How to tell if your page is good
On The CRO Podcast, we cover how to judge a landing page, plan tests that give a clear result and use AI sensibly.
You can have the right audience and a healthy ad budget and still lose people the moment they land. Most businesses we speak to sense their page could do better, but have no clear way of knowing.
This episode of The CRO Podcast was just the two of us, tackling the questions we’re asked most: how to tell if a page is any good, whether to test one change or several, and which tests are worth running.
We kept coming back to one idea: a page is good if it does the job you need it to do, and making it better takes a clear hypothesis, a test big enough to read and a human eye on the details.
Here are five lessons from the conversation, finishing with two real landing pages we reviewed on air.
A good page is one that does its job
Courtney’s answer to “is my page any good?” was to reframe the question: a page is good if it does what you need it to do. If you’re getting enough high-quality leads to keep the business healthy, it’s probably doing its job.
That doesn’t make it finished. Courtney compared CRO to swimming with flippers: if a competitor is optimising and you aren’t, they have an edge you don’t.
Send your page to someone who has never seen it and ask them to explain it back. You know the context, so ambiguity is hard to spot in your own work. It’s Will’s top tip, even if the feedback isn’t always fun to hear.
Start with data and heuristics, not your competitors
Competitor analysis is where a lot of people begin. Will’s view is that your own data and a heuristic review are a better start, because a competitor’s page sits in a context you can’t see.
A heuristic review looks at the page the way a visitor does when making a snap decision, mostly above the fold. Courtney listed the checks people run without realising it:
- Relevance. Is this what I expected when I clicked?
- Trust. Do I trust this place on the internet?
- Time. How long will this take?
- Effort. How much work am I being asked to do?
- Value. What do I get, and is it clear?
Will framed this as risk management: visitors look out for risks, and your job is to reassure them. He also reads bounce rate by source: high bounce on cheap traffic may not matter, but on organic traffic it suggests the page isn’t what people expected.
Bundle small changes into tests big enough to measure
Our audits end with a master test pipeline of every change we’d make. It can easily run to 50 items, and most companies don’t have the traffic to test each one alone.
So we group changes that serve the same hypothesis, such as a set of trust changes. Courtney’s rule of thumb is to aim for an impact of around 10%, a size you can see in the data.
Many small tweaks might lift conversion by closer to 5%. Will’s rough recollection was that detecting a 5% uplift needs somewhere between four and eight times as much data as detecting a 10% one.
These are rules of thumb, not guarantees. The data you need depends on your traffic, so check a sample size calculator first. A change only belongs in a bundle if it fits the same hypothesis.
AI review is a starting point, not a verdict
We’ve been training our own AI to analyse landing pages. It helps, but it still needs us to keep it in check. As Will put it, “It’s not a silver bullet.”
The main problem is generic advice. AI blends best practices from many industries and applies them without context, which is one of the most common mistakes we see.
Will’s example of a principle that isn’t universal: a very short form is often worse than a longer one, because you haven’t yet built the trust needed to ask for contact details.
Two paid landing pages show the cost of clutter
We ended by reviewing two PPC landing pages. These are opinions from a quick look, not test results.
The first, a London estate agent comparison page, we rated highly: a short, clear headline, a Trustpilot rating near the top, reassurance ticks under the CTA, and a CTA asking for your London postcode that echoed the headline. On mobile, the button sat in thumb reach and the form ran in steps with a progress bar. We’d still test turning its subtext into bullets, which Will says often wins (experience, not a rule).
The second, a hearing aid provider’s page, looked built by committee, with at least 13 calls to action by Will’s count. Its comparison table was hard to read on a phone, a problem Courtney flagged for an audience that may struggle with small text. It asked for contact details after one question. In Courtney’s words, “it’s like saying, I love you on the first date.”
Check your own page:
- Is there one clear call to action?
- Does the page build trust before asking for personal details?
- Have you checked it at phone size? Will uses iPhone 12 Pro and iPhone SE views.
Try this before you change anything
Send your landing page to someone who has never seen it. Ask them to open it on their phone and tell you, in their own words, what you offer and what happens next.
- Could they explain the offer from the headline alone?
- Did they see a reason to trust you before you asked for details?
- Was it clear how long the next step would take?
Write down every point where they hesitated. That list is the start of your test pipeline.
Get your free toolkit
Enter your email and we’ll send you a copy.