Most content teams have a few hundred posts and no idea which earn anything. This audit takes an afternoon and produces a decision for every URL.
Pull three numbers per post
- Organic sessions over the last twelve months, from Search Console or GA4
- Conversions attributed to the post, using whatever definition you already trust
- Last meaningful update date, not the date a plugin touched the file
Then add one judgement column: could a language model have produced this from public sources? Yes or no. Be honest. Most teams find eighty percent or more are yes.
The four actions
Keep and leave alone: real traffic, real conversions, model could not have written it. Do nothing. These are working.
Update: real traffic but stale, or a good topic executed thinly. Worth investment, especially if you can add something a model cannot copy, such as your own numbers or a tool.
Consolidate: several posts on near-identical topics splitting the same intent. Merge into one, redirect the rest. Common on any site older than two years.
Kill: no traffic, no conversions, model could have written it. Delete and return a 410, or redirect if a successor exists.
The number people argue about
On a typical Indian B2B blog running three years, the split lands near ten percent keep, twenty update, twenty consolidate, fifty kill. The last figure causes the argument, because someone commissioned every one of those posts.
The argument to have is not whether they were a mistake. It is whether keeping them costs anything. It does: crawl budget, internal link dilution, and time maintaining pages nobody reads.
Make it quarterly
Done once it is an event. Done every quarter it stops the archive rotting, and takes about an hour once the spreadsheet exists.
General information, not legal advice. Rules current as of July 2026.