Product-led SEO

The platform was built on WordPress and leaned on an off the shelf implementation for its core functionality. It worked. It also carried limits we could see as soon as we started testing it properly: reliability gaps, patchy coverage, and a system that was never designed to be developed continuously.

That mattered because driving more traffic to a product with unresolved technical problems only widens the audience for those problems. So we made an early call that shaped the whole engagement.

An average position of 7 told us the site was already reaching the first page for the queries in the report, and 5.61K impressions said Google was exposing it to real demand. The gap between impressions and clicks was the opportunity.

We did not answer that with a content sprint or a link push. We went back to the product.

Our developers moved past the original implementation and built a more capable pipeline underneath the product. The most valuable change was detecting every relevant board properly, which is the core job the SaaS exists to do.

We then stopped treating the platform as finished. Releases were documented in a changelog, feedback fed the next cycle, and the development loop settled into something repeatable: user feedback, diagnosis, development, release, observation, iteration.

By month two the numbers had barely moved: 236 clicks from 5.67K impressions, CTR flat at 4.2%, average position drifting from 7 to 7.7. Clicks had gone up by one.

That flat line was useful. It proved that an operational product and some existing rankings were not going to compound on their own. The next move had to be structural, so we stopped asking how to create more pages and started asking how to make the existing ones easier for Google to discover, understand and prioritize.

We started with metadata and page level targeting, giving every important URL a single clear purpose instead of overlapping signals. Then we reorganized the site into a hub and spoke model, so core pages, supporting pages and related content reinforced each other rather than competing.

Internal linking carried most of the weight. We used contextual links with descriptive anchors, removed orphan pages, and kept the pages that matter within a shallow click depth.

We are not claiming a single change produced every indexing gain. Several things were improving at once and SEO is cumulative. What the report shows is the site's footprint becoming more established, with Google processing a growing share of the URLs it knew about.

We judged keywords on their relationship to the product: intent, topical relevance, SERP competition and the actual value of the traffic. Volume on its own never justified a page. That kept the site from drifting into traffic it could not serve.

Related queries were grouped by the need behind them. Where several keywords meant the same thing, they became one stronger page instead of four that cannibalized each other. Where the intent genuinely differed, they earned separate pages.

Content was published gradually and deliberately. We let the product and the architecture mature instead of racing ahead of them, because a content farm attached to a fragile product helps nobody.

Link building started late on purpose. By the time we pushed on authority, the SaaS had a working and improving product, useful pages, organic traction, a stronger architecture and a clear search strategy. Pointing links at a weak foundation would have wasted them.

The mix was digital PR for brand mentions and coverage, niche edits where the surrounding content actually related to the product, and guest posts through personal outreach rather than marketplaces, which kept us in control of context and placement quality.

As traffic grew, mobile performance became the bottleneck. The product scored 59 on mobile with a 5.9 second First Contentful Paint and an 8.9 second Largest Contentful Paint. Accessibility and SEO were already strong. Loading was what held the experience back.

We treated this as product quality rather than a Lighthouse score. Someone arriving from search should not wait several seconds before the product becomes usable, and that fix serves the user and the ranking at the same time.

The first symptom was speed. A site serving more than 1,500 visitors a day started taking around ten seconds to load, and the loading metrics fell apart. In DevTools we found a request going through a /rpc_proxy endpoint that returned Base64 encoded JavaScript, which the browser then decoded and executed.

We checked the WordPress files, plugins, themes, database and configuration. Nothing. That absence was the clue, so we followed the entire request path instead of assuming the problem lived on the origin.

It was sitting at the Cloudflare edge. A Worker had been attached to routes covering the site, fetching the legitimate WordPress response, modifying it, and injecting JavaScript before anything reached the browser. Nothing in WordPress was ever compromised, which is exactly why searching WordPress found nothing.

We traced the injected script, located the Worker, removed the affected routes, revoked the unrecognized token, secured the account and retested until the injected behavior was gone.

We are not claiming Cloudflare was breached. What the evidence showed was unauthorized access to the site's Cloudflare configuration and a malicious Worker deployed through the API. How that access was obtained is not something we can establish from the evidence available to us.

With the product rebuilt, the architecture clarified, indexation healthier, content aligned to intent, authority earned and the infrastructure secured, the growth curve changed shape.

Against the early benchmark of 236 clicks and 5.67K impressions, the site reached 42.9K clicks and 783K impressions while improving CTR from 4.2% to 5.5% and average position from 7.7 to 5.1. Rising volume alongside a better position and a better click through rate is the combination worth having.

There was no single tactic behind this. Ten things improved together, and the system is what produced the result.

The biggest lesson was not a keyword, a link or a technical trick. SaaS SEO works best when the product and the search strategy improve together. The rankings followed because the whole system got better.