Publisher search work looks like ordinary SEO and is dominated by two things ordinary sites do not have. The first is an archive of thousands or millions of pages, most of which earn nothing and some of which are the site's most valuable assets. The second is an advertising stack that loads scripts from many third parties and is the main reason publisher pages are slow. Both constrain what any recommendation can achieve.
The archive is an asset and a liability at once
Old articles accumulate links and occasionally rank for years; most of the rest dilute crawling and add nothing. The work is triage: identify the pages that earn, update them deliberately with a visible date and substantive change, consolidate near-duplicates, and remove or noindex what has no value. Doing this well is worth more than any volume of new publishing.
The advertising stack is the performance problem
Header bidding, multiple exchanges, consent frameworks and tracking scripts make publisher pages heavy, and the revenue team owns all of it. Improving page experience therefore requires a negotiation rather than a technical fix. Measure the revenue effect of each change honestly, because a recommendation that costs more advertising revenue than it earns in traffic will rightly be refused.
Internal linking from the front page is a real lever
What the homepage and section fronts link to determines what gets crawled and how quickly new content is discovered. Publishers routinely link only to the newest items and bury evergreen pieces that earn continuously. Deliberate placement of the durable content, and links from new articles back into it, costs nothing and compounds.
News surfaces have their own requirements
Appearing in news and discovery surfaces depends on content policies, technical requirements and site reputation rather than classic ranking factors, and the rules change. Treat that traffic as volatile and valuable rather than as a foundation, and make sure the business does not build a cost base on a channel a policy update can halve overnight.
Questions people ask about SEO for publishers
Should old articles be deleted or updated?
Update what has value and a reason to be current, consolidate near-duplicates into one strong page with redirects, and remove or noindex what nobody would miss. Deletion without redirects discards accumulated links, which is the most common mistake in archive clean-ups.
How do we handle syndicated content?
Agree with partners who carries the canonical version and make the technical implementation match the agreement. Syndicating without that produces two competing versions, and the partner with more authority usually wins the ranking for your own reporting.
Is AI-generated content viable for publishers?
Scaled content produced primarily to manipulate rankings is treated as spam regardless of how it was made. The workable use is assistance under editorial control, with a human accountable for accuracy and a byline that means something.
What should a publisher measure?
Sessions and revenue by article cohort and by age, not just totals, so you can see whether the archive or the new publishing is carrying the site. Also track how much traffic depends on a single volatile surface, because that number is a risk figure.