Keyword analysis revealed fundamental intent mismatch
The most expensive page problems might not be content, design or functionality. Pages can work but just for the wrong users.
Penny Appeal's Sadaqah Jariyah fundraising appeal was underperforming. On the surface, the page looked fine: clean design, clear copy, functional donation form. But Google Analytics data confirmed the bounce and exit rate was notably higher than other appeal pages, and the PPC manager suspected a page-level issue based on quality score data.
The default diagnostic approach in this situation is to look at the page: layout, copy, form friction, trust signals. Run heatmaps. Test variants. Iterate on design.
But there was a prior question: if users are landing and immediately leaving, the page might not be the problem. The mismatch between what the traffic came to find and what the page offered might be the problem. And if that's true, optimising the page has no bearing on the performance.
I downloaded the complete keyword list driving paid traffic to the Sadaqah Jariyah page — every search term users had typed before clicking the ad. Then I mapped these terms across core themes and analysed how each theme correlated with on-page behaviour (bounce rate, time on page, scroll depth, conversion).
The pattern that emerged was unambiguous. A significant proportion of traffic was coming from users searching for 'sadaqah' an immediate, one-off form of charitable giving. But they were landing on a page about Sadaqah Jariyah which relates to long-term giving projects (wells, schools, orphan sponsorships) with higher price points and ongoing commitments.
This was a classic intent mismatch. Users weren't bouncing because the page was poorly designed. They were bouncing because the page was answering a question they hadn't asked. Users seeking a simple one-off donation were effectively being told 'we don't offer that here.'
The solution became clear once the problem was properly framed: don't change the Sadaqah Jariyah page. Build a dedicated Sadaqah page for the traffic that's actually searching for one-off giving.
The new page was designed around three requirements that fell directly from the intent analysis:
First, explain the difference between Sadaqah (one-off) and Sadaqah Jariyah (long-term) clearly, without jargon so users understood what they were choosing between. Second, offer a flexible, low-entry donation experience: any amount, any appeal, no long-term commitment. Third, give users agency over where their donation went, removing the friction of a single appeal with a fixed price point.
Collaborated with the design team to translate this intent architecture into a page that felt welcoming for small acts of giving rather than structured, high-commitment projects.
The new page generated over £500k in donations and became the second highest converting appeal across more than 18 live appeals.
The secondary outcome was equally instructive: the original Sadaqah Jariyah page also improved after the new page launched. Why? Because users not seeking long-term giving were now filtered to the right destination. The traffic quality on the Sadaqah Jariyah page increased, which meant the people landing there were more likely to convert because their intent matched the offer.
This compounding effect of how solving one problem improves two outcomes simultaneously. You're not just fixing the underperforming page. You're also improving the page it was cannibalising traffic from.
The standard CRO playbook would start with the page: run heatmaps, test layouts, change copy. But when users are exiting and bouncing from a page, the prior question is 'why did they land here in the first place?'
Keyword intent analysis is well-established in SEO and PPC contexts but underused in CRO. When you reverse the diagnostic flow and start with the traffic, then work backwards to the page you can find that the 'conversion problem' is actually a 'traffic problem.' The page isn't failing. It's being asked to serve an audience it wasn't designed for.
The other principle at work here was precision in problem definition. 'The page has a high bounce rate' is a symptom. 'The traffic expects one thing and the page offers another' is a diagnosis. The former leads to layout tests. The latter leads to a new page and £500k in revenue.
If I were to approach this today, I'd pair the keyword analysis with session recording review filtered by bounce sessions. Watching a handful of users land, scan, and leave would surface the micro-moments where intent misalignment becomes visible .
Transferable principle: Before optimising a page, understand what the traffic came to find. High bounce rates are often intent mismatches, not design failures. Fix the match, not the page.