The Quattr Method
L3 · Refresh & content quality, why refreshing content beats net-new
Recoverable clicks as the ranking function; days-to-proof economics.
Your site has more pages than anyone still working there has read. Some earned real traffic once and then quietly stopped. Nobody can say which ones, or whether the demand that fed them is still out there.
L3 · Refresh & content quality is where the Quattr Method breaks with the industry's default instinct, which is to write more. It refreshes first, and it can defend that ordering with arithmetic instead of taste.
On the cross-map this work is primary on R2 · Retrieved, the second question: do you enter the consideration set, the pool a search engine or an answer engine like ChatGPT draws from at all? Its secondary effect lands on R3 · Referenced, the third question: are you the source answers are built from?
What gets refreshed first?
The pages with the most recoverable clicks. Your worklist starts with a decay diagnosis: pages that earned clicks and lost them since their own peak. Each candidate ranks by traffic it already proved it can win. Not by age. Not by whoever complained loudest.
Decline is a statistical question before it is a work question. Is the drop bigger than that page's ordinary week-to-week movement? Only pages whose decline survives that check reach the queue. Otherwise you rewrite a page that was never falling.
The content-decay-refresh skill runs the diagnosis and hands back a ranked list, one recommendation per page: refresh, consolidate, or leave alone. A skill is a packaged analysis you run by name in your assistant.
Recoverable is a modeled word and it carries that label. The page proved the demand once. What a refresh wins back is an estimate. Its job is to rank the queue rather than promise the outcome.
The workflow that does this: Category decline, verdict included →
Why refresh beats net-new
Because a decayed page is already indexed, already linked, already known to the engines that matter. Improving it moves numbers in days to weeks. A new page starts at zero on all five questions: unfetched, unlinked, untrusted, unproven.
The classic shape is the guide that owned its query two seasons ago and slid while competitors refreshed theirs. The demand never left. The page aged out of the consideration set. Bringing it back beats waiting for a new page to earn all of that from scratch.
That is the economics behind refreshing first. It is not an argument against new content. It is an argument about which proof arrives fastest, and fastest-proof-first is how the method sequences: prove in weeks before you fund what takes quarters.
Learning speed compounds too. A refresh program gets readouts back while a net-new program still waits on its first indexation. Your editorial judgment improves before the expensive writing starts.
How do you know the refresh worked?
You hold pages back on purpose. Refreshes ship in pilot batches, and you read each batch against comparable pages nobody touched. That is a holdout: change some pages, leave similar ones alone, read the difference. The season everybody lived through cancels out.
A page that rose after a refresh proves nothing alone. A batch that rose while its holdout stayed flat proves the program. That is the proof standard behind every lever here: intervention against holdout, never a trend line.
Only after the pilot lift confirms does the backlog scale. The gate keeps enthusiasm from outrunning evidence, which is how refresh programs usually die. A good first result, a hundred pages queued behind it, and no way to tell which half did anything.
You ask the numbers the same question twice. Once of the decline, before the work starts. Once of the lift, after it ships, at the same significance bar the rest of the method runs on: a change is real only when the test says it beats ordinary wobble.
Ask it yourself
What should we refresh next, ranked by recoverable clicks, and what would we leave alone?
When do you prune instead?
When the page never earned anything. The same diagnosis that surfaces refresh candidates surfaces those too, and each one gets a new angle or comes down. A library that only ever grows is a liability wearing a content strategy's name.
Quality is the other half of this work's title, and it scores relatively. Not against an abstract bar. Against the pages it has to beat. Re-optimization waits for real signals after launch instead of thrashing on day two.
Consolidation deserves its own line. Two pages splitting one query usually beat each other long before they beat anyone else, and merging them is a refresh with a demolition permit.
Some refresh targets serve demand the keyword tools barely register. Your own impressions say it exists, and on your own site your data outranks an industry average.
From here the work hands off upward. Net-new content opens with briefs the decay data helped write, and the pruning clears space the new pages get credit for. Internal linking is the other half of the wedge, the two kinds of work the method proves fastest.