A charity website that scored 14 on PageSpeed, and the parents who could not find it
PageSpeed 14 to 100, a dead application form that had been collecting 212 searches a quarter into a 404, and ten new pages written for the families the charity exists to help.
The numbers
The problem was not the one anybody expected
Buttercup Children's Trust is a national UK charity, registered in England and Wales and in Scotland, that funds equipment for children under 18 with life-hindering illnesses and disabilities. Specialist beds, specialist seating, wheelchairs, sensory equipment, hospital travel costs.
It had a website that scored 14 out of 100 on Google PageSpeed for mobile, which is roughly where a site ends up when nobody has looked at it for a few years. That was the visible problem, and it was the easy one.
The real problem showed up in Search Console. The charity ranked for its own name and almost nothing else — and the pages it did have were written for regulators and funders rather than for families. Complaints procedures, terms and conditions, partnership enquiries, a photo gallery. Out of fourteen pages, exactly one was written for a parent looking for money for a piece of equipment: the application form.
So when a mother in Norwich searched for a charity that funds specialist beds for disabled children, Google had nothing on that site to show her.
What we found in the index
Three things were actively costing the charity applications, and all three had been there long enough to be invisible.
A dead application form was ranking. A PDF at `/files/grant-application-form.pdf` was drawing 212 searches a quarter and returning a 404. Every one of those was a family looking for the form and finding nothing.
The homepage was indexed twice. `/index.html` returned a 200 alongside `/`, so the site was competing with itself and splitting its own ranking signals between two URLs.
HTTPS was not enforced site-wide. Beyond the obvious, this quietly disqualified the charity from Google Ad Grants, which is worth about £7,900 a month in free advertising to a registered UK charity.
We fixed all three: 301 redirects at the Cloudflare edge for the PDF and the duplicate homepage, and HTTPS enforced with TLS raised to 1.2. Because those live in Cloudflare rather than in the site files, we documented them — an undocumented edge fix is one that silently disappears if the site ever moves host.
We let the charity’s own data write the content plan
Rather than guessing what to write, we read the charity's application database: 816 applications and 80 enquiries, aggregated, with nothing individual leaving the room.
The applications and the enquiries rank needs differently, and the difference is informative. Enquiries — people asking before they formally apply — are dominated by specialist beds, mobility equipment and seating. Applications — the fuller ask, once a family is engaged — put respite and therapy near the top.
Ranking both funnels together produced the content plan: specialist beds, specialist seating, mobility equipment, sensory equipment, travel costs, therapy, and respite. Specialist seating was second in applications and third in enquiries, and it had not appeared anywhere in our first draft of the plan. The data caught it; we would not have.
We wrote eleven pages against that order, in the words families actually use, each one a plain answer to a question somebody types at eleven at night.
A contradiction worth stopping for
Respite was the single most-requested thing in the application data — 35 mentions. It should have been the flagship page.
Except the application form said, twice, that the charity could not fund family holidays, while the appeals page was actively fundraising for one. We stopped and asked rather than writing around it.
The answer turned out to be a real distinction that the site had never made: the charity does not give money towards a holiday, but it has access to caravans on the east coast and offers families the break itself. The form was not wrong. It was incomplete, and it had been reading as an absolute no to the most-requested thing the charity does.
That is a content problem worth more than any amount of keyword research, and it was only findable by refusing to publish something that did not add up.
Making the donations visible
The charity took donations through CAF Donate, which is sensible — CAF handles Gift Aid, worth 25% on top. But the donor left the site and never came back, so nobody could answer the most basic questions: how many people reach the donate page, how many click through, how many finish.
We added outbound click tracking on every link to a donation processor, and built a thank-you page to set as CAF's return URL, so an arrival there is a completed donation. Two events, on the analytics the charity already had.
This is also a hard requirement of Google Ad Grants. Accounts without conversion tracking get suspended, and it is the most common reason a charity loses a £90,000-a-year advertising grant in its second month.
Honestly
The traffic results are not in yet — the content shipped in September 2026 and search takes months, so anyone quoting ranking improvements this early would be making them up. What is measured today is the technical work and the reach of the fixes: 14 to 100, a 404 that no longer catches 212 searches a quarter, and a donation funnel that can now be seen. We also found the charity is not recording grant amounts — only nine of 816 applications hold a figure — which means it cannot yet publish an impact number that would survive a funder checking it. We told them that rather than helping them publish one.