Content Network Takeover — What a 55-site network looked like on handover
The 55-site content network I took over, and the audit that found why its numbers were sliding — a SERP pipeline dead for months, thousands of articles written blind, duplicate slugs slipping under the gate. A diagnosis, not a turnaround.
Responsible for the network since August, I inherited a downward line — impressions −39% in a week, sessions −22% in another — and a brief to explain it, not to spin it. The audit found the cause. The SERP research feeding the whole engine had been dark since April: 109 nights in which 5,000+ articles were written without the source the prompt itself called the most important. 172 near-duplicate slug pairs hid just under the gate built to stop them; 456 more ran between sites, outside its view. Lead routing was recoverable to 113 of 132 by reading a field already present. That is the work — a clear map of an inherited system, drawn in one pass.
Own operator engagement · audit findings · anchored in docs/claims.md
A diagnosis, not a result I produced. The −39% and −22% are the state of the network at handover — not an outcome I caused, and not one I claim to have reversed. Presented as my own operator work; the network is not named here.
The story.
In August I became responsible for a network of 55 Dutch niche sites — a nightly engine writing one article per site, per night. The line since was down: impressions off −39% in one week, sessions off −22% in another. The job was to explain that, not to claim I'd reversed it. So the audit did the talking. The SERP research the whole engine leaned on had been failing silently since April: 109 nights in which more than 5,000 articles were written without the one source the prompt itself labelled the most important. Inside sites, 172 near-duplicate slug pairs sat just under the threshold of the gate built to catch them; another 456 pairs ran between sites, where that gate never looked. Lead routing could be repaired to 113 of 132 by reading a field that was already there. None of that is a result I produced — it is the honest shape of what a fractional operator inherits, and how quickly a clear-eyed audit maps it.
The five phases, applied.
Every build under Juan Diaz, LLC runs the same playbook. Here's how it ran for Content Network Takeover.
01
Survey
Read the system before touching it: pulled Search Console, the nightly logs and the article archive, and asked the one question a handover turns on — is this line going up or down, and does anyone know why?
02
Blueprint
Traced the decline to its source instead of its symptoms. Mapped every gate the pipeline claimed to run and checked which ones actually fired — the gap between the two is where the damage hides.
03
Build
Wrote the recovery, not the spin: the SERP feed reconnected, the duplicate slugs merged, the lead field read instead of ignored. Each fix costed and ordered by what it returns.
04
Commission
Pressure-tested every finding against the raw data before writing it down — a diagnosis that can't survive its own numbers is worse than none at all.
05
Operate
Handed back a map, not a promise: what the network needs to run, what it silently skips, and the order to repair it in. The audit is the deliverable.
Built on.
SQLSearch ConsoleScreaming FrogPythonPostgres
◉ Want something like this for your operation?
Same five phases. Different sector. Let's draw the blueprint.