There is a sentence I hear all the time during website projects:
“When will the website be finished?”
And obviously, there needs to be a launch date.
There needs to be a point where designs are approved, development is complete, redirects are tested and everyone can finally stop staring at staging links.
But I think we need to stop treating launch day as the moment a website becomes finished.
Because launch is actually the first time we get one very important thing:
real users.
Before launch, we’re making educated decisions.
After launch, we have evidence.
And the website industry is finally starting to build around that idea.
Website platforms are quietly becoming optimisation platforms
Look at what’s happening with modern website tools.
Webflow now has Optimize, allowing teams to create variations, test experiences and target different audiences.
Framer has built A/B testing, conversion funnels and click tracking directly into its analytics ecosystem. Its current A/B testing guidance specifically recommends focusing experiments on pages with clear conversion goals such as form submissions and link clicks.
That’s quite a big philosophical change.
Website builders used to help you publish websites.
Increasingly, they’re helping you improve websites after you’ve published them.
And I think marketing teams need to change their mindset with them.
The redesign-every-four-years model makes absolutely no sense
This is still how loads of businesses manage their website:
Launch new website.
↓
Everyone celebrates.
↓
Nobody meaningfully touches it for three years.
↓
Marketing starts complaining that it isn’t generating enough leads.
↓
Everyone decides the website is “outdated.”
↓
Spend £20k rebuilding everything.
↓
Repeat.
It’s such a weird system when you think about it.
Imagine running paid advertising like that.
You wouldn’t launch an ad campaign, refuse to look at its performance for three years and then completely rebuild the campaign because it “feels dated.”
You’d look at the data.
Change something.
Measure it.
Learn.
Improve it.
Websites should work the same way.
Your first homepage is a hypothesis
This is probably the biggest mindset change I’d like businesses to make.
Your website design isn’t the answer.
It’s a hypothesis.
We believe this headline will communicate the value proposition.
We believe this case study will increase trust.
We believe this CTA will encourage enquiries.
We believe users will understand this navigation.
We believe this form asks for the right amount of information.
Strategy, UX research and experience can make those hypotheses very good ones.
But they’re still hypotheses.
Once real users arrive, you get to find out whether you were right.
That’s where CRO becomes much more interesting than endlessly redesigning things based on opinions.
Stop changing websites because someone in a meeting “doesn’t like it”
This is another reason I love the idea of continuous optimisation.
Website meetings can become ridiculously subjective.
Someone doesn’t like the wording.
Someone wants the button bigger.
Someone thinks the hero should have more information.
Someone’s CEO wants a carousel because they saw one on another website.
And suddenly the website is being redesigned around whoever speaks loudest in the meeting.
Instead, ask:
What problem are we trying to solve?
If the concern is:
“I don’t think people understand what we do.”
Great.
That’s testable.
Try a clearer value proposition.
If the concern is:
“People aren’t finding our case studies.”
That’s testable.
Move the case studies higher or change the navigation.
If the concern is:
“The enquiry form feels too long.”
Again: test it.
A/B testing doesn’t remove creativity or expertise.
It stops opinions pretending to be evidence.
But please don’t A/B test everything
This is where CRO can get silly.
Changing a button from slightly rounded to very rounded probably isn’t transforming your pipeline.
Neither is obsessing over whether your CTA says:
“Get Started”
or
“Let’s Get Started.”
Your experiments should be attached to meaningful business questions.
For example:
Does showing pricing guidance increase qualified enquiries?
Does putting a relevant case study beside the enquiry form improve conversions?
Does a problem-led hero outperform a service-led hero?
Do visitors respond better to “Book a call” or a lower-commitment website audit?
Does reducing the number of form fields increase enquiries without destroying lead quality?
Those experiments teach you something about your customers.
That’s the point.
Marketing teams need an experimentation backlog
Instead of having a random list called “website changes,” create an experimentation backlog.
Every idea should contain four things:
Problem
What are we seeing?
Example: lots of service-page traffic but very few enquiries.
Hypothesis
What do we think is causing it?
Example: visitors don’t have enough evidence that we’ve solved this problem before.
Experiment
What are we changing?
Example: introduce an industry-specific case study immediately after the service explanation.
Success metric
What would tell us it worked?
Example: increase qualified enquiry conversion rate from that page.
Now your website roadmap is connected to actual business outcomes.
Not aesthetics.
Don’t optimise for form submissions and accidentally destroy lead quality
This is especially important for B2B businesses.
Imagine your current website generates:
20 enquiries → 10 qualified → 4 sales conversations.
You shorten your enquiry form and suddenly generate:
35 enquiries → 11 qualified → 4 sales conversations.
Your conversion dashboard might celebrate.
Operations probably won’t.
You’ve created 15 additional enquiries for someone to process without creating any additional commercial value.
The better metric isn’t always:
Did more people submit the form?
It might be:
Did more of the right people submit the form?
That means website analytics shouldn’t end at the thank-you page.
Where possible, connect website activity to CRM outcomes:
Website visitor → enquiry → qualified opportunity → proposal → customer.
That’s when CRO becomes commercially useful.
Your highest-traffic page might not be the page worth fixing
Another mistake is automatically optimising whatever gets the most traffic.
I’d prioritise pages using a mixture of:
Traffic × Intent × Commercial value × Current performance.
A blog receiving 8,000 visitors might generate very little commercial intent.
Meanwhile, a service page receiving 400 visitors might influence £100,000 worth of pipeline.
Guess which page I’m looking at first?
The closer a page sits to a commercial decision, the more interested I am in understanding its performance.
For most B2B businesses, I’d start with:
Service pages.
Pricing pages.
Case studies.
Contact/enquiry journeys.
High-intent landing pages.
Then work backwards through the funnel.
AI makes experimentation easier — but also more dangerous
This is where things get interesting.
Framer is already pushing AI Agents deeper into website creation; its July 29, 2026 update expanded Agent support across accessibility attributes, interactions, CMS video, page effects and more.
Generating another landing-page variation is becoming ridiculously easy.
That should increase experimentation.
But it also creates a new problem:
Just because AI can generate 15 variations doesn’t mean you should test 15 random ideas.
The bottleneck is shifting.
Creating the variation isn’t the difficult part anymore.
Knowing what is worth testing is.
Strategy becomes more valuable, not less.
Small businesses don’t need enterprise-level experimentation
You also don’t need 100,000 monthly visitors and a dedicated CRO team to adopt this mindset.
If your traffic is too low for statistically meaningful A/B tests, use other evidence.
Look at:
Search Console queries.
Form abandonment.
Sales-call questions.
Customer interviews.
Heatmaps.
Session recordings.
CRM data.
Support tickets.
On-site search.
Repeated objections.
Then make deliberate improvements and measure performance over a longer period.
Not every optimisation needs to be a laboratory experiment.
The important part is that there is a reason behind the change.
Build websites that are easy to change
This also changes how I think websites should be developed.
A marketing team shouldn’t need to raise a developer ticket every time they want to test:
A new hero.
A different testimonial.
A landing page.
A CTA.
A case study.
A campaign message.
That doesn’t mean giving everyone permission to destroy the design system.
It means building flexible components, CMS structures and guardrails that let teams safely iterate.
When we’re choosing between WordPress, Webflow, Framer or another platform, this should genuinely be part of the decision.
Not:
Which platform is coolest?
But:
Which platform allows this organisation to manage and improve its website properly after we leave?
That’s a much better question.
The website retainer needs a rethink too
This shift has implications for agencies.
Traditional maintenance retainers often focus on:
Plugin updates.
Backups.
Security.
Bug fixes.
Those things absolutely matter.
But that’s website maintenance.
There’s another category businesses increasingly need:
Website growth.
Every month:
Review performance.
Identify friction.
Choose an opportunity.
Make an improvement.
Measure the result.
Feed what you’ve learned into the next experiment.
Instead of paying an agency purely to prevent the website getting worse, you’re paying them to help it become better.
I think we’ll see much more of this over the next few years.
What I’d do this month
Don’t redesign your website.
Pick one commercially important journey.
Maybe:
Google → service page → case study → enquiry.
LinkedIn → homepage → service page → call.
Paid campaign → landing page → demo request.
Then follow it from beginning to end.
Where are people disappearing?
What questions aren’t being answered?
Where is trust missing?
What assumptions are you making?
Choose the biggest friction point.
Create one hypothesis.
Make one meaningful improvement.
Measure what happens.
Then do it again.
Your website should be better six months after launch
This is ultimately the standard I’d use.
If your website launched six months ago and you’ve learned absolutely nothing from the people using it, something has gone wrong.
You should understand your customers better.
Your messaging should be sharper.
Your case studies should be stronger.
Your conversion journeys should have fewer unnecessary steps.
Your highest-value pages should be performing better.
Because the website you launch is built using everything you know at the time.
The website you have six months later can be built using everything your customers have taught you since.
Launch isn’t the finish line.
It’s when the useful data finally starts.
Discover more from Raccoon Web
Subscribe to get the latest posts sent to your email.
