Building a content strategy that survives algorithm updates

Every major search engine update sorts through content strategies, and it always sorts in the same direction.

What never holds up

• Content produced solely to target a spotted algorithmic loophole
• Volume without depth, designed to cover keywords rather than needs
• Duplicate or near-duplicate content across pages

What holds up over time

Content that genuinely answers a question, written by someone who knows the subject, structured to be useful regardless of the engine that displays it. This is not an SEO strategy; it is an editorial strategy that also happens to rank well.

How to test it before publishing

One simple question is often enough: if this content were indexed by no engine at all, would it still have value for someone who came across it by chance? If the answer is no, it will not survive the next update.

What Google actually evaluates

Google does not hide its criteria; it publishes them. Its page Creating helpful, reliable, people-first content defines people-first content as content "created primarily for people, and not to manipulate search engine rankings". It offers very concrete self-assessment questions, including this one: will someone reading your content feel they have learned enough to achieve their goal?

The same document describes the E-E-A-T framework (experience, expertise, authoritativeness, trustworthiness) and specifies that, among these aspects, trust is the most important. It also invites you to ask three questions about every piece of content: who created it, how, and why. The third is the most decisive: content produced primarily to attract traffic from search engines starts with a handicap that no technical optimisation can make up for.

The helpful content system, launched in 2022, is no longer a separate filter. With the March 2024 core update, Google folded this evaluation into its core ranking systems and stated that no single signal or system is used, on its own, to judge the helpfulness of content any more. On the same day, it created the scaled content abuse policy: generating many pages with the primary purpose of manipulating rankings is penalised, whether those pages are produced through automation, by humans, or by a mix of both.

On AI, the position has been consistent since February 2023: Google rewards quality, however the content is produced, and only automation intended to manipulate rankings breaks its rules. I analyse this point in detail in AI slop: Google does not penalise AI, it penalises scale.

Content audit: keep, merge, delete

A strategy that survives updates starts with an honest inventory of what already exists. For each page, I cross-reference three readings: its Search Console data over twelve months, its overlap with other pages on the site, and the answer to the question "who, how, why".

Page situation	Signal in Search Console	Decision
Meets a specific need, up to date	Stable or rising impressions and clicks	Keep, update dated facts
Covers the same topic as another page	Two URLs alternate on the same queries	Merge into the stronger one, 301 redirect the other
Thin but on a real need	Impressions without clicks, position beyond 10	Enrich with experience and sources
Written for a keyword, with no real need	No impressions for months	Delete or merge, as a last resort

Deletion comes last, and Google says so itself on its page about core updates: removing content is a last resort, to be kept for what cannot be improved. If you are considering deleting entire sections, it is often a sign that they were written for search engines rather than for readers.

Reading the effect of a core update in Search Console

After an update, the temptation is to fix things straight away. The method Google recommends is more patient, and I apply it as is. Wait for the rollout to finish, as announced on the Search Status Dashboard, then for at least one full week. Then compare that week with a week before the rollout began, isolating the pages and queries that moved.

Scale matters. Google distinguishes a small drop, such as from position 2 to 4, which does not warrant any drastic action, from a fall from position 4 to 29, which calls for an in-depth assessment of the site as a whole. It also advises against "quick fixes" such as removing a page element because you heard it hurt rankings. The effects of an improvement can take several months to show. To run this diagnosis page by page, my free SEO diagnostic compares up to three pages side by side.

The same filter on the generative engine side

The same logic now applies to generative engines (ChatGPT, Perplexity, AI Overviews): content cited in an AI answer must bring verifiable, specific information, not a generic rewording of what already exists elsewhere. The survival criterion has therefore become doubly strict: withstanding both classic ranking updates and the selection made by generative engines, which, at heart, both apply the same filter.

A concrete example: faced with the question "what should I do after a core update?", a sentence like "you need to produce quality content" has no chance of being picked up. The sentence "Google recommends waiting at least one week after the rollout ends, then comparing with an earlier week" has every chance: it is dated, attributable and verifiable. I have detailed these mechanisms in what makes content citable by generative engines.

Algorithms change. Users' real information needs, much less so.

It is this distinction, between opportunistic content and content that holds up, that I assess in my work on GEO.

Further reading: SEO in 2026: why brand authority matters more than the keyword

Frequently asked questions

What should I do after a drop linked to a Google core update?

Wait for the rollout to finish and then at least one more week, compare with an earlier week in Search Console, and only react strongly in the event of a major fall. Google advises against quick fixes and recommends improving content rather than deleting it.

Does Google penalise content written with AI?

No, not as such. Since February 2023, Google has stated that it rewards quality regardless of how content is produced. What gets penalised is the large-scale production of pages intended to manipulate rankings.

Should I delete old articles that no longer generate traffic?

Not as a first reflex. An article on a real need can be enriched or merged into a stronger page. Deletion is reserved for content written for a keyword with no real need behind it.