Volume is not value: what the March 2024 update measures
The structure and markup of this piece were refreshed for current answer engines; the original analysis is preserved as written.
Generative tools made publishing a thousand pages almost free — and the temptation to flood a keyword space has never been higher. This month Google answered, naming mass production itself as abuse. The line it drew is not about who writes the content. It is about whether the content deserves to exist.
the short answer
On 5 March 2024 Google made "scaled content abuse" a spam policy: mass-producing pages to manipulate rankings, whether written by AI, humans, or both. The core update aims to cut low-quality content by about 40 percent. The lesson for AI visibility is one of measurement — count whether a page answers better than the alternatives, not how many pages you shipped. The content that survives is the content AI cites.
key takeaways
- On 5 March 2024, Google launched a core update and a spam update together, with three new policies; the most discussed is "scaled content abuse."
- Scaled content abuse targets mass-producing pages to manipulate rankings — whether written by AI, humans, or a mix. What counts is the lack of value, not the method.
- The core update aims to cut low-quality, unoriginal content by around 40 percent, per Google: generic volume stops paying.
- For the AI era, where producing at scale is nearly free, the lesson is one of measurement: do not count pages or words (vanity), measure whether each piece answers better than the alternatives.
- The content that survives a quality-focused core update — useful, original, better than the alternatives — is the same content AI chooses to cite. Quality is the shared filter.
what the quality filter keeps
The same filter sits in both rows, and it does not care how a page was written. Mass-produced material has nothing worth keeping, so it is dropped; genuinely useful material is surfaced and, increasingly, cited. The variable that decides the outcome is value, not volume and not method.
Why this lands harder in the AI era
A crackdown on mass-produced content would have mattered in any year, but it lands with particular force now, because the cost of producing plausible content has collapsed. When generating a thousand on-topic-looking pages takes an afternoon, every incentive that used to be held in check by the sheer effort of writing is suddenly unleashed, and the obvious move — for anyone optimising for output — is to flood. Google’s policy is a direct answer to that incentive: it says, in as many words, that producing at scale to manipulate ranking is abuse no matter how it is produced. Read as a strategy signal rather than a threat, it is clarifying. The cheap path just got cut off, which means the effort you might have spent spinning up volume is better spent making fewer things genuinely excellent — and that reallocation is not a concession to the algorithm, it is the thing that actually wins readers and citations. Seen this way, the policy did you a favour: it removed the option that was always going to lose eventually, and forced the question of value to the front while you still have time to answer it well.
It also exposes a trap that the AI era makes easy to fall into: confusing the ease of production with the value of the product. Because a tool can write a competent-looking page in seconds, it is tempting to treat that page as an asset the moment it exists — to feel that publishing it was progress. But an asset that no reader needs and no system will surface is not an asset; it is clutter that can now actively count against you. The honest posture is to be slower and more selective precisely because production is fast: to let the cheapness of making content raise, not lower, your bar for publishing it. The teams that thrive through updates like this one are not the ones that produced the most; they are the ones that, for each thing they put out, could answer the only question that matters — does this deserve to exist, and is it the best answer a reader could find?
The lesson, in three parts
What "scaled content abuse" actually means, why volume stopped paying, and the measure that matters instead. Open each layer for the part that changes how you judge your own content.
01 What "scaled content abuse" actually means
The headline policy from the March 2024 update is worth reading precisely, because its wording is the lesson. Google defined scaled content abuse as generating many pages primarily to manipulate search rankings, with little or no value to users — and, crucially, it applies regardless of how the content is made: automation, human effort, or a combination of the two. That is a deliberate shift. The older framing singled out "automatically generated" content, which let people argue about whether a given page was technically machine-written. The new framing drops the method entirely and asks two questions instead: was this produced at scale to game ranking, and does it actually help a reader? If the answers are "yes" and "no," it is abuse, whoever or whatever typed it. The reason this matters now is that generative tools have made producing plausible-looking pages almost free, so the temptation to flood a keyword space has never been higher — and Google has just said, in policy language, that the flood is the problem, not the faucet. It is a clean line, and a fair one: it does not punish you for the tools you reach for, only for shipping pages that no reader needed in order to occupy ground you had not earned.
02 Why volume stopped paying
For a long stretch of the SEO era, volume was a viable, if cynical, strategy: produce enough pages targeting enough terms and some of them would rank, and the cost of producing each one was high enough to act as a natural brake. Generative tools removed the brake. When a thousand pages cost what one used to, the old volume play scales without limit, and search results fill with material that exists to occupy space rather than to help. The March 2024 update is the response to that: a core ranking refinement aimed at cutting low-quality, unoriginal content substantially, paired with a spam policy that names mass production as abuse. The effect is to break the economics of the volume play. If generic pages no longer rank — and increasingly are not cited by AI either — then producing them at scale is not a cheap growth tactic; it is wasted effort at best and a liability at worst. The lever that volume used to pull has been disconnected, and the only thing still wired to results is whether the content is genuinely the best answer to a real question. That is a harder game than volume ever was, but it is also a more honest one, and it rewards the kind of work that does not decay the moment a competitor spins up a thousand pages of their own.
03 The measure that matters
This is, underneath, a measurement problem, and getting the metric right is most of the battle. The seductive numbers are output metrics: pages published, words shipped, posts per week. They are easy to count, they feel like momentum, and they are exactly what mass production optimises — which is the tell that they measure activity, not value. The numbers that actually matter are outcomes: for the real questions in your category, is your answer the best available, and are you found and cited when someone asks? Those are harder to measure, which is why they are trustworthy — you cannot inflate them by producing more, only by being better. The discipline this update rewards is to stop celebrating how much you made and start scrutinising whether what you made deserves to exist, question by question, against the alternatives a reader or a model would compare you with. That is slower and less flattering than a publishing dashboard, but it is the only measurement that survives a world where output is nearly free. It is the discipline the AC Group has held to for 27 years: count what helps, not what fills. The publishing dashboard will always look healthier than the honest map of where you are genuinely the best answer, and that gap between the flattering number and the true one is exactly where most content strategies quietly fail.
How to audit your own library against this
The policy is abstract until you turn it on your own pages, so here is the concrete version. Go through your content, page by page, and ask one question of each: if a reader stopped and asked "why does this exist?", could you answer with a real reason — "it is the clearest explanation of X," "it answers a question people actually have and answers it better than what is out there" — or would the honest answer be "to rank for a term"? That single question sorts a library faster than any tool. The pages with a real reason are assets; keep them, and keep improving the ones that are close. The pages whose only reason is ranking are the liability the March update describes, and they do not improve by being left alone — they either get rewritten into something a reader needs, or they get removed, because Google has said plainly that pruning unhelpful pages can lift a site rather than shrink it.
Then change what you report. Instead of a publishing count — pieces shipped, words written — build a map of the real questions in your category and mark, honestly, where your answer is the best one available, where it is merely adequate, and where you are absent or beaten. That map is uncomfortable because it does not flatter activity, but it is the only report that tells you what to do next: close the gaps where the best answer is someone else's, deepen the ones where you are merely adequate, and stop manufacturing pages for questions no one is asking. Run it as a habit, not a one-off, because both the questions and the competition move. The work is slower than a content calendar that rewards throughput, and that slowness is the point — in a world where producing volume is nearly free, the only edge left is the judgement to make fewer things that genuinely deserve to exist.
There is a quieter benefit to running your library this way, and it compounds. Every page you keep because it is genuinely the best answer becomes a piece of evidence about what your site is for — a consistent signal that this is a place where questions get answered well, not a place where pages accumulate. Systems that rank and systems that cite both read that consistency over time; a site that is uniformly useful is easier to trust than one where a few strong pages are buried under filler, because the filler dilutes the signal the strong pages send. So pruning is not only defensive. Removing the pages that exist for no reader sharpens the identity of the ones that do, and a sharper identity is exactly what earns the benefit of the doubt when a model is deciding whose answer to quote. The question "does this deserve to exist?" pays twice: once in the page you fix or cut, and again in the credibility every remaining page inherits from a bar held high.
Volume, value, and citations: quick answers
Is Google banning AI-written content?
No, and the distinction is the whole point of the policy. Google has been explicit that it is not against AI or automation as such; what it acts on is content mass-produced to manipulate rankings, regardless of whether a machine, a person, or a combination wrote it. The new "scaled content abuse" framing deliberately removes the method from the question and puts the purpose and the value in its place: many pages generated primarily to game search, with little or no use to a reader, are the target — and that net catches low-value human content just as readily as low-value machine content. So a single, genuinely useful page written with AI assistance is not what this is about; ten thousand thin pages spun up to blanket a keyword space are, no matter who or what produced them. The takeaway is liberating if you read it correctly: you are free to use the tools, and you are accountable for the value, which is exactly the right place to put the line.
I publish a lot of content. Am I at risk?
Volume by itself is not the problem; volume in place of value is. Plenty of sites publish frequently and are entirely safe because each piece earns its place — it answers a real question, adds something the alternatives do not, and is made for a reader rather than for an algorithm. The risk lives in the other pattern: pages produced primarily to occupy search real estate, where the page count is the strategy and usefulness is an afterthought. A useful test is to ask, page by page, whether you could defend its existence to a reader who asked "why does this exist?" If the honest answer is "to rank for a term" rather than "to genuinely help with X," that page is the kind the policy targets, however it was written. So the question to ask is not "do I publish a lot?" but "does each thing I publish deserve to exist?" — and if some of it does not, the prudent move is to improve or remove it rather than wait to find out the hard way.
How does this connect to getting cited by AI?
Directly, because the filter is shared. A generative engine assembling an answer is, like a ranking system, trying to surface what is genuinely useful and credible on a topic, and it has the same reason to pass over thin, mass-produced material — there is nothing in it worth quoting. So the content that survives a quality-focused core update and the content an AI chooses to cite are, in practice, the same content: clear, original, better on its question than the alternatives. That is good news, because it means you are not maintaining two separate strategies. The work that makes you resilient to a spam crackdown is the work that makes you quotable, and the work that makes you quotable is the work that serves a human reader well. Mass production fails all three at once; genuine usefulness passes all three. Treating AI visibility and content quality as the same problem, rather than competing priorities, is the efficient and the honest reading.
What should I measure instead of volume?
Measure outcomes, not output — the distinction that this whole episode rewards. Output is what you produced: pages shipped, words written, posts per week. It feels like progress and it is easy to count, which is exactly why it becomes a vanity metric that mass production games effortlessly. Outcome is what your content actually achieved: whether a real question is answered better by you than by anyone else, whether you are found and cited when someone asks, whether a reader leaves helped. Those are harder to count, which is precisely why they are worth counting — they cannot be faked by producing more. A practical reframe is to stop reporting "we published N pieces this quarter" and start reporting "for the questions that matter in our category, here is where our answer is the best one and here is where it is not." That second report tells you what to fix; the first only tells you that you were busy. In a world where producing volume is nearly free, the only metrics that mean anything are the ones volume cannot buy.
A note on sources and timing
This is written in March 2024, days into the rollout. We have described what Google announced on 5 March: a core update alongside a spam update, with three new policies — scaled content abuse, site reputation abuse, and expired domain abuse — and Google’s own stated aim of cutting low-quality, unoriginal content by around 40 percent. We have leaned on Google’s wording that producing content at scale to manipulate ranking is abuse regardless of method, because that wording is the substance. We have deliberately not reported which specific sites rose or fell, or quantified the eventual impact, because a core update takes weeks to settle and those results were not yet in as of this writing — citing them now would be guesswork. The durable point does not need them: in an era where producing volume is nearly free, the only content that earns rankings or citations is content that genuinely helps, and measuring value rather than output is the discipline that follows. It is the one the AC Group has kept for 27 years, across every version of this same temptation.