves, let me first recap on the other two main themes of this blog.

Agile working and the KISS principle

These two philosophies have helped me come a long way in understanding what can go wrong with web projects.

Agile reminds us that fixing the outcome of a project in advance can be problematic. Knowledge comes from the process and outside influences need responding to.

Keep it Simple, Stupid reminds not to distract from the clarity of the message or add unnecessary complexity that makes it harder to adapt to change and maintain.

What I feel has been missing is a simple way to explain what many of us are trying to do for clients. Something that distinguishes us from those selling websites as a more templated commodity.

So what is Data-driven design?

You will see different definitions. Here are two:

“Data-driven design can be defined as a decision-making approach to the design process that heavily relies on collected data about customers’ behaviour and attitude. Information about how customers interact with your design acts as feedback that informs you whether your design fulfils its purpose”.



“Data-driven design uses UX research methods such as surveys, usability testing, behaviour flows and tracking analytics on websites”.

I think of it as essentially a commitment to working with facts rather than opinions.

We often can’t escape the politics that creates “design by committee”, but we might be able to steer it to something more productive.

Data-driven design is not just number crunching. Qualitative data is equally important.

The information can come from an existing site or any evidence based research that can be used to test a hypothesis.

You might be thinking that agile methodologies already tend to employ this with it’s iterative testing approach and you’ll be right.

However, this is part of individual methodologies rather than in the agile philosophy itself.

As freelancers and small agencies we are less likely to get clients who will commit someone to being permanently active in a Scrum team.

Even if they do, this concept might help to explain the approach to those out of the loop.

Give us the problem rather than the solution

I’m sure everyone who has done client websites for a while has hit a point where the client wants something we feel will do them more harm than good.

It’s a dilemma.

How do we tell the person paying us their idea is not sensible without putting them on the defensive?

Saying we are the experts only makes us look the opposite.

Do we honour the slogan of good service and conclude the customer is always right?

Tricky! Our knowledge is the value we bring to projects. We may also find ourselves being held responsible for problems we did not flag up?

How does this happen?

Perhaps it begins with “I need a website”?

In 2011 I decided I didn’t want to build clients’ websites. I attended a course on Google Analytics aimed at business owners. I got booed for introducing myself as an inspiring web designer.

Those  on this course had sites built by web designers and they were here because they did not give them the business they expected.

What I learned was that none of the course attendees had asked for traffic and conversions. They got what they asked for:- a website. They came with the solution, not the problem. They were happy with their designers at the time.

Naturally, I found their view of our profession depressing, but equally their new solution to find out more about web traffic.

It’s one component in an online marketing strategy. How would these busy business owners find the time to join the dots together?

Strategies are confusing

Occasionally, I am very pleased with the chats I have had with clients.

They are nodding enthusiastically at my explanations of how the web is different to traditional marketing. How keyword and competitors research can impact on the traffic. How conversion techniques can overcome visitors bouncing and never returning etc, etc.

I think I have brought clarity to a multiple discipline subject that took me years to digest. Perhaps for a while.

When I stop talking a century of mass media culture kicks in and it is usually only time before suggestions come in that run counter to what I thought we had agreed.

When in doubt Keep it Simple

What I like about data driven web design is that it’s term sums up an ideal way of working. More memorable for clients than an online marketing strategy.

I suspect the term could be replaced with “results based” , “experimental” or “research based” design

There is a 2006 book (free to read on Google books) by U.S. Dept. of Health and Human Services called Research-based Web Design & Usability Guidelines.

The book is dated, but it seems to have the same aim of not having decisions made entirely subjectively or (equally as bad) based only on “expert” opinions.

Of course, expert knowledge is important, but if we are going a scientific route we should be seeking to falsify our theories.

Experts tend to end up with more questions than answers. But, at least they are good questions.

What does a data-driven web design approach look like?

As with agile, the philosophy is clear, but the methodology isn’t.

Personally, I like to visualise the design process as layers that make up an user experience.

I’ll expand on this more in other posts, but the basic overview could look like this:

1. Strategy

This generally starts with clarifying what the company does, for whom and what it stands for. We are essentially part of the storytelling process for visitors and the brand alike.

But that’s not much use unless it’s seen, so often my first concern is with where the traffic will come from.

We can look at any existing data and do keyword and competitor analysis to see what our best chances are in search.

With some industries what they offer may not be typically what enough people search for. Then, an emphasis on social media and offline promotion could be more of a priority in the spending of the overall marketing budget.

This acknowledgment offers a way to turn the client’s initial solution “I need a website” into a research question about what it needs to do in the overall strategy.

I’m keen to do this as I want to counter the assumption that I will try to get as much business from them as I can.

I do, of course, but on the basis of data based results rather than theory.

2. Structure

The data we have gathered could give us clues on how to structure the content.

Recently on a client project I noticed a familiar structure among competitors. Most seem to have followed a common website template structure of a hero section followed by 3 or 4 boxes that lead to other main pages.

To fill these in, most had structured the content into types of people their industry serves.
What the data told us was that literally no-one searches by type. They searched for what they are interested in.

There was no sensible reason to follow the same structure. If we did we would be letting a design tread lead the strategy not the other way around.

3. Web Copy/Messaging

With the structure in place (and the keywords we want to rank for), it’s probably logical to start writing the copy.

If someone was brought in, they would probably understand how copy needs to be visually grouped to be scannable and needs to take the visitor on a journey that encourages them to hit that “call to action” button.

If we do it ourselves there is no shortage of research and frameworks by people who do conversion experiments and copywriting.

We can use these with clients to turn their words into something that will have an impact on their target audience.

4. Visual design

This may not come entirely last as they may already be branding in place. If we are lucky that might have even been done as a strategy.

As there is no right and wrong with personal aesthetics, data driven decisions can really come into its own:

  • To help us focus on the target audience.
  • To show that good looking and instinctive design follows certain principles.

Once we have the words, content and basic structure on a page it’s mostly a case of adding emphasis to the key areas and reducing the cognitive friction for visitors.

We can do that from a data driven understanding of how humans process and group information. There is no shortage of scientific theory on the use of

  • Space, alignment and proximity.
  • Fonts and typography.
  • Colour.

Of course, this alone will not make up for the instincts of designers.

But there’s a way to make the process transparent so clients have a basis for their feedback and contributions and will not go altering work blindly.

I’m only scratching the suffice

My example process is simplified. I don’t even mention ways to get qualitative data on visitors.

In many ways there is nothing new here .

The very first “traditional” design process I experienced (over a decade ago when working for the UK government ) was based on trying to make what could be a subjective process into an objective and transparent one.

Also, for many of my clients there will not be much testing. The cost can’t be justified.

But the key thing for me is that more of my clients get a better understanding:

  • What we are aiming for professionally.
  • Why UX is much more than the look of a website.
  • Where we don’t need to reinvent the wheel each time.
  • How random of copying cool elements from other sites is a Frankenstein’s monster approach.

As always I am interested in your thoughts. Have I touched on common issues ? Do you have other ways of dealing with them?


Leave a Comment