Scaled content abuse and AI content: what Google looks for

Look for the production pattern

Google’s policy concerns large numbers of pages created primarily to influence rankings, with little value for the reader. Start by grouping URLs by template, publication date, content source and target query. If five hundred town pages came from one brief, inspect examples from different batches and check what actually changes beyond the town name.

Compare the page with the job it promises to do

Suppose a page claims to compare local installers. Look for the names of installers that actually serve the town, how the selection was made, evidence for quoted prices and a way to act on the comparison. If every location repeats the same generic advice, editing adjectives or adding another paragraph leaves the missing local information unresolved.

Choose a treatment for each group

Keep the pages you can substantiate and identify where first-hand data or product knowledge would make the others useful. Consolidate genuine overlaps where one destination can answer the same need, and remove or exclude content that has no useful audience. Record the intended treatment against each group before making a bulk change, then test representative URLs afterwards.

Change the publishing decision

Give the editor a concrete acceptance test. For the installer example, require evidence of local coverage, a checked date for each material claim and an identified person responsible for updates. If a manual action is present, preserve examples of pages repaired or removed and describe how the revised approval process prevents the same batch from being generated again.