Experimentation Elite: Good copy doesn’t fit in a box

Published on April 28, 2022 by Josef James

Taken from my 2022 talk at Experimentation Elite

This talk was born out of the frustrations that come with being a wordist who works on his fair share of SEO projects in an agency setting. I work on a lot of website projects where the input of the writer tends to be… predefined. This is pretty shite, to be honest, so I decided to talk about how projects such as these should take a copy and experimentation-first approach. 

*disclaimer, I am aware that a lot of good copy could, in fact, fit in a box. But not all boxes are created equal.

*disclaimer no.2: no flame to any designer or developer, y’all do an awesome job. 

The TLDR (too long didn’t read)

  • Website foundations that don’t consider their visitors won’t perform.
  • Designs that restrict copy from providing value won’t inform.
  • Experimentation in your copy and design should become the norm.

How 96.92% of new websites come into existence 

Opinions. That’s it. 

Only kidding, but not really. A lot of decisions that go into the existence of a new website (in my experience) are based on opinions or assumptions. These opinions generally come top-down from our arch-enemy the HiPPO (highest paid person’s opinion). And even some of the more informed assumptions from Google Analytics can be misleading. 

“We’re data-led!” they’ll profess, “the bounce rate is too high” and “nobody contacts us”. These claims may be true enough, and there might just be something fundamental in the build that is hindering performance. But more likely than not, the decision for redesign is based on gut feeling. 

So, they take these ideas to a design and development team and vomit them all over the place, and before we know it, a new site is in the works. 

Enter the wireframes 

Projects like this involve wireframes and the infamous Lorum Ipsum text. The nonsense latinate placeholder words that basically mean: words go here. What words do you ask? Who knows. That’s not the designer or developer’s job. Those words should have been well mapped before now… right?


So these pretty looking designs and wireframes land on my desk and the task is: can we write up this series of pages? You can use the old website as a guide. 

“But why would I be using the old site as a guide if we’re building something new?”, You ask? Good question. Anyway, that’s not on the top of our list of niggles and itches here, there’s something else. 

Word and character limits. 

And I just want to point out that I’m no stranger to limitations. Ads and other mediums have limitations and I can manage that. But I tend to write a fair few variations for ads, so we can see the performance over a series of live ads and continuously optimise and iterate. 

It’s quite likely there will only be one landing page, one homepage and so on. So this limitation suddenly looks very bad indeed. It, in my opinion, could even be costly. What if we need more words/characters for that section? Or want an extended CTA button? Or on the flipside what if there is too much space for copy with placeholder images, videos and other assets. What if the information we need can all go above-the-fold? 

When copy comes last, it’s normally too late to make these changes, and we have to play by the rules. 

What most website build phases aren’t including

Nobodies collecting qualitative data – UX analytics and all that. Google Analytics is pretty good, it tells you the what and the how. There’s one key ingredient missing though: the why. And in the why is a lot about the customer/visitor/client/whoever. From a writer’s perspective, almost everything about who they’re writing for, and how they’re writing, is missing. 

These projects all have one thing in common:

  • Nobody asks the visitor. 
  • No experiments are run. 
  • No qualitative data is gathered. 
  • The words on the page are considered last.
  • No alternative messaging & design combos are trialled. 

A lot of weight is carried in the WHY

I want to know what people want the page to say. What do they need to know? What motivates them? What makes them anxious? What’s crucial for them to consider taking the next logical step? What is the next logical step? 

If we don’t gather UX data, don’t run surveys, and don’t trial different messaging – how will we find the sweet spot? How will we incrementally improve our messaging, our landing pages, our website?

Because the honest truth is that we don’t know what the optimal messaging can be for a page until we’ve asked our visitors, gathered a bunch of qual data and mapped out our copy. Sure we can revisit these things in the future, but why aren’t we just killing two birds with one stone? 

Your landing pages don’t have to be like a print ad, and go live and hope to have an impact. You can iterate on them. Again and again and again. 

A copy and experiment-first approach

Enter the copy and experiment-first approach. In my head, doing things this way means we can embed a culture of gathering data, testing and measuring into a new site build that will do more than just benefit your messaging. Trust me, this approach will benefit your copywriters to NO END – and you should care about your copywriters.

When any new website or page is being built it should consider these steps: 

  1. Qualitative research incl. UX analytics 
  2. Writing and then designing 2/3 priority page variations
  3. A/B and multivariate testing in a staged or live environment

And that’s it. 

Hey, a site build is a really complex project and throwing too much more into it may mean spontaneous combustion. In my experience, keeping it simple but making the biggest impact is most important here. With the additional steps above:

  • Every page will be better inform
  • Messaging and its accompanying design will be based on more than guesswork and industry best practice
  • We can validate performance with testing and make continuous incremental improvements


Thanks again to Jackie & Craig for putting the event together. It was ace. And if any of you reading this made it this far and want to continue the conversation – reach out.