Client under NDA. Market, product and timeline are real; figures are rounded and shared with the client’s consent.
The situation
A casino and sportsbook affiliate network operating comparison and bonus sites across four GEOs — the UK, Germany, Canada and Brazil. Over two consecutive core updates the network had lost roughly 60% of its organic traffic. Revenue per visit was falling faster than traffic, because the pages that survived were the ones with the weakest commercial intent.
The previous strategy had been volume: thin comparison pages generated from a template, translated across markets, supported by links bought in bulk. It had worked for eighteen months. Then it stopped working in a weekend.
The diagnosis
A forensic audit of what was lost, which competitors absorbed it, and what the survivors had that the network didn’t.
- The link profile screamed bought-in-bulk. Several hundred referring domains from two marketplaces, an expired-domain network with a visible footprint, and anchor-text ratios no natural profile has.
- Topical depth close to zero. Four hundred pages that each answered one query badly, no clusters, and internal linking that followed the CMS template rather than search intent.
- Translated, not written. The German and Brazilian sites read like the UK site run through a translator. They mentioned payment methods and licensing rules that do not apply in those markets.
- Fact-checking was absent. Bonus terms, wagering requirements and withdrawal times were out of date on a majority of the top-earning pages — exactly the kind of YMYL failure a core update targets.
What we did
- Months 1–2. Clean-up. Disavowed the bulk and PBN links, removed or consolidated 60% of the URLs, and set up a migration map with 301s so nothing that still ranked was lost.
- Months 2–5. Rebuild by cluster, one GEO at a time. Starting with the UK: topical clusters around specific game types, payment methods and licensing, with pillar pages that could actually be the best result for the query.
- Months 3–11. Native-language writers per market. German, Brazilian Portuguese and Canadian English writers who bet, plus an editorial process that checks every bonus and payment claim against the live operator site monthly.
- Months 4–11. Links that look like a brand growing. Digital PR around data pieces, industry publications, partnerships — at a velocity that matches the content output rather than a budget.
- Ongoing. SERP and competitor monitoring, and a monthly report tied to revenue per visit rather than to positions.
The numbers
Eleven months. The first two months show no growth at all; that is what a clean-up looks like.
- Sessions: 4.1M per month, above the pre-update peak, on roughly 40% fewer URLs.
- Revenue per visit: +92%, because the traffic that came back has commercial intent and the pages it lands on are accurate.
- Core-update losses since the rebuild: 0. Three updates have shipped since; each one has been flat or positive for the network.
- Manual actions: none. The disavow was submitted pre-emptively, before any penalty was applied.
What we’d do differently
We rebuilt the UK site first because it was the largest revenue line. In hindsight, Brazil should have gone first: the market was less competitive, the content gap was larger, and the recovery would have shown up in the report two months earlier — which would have made the “nothing improves for two months” period easier for the client to sit through. We also underestimated the editorial load of monthly fact-checking across four markets and had to add a second editor in month six.
Lost ~60% of organic traffic across two core updates. Bought links, thin content across four GEOs. We disavowed and cleaned, rebuilt topical clusters and replaced the content engine with native-language writers.