Content marketing advice tends toward volume. Produce more, publish faster, cover every keyword. Generative tools have made that trivially easy, which is precisely why it stopped working.
What changed
Google's March 2024 spam policy update introduced scaled content abuse as a named violation: generating many pages primarily to manipulate search rankings rather than to help users.
The important detail is that the policy is not about how the content was made. It is about purpose and pattern. Human-written pages produced at scale for ranking purposes are inside the policy; AI-assisted pages genuinely helping readers are not.
The signal is not word count or authorship. It is a site that suddenly behaves like a content farm.
The pattern to avoid
A domain with a handful of pages and minimal search presence publishing fifty articles in a week is exhibiting exactly the shape the policy describes. So is a set of pages that differ only by a swapped city name — which Google has called doorway pages for well over a decade.
We apply the same rule to our own county pages: three of them, genuinely different, rather than thirty templated ones. If a page could be about anywhere, it will rank nowhere.
How we stage
A cadence, not a dump. Two to four substantial pieces a week, sustained. Slower than the alternative and it compounds instead of triggering a review.
Pillars first. The pieces you want to rank for go out first and get the internal links. Supporting articles follow and point at them.
Internal links written in, not generated. Every article here links to others because the argument genuinely continues there, not because a plugin inserted them.
Citations to primary sources. We keep a single file listing every citation used across the blog and what it supports, and nothing gets cited that was not read at the source. A fabricated citation is the one error a skeptical reader catches in ten seconds.
Nothing published that we would not defend on a call. Which is the real filter. If an article exists only to rank, it will read like it.
What this costs
Speed. A staged rollout of fifty pieces takes three to four months rather than a week, and the traffic arrives correspondingly later.
What it buys is that the traffic arrives at all, and that a manual reviewer looking at the site sees a practice publishing regularly rather than a domain that was inflated overnight.
For your own site
If you have had a large batch of content written, stage it. Set most of it as drafts, publish on a fixed cadence, and link each new piece into the ones already live.
It is the same content either way. The difference is whether it arrives looking like a publication or like an inflation.
We do this for ourselves, which is either reassuring or self-serving depending on your view. The articles are dated so you can check the cadence.