The Case for Publishing Less

Here is the direct answer: publishing more is not automatically publishing better, and a site full of thin pages is harder for an AI system to characterize than a smaller site of strong ones. Every page you publish is a statement about what you are about. Weak pages are weak statements, and they average in.

This runs against the instinct of the last decade, where volume was a reasonable proxy for effort and effort was eventually rewarded. That logic has weakened considerably. A system assembling an understanding of your site is looking for a coherent signal about what you cover and how well. Two hundred pages where thirty are good and a hundred and seventy are filler produce a blurrier picture than thirty good pages alone.

What a thin page actually costs

Three things, none of them obvious from a traffic report.

It dilutes the topical signal. If most of your pages are generic, the generic reading of your business is the well-supported one. Your genuinely expert pages become outliers in your own corpus, which is a strange position to put them in.

It competes with your own best page. Four pages circling the same question mean a system has to pick one, and it may not pick the one you would have. Worse, the corroboration splits across them so none accumulates enough to win outright.

It costs retrieval attention. Crawling and retrieval budgets are finite. Pages that exist mainly to exist consume attention that could have gone to the ones that matter.

What counts as thin

Not short. Short is fine — a page that answers one question completely in three hundred words is a strong page. Thin is something else:

• Pages that restate what another page already says better.
• Pages written for a search term rather than for a question someone actually has.
• Tag and category pages that list things and say nothing.
• Old posts tied to a product, price, feature or year that no longer exists.
• Location or service pages generated from a template with a word swapped.
• Posts that exist because something had to go out that week.

That last one is worth sitting with. A publishing cadence is a useful discipline, but cadence for its own sake manufactures exactly the pages this article is about.

What to do with them

Four options, in rough order of how often they are the right call.

Merge. Several pages circling one question become one page that answers it properly, with the others redirecting there. This is usually the best outcome: nothing is lost, the signal concentrates, and the surviving page is stronger than any of its inputs.

Improve. If the question is genuinely worth answering and the page simply does it badly, fix the page. Worth doing for anything with real intent behind it.

Remove. For pages about things that no longer exist and are not worth preserving. Return a proper gone status rather than redirecting everything to the homepage, which teaches a system nothing except that your redirects are meaningless.

Keep and stop worrying. Some pages are thin and necessary — contact, privacy, terms. They are not trying to establish expertise. Nobody is confused by them. Leave them.

The honest caveat

I would not do this aggressively or all at once. Pages that look useless in an analytics tool sometimes earn links, get cited, or serve a small number of people extremely well. Pruning is a scalpel, and the failure mode — removing something that was quietly working — is harder to reverse than leaving a weak page up another month.

So: list the candidates, check each against links, citations and any traffic it does get, and merge far more often than you delete. Do it in batches you can watch rather than in one sweep you cannot evaluate.

The reframe

The question is not "what should we publish this week." It is "what is the clearest possible picture of what we are expert in, and does every page on this site contribute to it." Most sites would be better represented by two-thirds of their current pages, and the work of getting there is cheaper than the work of writing more.

My read: for a site that has been publishing steadily for a few years without pruning, a careful consolidation pass is frequently higher return than the next three months of new content — and it is finite work with a clear end, which the alternative never is.

About the author

Scott Tischler is the Founder & Chairman of AIrecommend.ai and a practitioner-authority on AI search and Answer Engine Optimization. With 20+ years in marketing technology — including American Express, MetLife, and UBS — and executive and professional study at Wharton, Harvard, and Oxford, he helps businesses become the ones AI recommends.