Content decay recovery with an untouched control group

Most pages that lose traffic year on year do not have a writing problem. Of the declining URLs I have worked through since 2019, a little over nine hundred of them across B2B SaaS, marketplaces and publishing, close to a third fell because the query changed shape rather than because the page aged: Google started rewarding a different page type, demand moved to a phrasing the page never targeted, or a product change left the page answering a question nobody asks now. Rewriting those pages spends budget and returns nothing. This engagement decides, one URL at a time, which of your declining pages earn a rewrite, which should be merged into a stronger page, which should be redirected, and which should be left to decline deliberately. Who does the work: I do. I am Sofia Almeida. I work as a fractional head of content for two or three companies at a time, in English, Portuguese and Spanish, and the decision on each URL is mine and signed. Before going fractional I ran organic content for a Lisbon travel marketplace through two replatforms and a merger, which is where I learned that the expensive mistake is not a bad rewrite. It is a good rewrite of a page that should have been retired. The threshold for entering the analysis: A URL is in scope if it recorded at least 200 impressions a month averaged over the trailing twelve months in Search Console and has lost 25 percent or more of clicks against the same period a year earlier. Below that, the data cannot separate decay from noise, and anyone who tells you otherwise is charging you to read randomness. Seasonal businesses get a year-on-year comparison, never month-on-month. Sites with fewer than twelve months of data get told to wait. The control group: Every rewrite wave holds back a matched set of declining URLs, matched on entry impressions, page type and starting position, left untouched for the full measurement window: eight weeks minimum, twelve by default. Positions move for reasons that have nothing to do with the work. A core update lands, a competitor ships a category page, seasonality turns. Without an untouched comparison you will credit me for a core update, or fire me for one. The control group is not optional in any of the three packages, and it is the reason my reporting is duller and more believable than the usual before-and-after screenshot. What you receive: One row per URL with the decision, the evidence behind it, and the order of execution, in a sheet you keep. A decision document that explains the pattern, not just the list, so your team can make the same call without me next quarter. A recorded walkthrough of forty minutes or so. In the engagements that include rewriting, the drafts themselves, in your CMS's format, with the redirect and internal-link changes each one depends on written next to it. When the answer is inconvenient: Roughly one engagement in five ends with me recommending fewer rewrites than the budget allows. Twice the recommendation has been to stop the refresh programme entirely for a quarter and spend the money on product templates, because the decaying blog was not the part of the site attached to revenue. I put that in the same document as everything else and defend it on the call. If your internal case for this project depends on a specific number of pages being refreshed, say so before we start, because I will not produce that number to protect a business case. Not included: Link building. Technical migration work. Net-new topic planning, which is a different engagement and quietly swallows a decay project if you let it. Changes made directly in your CMS unless we agree that separately. I do not touch robots directives, canonical logic or redirect configuration myself. I specify them; your engineers ship them. Who this is not for: Sites under about a thousand organic sessions a month. There is no decay to recover there, only a page that never ranked, and the analysis would find nothing worth the fee. Anyone whose traffic fell in a single week rather than over quarters: that is a technical fault or a manual action, and it needs diagnosis first. Teams that want the rewrite executed this month regardless of what the analysis says. And anyone who needs the report to name a competitor as the cause. The usual cause is that the page stopped being the best answer, and I will write that down. Working languages are English, Portuguese and Spanish, including Brazilian Portuguese SERPs, which behave differently enough from European Portuguese that treating them as one market is its own source of decay.