Case study · US standards body · 2021 to present · name withheld
The site was never the bottleneck.
Our client, a US standards body whose name we keep confidential, has its own SEO and content lead. He brought the traffic. Our job was everything it landed on: one design system, shipped in 2022, that carried 446 pages and six languages for four and a half years before it needed replacing.
A standards body sells one thing: credibility.
The client has been auditing and approving manufacturers for decades, under a scheme aligned to ISO 17065 and recognised by eleven national authorities. Its buyers are quality managers comparing two or three providers before a board decision. The website is where that credibility is proved or lost, long before anyone picks up a phone.
Who did what
Worth stating plainly. The client runs search and content in house: their SEO lead owns keyword strategy and article production, and the growth in visitors is his work, not ours. We were hired for the site, and everything the site touches.
What we own
- ›Design and the design system
- ›Build, templates and page architecture
- ›Forms and conversion flows
- ›Conversion tracking and reporting
- ›Third-party integrations
- ›Translations and market variants
- ›Hosting, security and performance
- ›Publishing and daily upkeep
A content programme is only as good as the surface it runs on. This one ran for five years without the surface becoming the problem.
A broken plugin and a developer who stopped answering.
October 2021. The client’s approval verification tool had stopped working after a WordPress update. The developer who built it was not responding. For a standards body this is not a bug, it is a credibility failure: a manufacturer’s customer types in a reference number and the site cannot confirm the approval exists.
“I am still leaning towards building our own system. I do not like to be dependent on them to fix every little thing. The current system has some limitations.”
Founder and CEO, client · 21 October 2021
We rebuilt the verification tool from scratch rather than reaching for another plugin, and rebuilt the site around it. Priced at $400 and $7,000. Live in January 2022.
A system, not a website.
Before a single page was laid out, we defined the parts every page would be made from. Colour ramps rather than colours. One heading scale. Buttons specified by padding rather than fixed width, so the system would survive translation into Chinese, Spanish and, four years later, Korean.
The least visible decision in the engagement, and the one that paid for itself the most: the site went on to absorb 446 pages and articles over five years without a redesign.
The design system, 2021
Built in Figma and still governing the site five years later. Type samples shown in system faces, not the licensed originals.
Traffic grew ten times over. The site it landed on did not need rebuilding for four and a half years.
37,861 sessions in 2022 to 375,316 in 2025, across 446 URLs and six language variants, all of it running on the design system we shipped in January 2022.
What the site had to carry.
These figures are not our results, they are our brief. The client’s content work brought this volume through the door. The question for us was whether the site could take it without slowing down, breaking, or quietly leaking enquiries.
Sessions the site served per year
All channels. Acquisition driven by the client’s in-house search and content work.
Engagement rate
Share of sessions where the visitor stayed and did something.
Engagement rate reflects content quality and page behaviour together: whether a page loads quickly, holds still while it loads, and works on a phone. The 2023 dip is a traffic mix effect, not a site regression. That year’s paid search sessions engaged at 17.6% and dragged the average down. *2026 covers 1 January to 7 September and is not a full year. Source: Google Analytics 4, client property, traffic acquisition report.
| Year | Sessions | Organic search | Organic share | Paid search | Engagement | Tracked enquiries |
|---|---|---|---|---|---|---|
| 2022 | 37,861 | 13,081 | 34.6% | 12,211 | 54.4% | not tracked |
| 2023 | 164,416 | 89,988 | 54.7% | 44,951 | 39.3% | not tracked |
| 2024 | 248,612 | 201,547 | 81.1% | 8,301 | 66.9% | 873 |
| 2025 | 375,316 | 307,215 | 81.9% | 6,143 | 90.6% | 2,141 |
| 2026 part year | 226,265 | 186,425 | 82.4% | 4,796 | 90.7% | 1,367 |
Every one of those columns is also a load test. Twice the site started getting in the way: an oversized hero image in March 2022, layout shift in February 2026. Both were found and fixed within days, roughly 300 KB cut from article pages the first time, Lighthouse back to 98% or better the second. Not ranking claims. The site staying out of the way of someone else’s ranking work.
Nobody was counting enquiries until 2024
This one is ours. For two years the analytics tracked visits and nothing else. We built the conversion tracking, and the first report immediately changed a spending decision, it also showed that one floating contact button we had built the year before was producing 19 enquiries every 28 days.
“This is awesome insight, especially on the paid side. We were spending too much.”
Founder and CEO, client · 6 June 2024, on the first conversion report
Measured enquiries went from nothing to 873 in 2024 and 2,141 in 2025. The value is not the number, it is that the client can now see which pages and channels produce enquiries and which only produce readers.
One long form became a three step conversation.
For four years the quote request was a single embedded form. It worked, but it asked a manufacturer to start typing without telling them how long it would take, how many questions there were, or what they would get back. The rebuild replaced it with a wizard that answers all three before the first field.
What a retainer actually looks like.
Not a monthly report. A standing capacity to notice something, argue about it, and ship it, on a site somebody else is publishing into every week. A representative slice.
-
Publishing, fast enough to stay invisible
Articles arrive from the client as documents and go live as designed pages, typically within one to three days. A content programme dies if the person writing it has to queue. Ours does not queue.
-
Infrastructure move
Off shared hosting onto Kinsta with Cloudflare in front, then a Stripe payment page and an integration with the digital credential provider, so an approval could be verified without leaving the client’s own domain.
-
Complaint form with an anonymity switch
An accreditation requirement turned into a usable form. Ticking “anonymous” hides the personal fields but keeps the business address mandatory, because the complaint still has to be investigable.
-
A floating contact button, built in house
Chosen over a third party chat widget on load weight. A year later that single button was producing 19 enquiries every 28 days.
-
A 50% drop in enquiries, diagnosed the same day
Conversions had fallen almost half in two weeks. The cause was three contact forms that had been permanently deleted from the trash. Restored that day, followed by a standing routine of testing every form every two days. Traffic can arrive perfectly and still convert nothing if a form is quietly broken.
-
Six language and market variants
Simplified Chinese launched in November 2022 and now runs to 29 pages. Spanish, Mexico, Brazil, Chile, Colombia and Argentina followed. Korean is in progress for a South Korea market push. Same design system, same templates, no separate builds.
The system we built has reached its ceiling. So we said so.
By 2026 the homepage design file told the story on its own. The first version was 5,333 pixels tall. The fifth was 8,569. Five years of good content had been added to a structure designed for less of it.
“At the moment, the website has too much text and is very information-heavy. From a CRO perspective, the main impact will come from restructuring the industry pages. We will use these pages as part of a funnel, showing more targeted and relevant content to users.”
Project brief, April 2026
The new site is a Next.js front end on a headless WordPress back end, currently in QA. Industry pages become funnel entry points, the homepage opens with a path selector rather than a wall of proof, and the application becomes the wizard above. Two commitments came with it, both testable against the live sitemap: keep the URL structure, keep every meta tag. Both hold.
There is a real trade-off, worth stating plainly. A headless build narrows what the client can edit alone: posts and SEO fields stay in their hands, page level edits come through us. Agreed before the work started, not discovered afterwards.
“Firstly, thank you for everything you do for us. You have been instrumental in our journey.”
Founder and CEO, client · 14 June 2024
If your content team is good, don’t let the site be the weak part.
We take on a small number of retainers and stay in them. If you have people producing the demand and a website that keeps getting in their way, that is the conversation we are good at.
Founder
info@woofocus.com · woofocus.com
About the figures. Traffic and conversion data from the client’s Google Analytics 4 property, traffic acquisition report, read on 8 September 2026. The 2026 row covers 1 January to 7 September and is not a full year. Conversion events were not configured before 2024, so 2022 and 2023 show no figure rather than zero.
About attribution. Search and content strategy is run in house by the client’s SEO lead; the growth in sessions is the result of that work. Paid media is handled by a separate agency. Woofocus is responsible for the website itself. Quotations are verbatim from project correspondence, published with the client’s permission. The client’s name, sector detail and screenshots are withheld at their request.