Google March 2023 Core Update: The Aggregator Reckoning

The update that punished repackaging

The March 2023 core update was the moment Google visibly stopped rewarding sites that merely repackage information living somewhere else, and concentrated visibility into the places that originate it.

Look at the two clearest movers and the thesis writes itself. Wikipedia, an authoritative destination, gained about 605 visibility-index points (+10.16%) in the US Sistrix dataset. Wiktionary, its dictionary sibling that largely restates definitions found across the web, lost about 141 points (-22.66%). Same content family, opposite outcomes. The differentiator was not domain authority or word count. It was whether the site was the source or the summary.

This was the first broad core update of 2023, arriving roughly six months after the September 2022 core update, and analysts widely judged it bigger and more volatile than that predecessor. It ran a long-but-not-extreme 13 days. What made it worth remembering was not its size. It was the unusually legible pattern underneath the volatility.

A tweet, a dashboard entry, and nothing else

Google's communication around this update was minimal even by core-update standards. There was no dedicated Search Central blog post and no special guidance for this release. Just a single tweet on March 15, 2023, and the later dashboard entry logging the rollout window.

Today we released the March 2023 core update. We'll update our ranking release history page when the rollout is complete.Google - Search Status Dashboard

The Search Status Dashboard later logged the rollout as lasting 13 days and 7 hours, starting March 15 and completing March 28, 2023. Beyond that, Google offered only its standard boilerplate: there is no fix for a core update, keep making great content for people. No distinctive advice was attached to this specific release.

That silence matters for how we read the update. The primary-source pattern below is analyst observation built on visibility data, not a documented Google instruction. We can describe what moved and infer why. We cannot quote Google explaining it, because Google never did.

Who rose, who fell, and one anomaly

The data here is solid. The figures below are from the US dataset analyzed by Lily Ray at Amsive via Sistrix visibility index. A separate Sistrix UK dataset shows the same directional pattern with different magnitudes, so the dataset matters when quoting numbers.

One caveat worth keeping honest. The widely repeated claim that AI-generated content sites declined in this update was an observation Ray surfaced via a third-party tweet, not a Google statement. Treat it as analyst inference, not confirmed fact. The e-commerce category also looked positive overall only because Amazon and eBay's gains masked broad small-merchant losses underneath.

Be the source, not the summary

The non-obvious, still-useful lesson is not a 2023 tactic. It is a directional verdict that has only hardened. March 2023 was the early tremor of a doctrine you can now state in five words: be the source, not the summary.

Consider what separated the winners from the losers. Wikipedia, Amazon, eBay, government health sites, and hotels all hold something that cannot be paraphrased from elsewhere on the web. Wikipedia is the canonical reference others copy. Amazon and eBay hold the listings, prices, and reviews. A hotel is the actual primary record of itself. The losers, by contrast, were sites whose entire value proposition was restating information that originated somewhere else.

• Originated data wins: if you generated the facts, the prices, the reviews, or the records, you are the primary source.
• First-hand experience wins: pethelpful.com and therighthairstyles.com gained on demonstrable hands-on content, not on summarizing it.
• Aggregation loses: dictionary clones and thin affiliate roundups added no information that was not already indexed elsewhere.
• Authority alone is not a shield: even Mayo Clinic slipped, so a strong domain does not exempt individual pages from the same test.

The timing reinforces the read. This update landed just months after Google added the extra E for Experience to E-A-T in December 2022, and amid the first wave of mass generative-AI content publishing. The market was about to fill with paraphrase at scale, and Google was already calibrating to reward what could not be paraphrased.

The pattern every later update amplified

March 2023 reads less like a standalone event and more like the origin point of a trajectory. If your only moat is aggregating data others originate, every subsequent core update has compounded the March 2023 verdict.

1. September 2023 helpful content. The helpful content system integration deepened the preference for content made for people from genuine knowledge, the same instinct that punished reference clones here. See our breakdown of the September 2022 core update for the predecessor it built on.
2. March 2024 core update. The anti-mass-production purge took direct aim at scaled, low-originality content. The March 2024 core update was where the paraphrase-at-scale problem March 2023 hinted at met an explicit enforcement mechanism.
3. AI Overviews and AI Mode. The way Google's AI surfaces choose what to cite favors original sources over summaries, the same logic at a new layer. Our look at Google AI Mode and the post-blue-link era traces where this goes next.

Read in sequence, these are not separate philosophies. They are the same judgment applied with steadily sharper tools. March 2023 was the first time you could see it clearly in the visibility data.

The takeaway for 2026

At three-plus years old, no one is diagnosing a March 2023 hit in isolation. Sites have cycled through a dozen later core and helpful-content updates since. So we do not treat it as a live recovery playbook. We treat it as the cleanest early evidence of a rule that still governs visibility in 2026.

Across the accounts we audit, the sites that have weathered the full run of core updates since 2023 share one trait. They own something. Proprietary data, real product, documented first-hand experience, original research. The sites that have struggled across the same window tend to share the opposite trait: their pages restate what is already available elsewhere, with a thin layer of formatting on top.

That is the question to carry forward from March 2023, not as a recovery step but as a standing test for any page that wants to keep ranking: are we the source here, or are we the summary?