News Publisher Case Study: Traffic to Zero at the March 2024 Core Update

The situation

The client is a Canadian news site running on high-volume automated publishing. Over the 16 months I reviewed it had earned 83,200 clicks from 5.45 million impressions, a click-through rate of 1.5 percent at an average position of 26.8.

Those averages hide what actually happened, which is why averages are a poor way to read a drop. The daily chart tells the real story: a strong run through the second half of 2023, a peak in early September 2023, a gradual decline through the autumn and winter, and then a cliff in mid-March 2024 after which the line sits flat against the axis.

Not a dip. Not a demotion of a few positions. The site effectively stopped receiving search traffic.

Matching the date

The break falls in the week of 13 March 2024, which places it inside the rollout of the March 2024 core update. That update ran from 5 March to 19 April 2024, the longest core rollout on record at 45 days, and it was not an ordinary one.

Two things shipped alongside it that matter here. The helpful content system was folded into the core ranking systems, ending standalone helpful content updates. And Google published three new spam policies, the first of which was scaled content abuse: producing many pages primarily to rank rather than to help people, regardless of whether they were written by people, by AI, or by both.

A site publishing copied articles automatically at volume is close to the definition in that policy. The date match and the policy match pointed the same way, which is the strongest kind of diagnosis available, because either signal alone can mislead.

What the content audit found

• Plagiarism checks returned complete matches on every post sampled. Why: Not paraphrased, not aggregated with commentary added, and not syndicated under an agreement. The articles existed elsewhere first and appeared here unchanged.
• Publishing was automated end to end. Why: No editorial step existed between the source and the live page, so there was no point at which a person decided whether an article was worth publishing.
• Titles were reproduced along with the body text. Why: This matters more than it sounds. A site that at least rewrites headlines is making an editorial decision. Copying them too signals that no decision was made anywhere.
• No original reporting, sourcing or author accountability anywhere on the site. Why: For a news publisher this is the whole product. Without it there is nothing for a reader or a quality rater to trust.

This is a straightforward violation of Google’s quality guidelines rather than a subtle quality gap, and it is worth being precise about why. A news aggregator is not inherently a problem. Aggregation with original selection, context and attribution is a legitimate publishing model. What Google acted against here was reproduction at volume with nothing added.

The honest prognosis

The client asked directly whether the site could recover. The answer I gave was yes, followed immediately by the qualification that this would be difficult, and that difficult meant something specific.

Most recovery work fixes a site that had a problem. This site did not have a problem so much as it was the problem. There was no thin section to prune while the rest carried on, no template to rebuild, no technical fault to clear. The archive was the violation, and the publishing process that created it was still running.

That distinction is the one I use to decide whether to take recovery work at all. When the business model is what Google demoted, recovery is not really recovery. It is becoming a different publication, and a client who wants their traffic back without changing what they publish is asking for something nobody can deliver.

The roadmap, and why it was demanding

• Stop the auto-posting immediately. Why: Everything else is pointless while the violation is still being added to daily. This was the one item that had to happen before any work started.
• Edit every retained post by hand, including the titles. Why: Not a rewrite tool, not a spinner. A person deciding what this article should say for this audience, which is the editorial step the site never had.
• Remove what cannot be made original. Why: On a site of this composition, a large share of the archive has no version worth keeping. Consolidation does not help when the source material is somebody else’s.
• Write genuinely new material going forward. Why: The site needs to demonstrate a different pattern over time, not just a cleaner archive. Reassessment looks at what the site is now.
• Build E-E-A-T from nothing. Why: Named journalists, visible accountability, real ownership details, sourced claims. A news site with no identifiable people behind it fails the trust standard before any other factor is considered. The E-E-A-T rubric covers what raters look for.
• Earn editorial links. Why: Authority for a publisher comes from other publishers referencing your work, which requires having work worth referencing.

Read that list as a whole and it describes founding a news outlet, not repairing one. That was the point of presenting it that way.

What this case teaches

A drop to near-zero is a different category from a drop of 40 percent. Partial losses usually mean a quality reassessment. A flat line usually means a policy violation, and the diagnostic path is different: check the spam policies before checking content depth. The spam and penalty recovery path applies rather than the core update one.

March 2024 is still the most consequential update for sites like this. The scaled content abuse policy introduced then remains the enforcement standard, and every spam update since has drawn on it. Sites still down from that period are usually down because the original cause was never removed, so each subsequent update reassessed the same site and reached the same verdict.

Saying the work is hard is part of the job. I could have quoted a standard recovery engagement and started work. The client would have paid for months of editing and then asked why the traffic had not returned, because the answer was always going to depend on whether they were willing to become a different publication. Telling someone that early costs a fee and saves them a year.

Figures are taken directly from the client’s Google Search Console during the audit. The client is not named and identifying details are withheld under the confidentiality terms set out in the editorial policy. Diagnostic findings and the prognosis are reproduced from the audit report delivered at the time. No recovery outcome is claimed here, because the roadmap required changes to the publishing model that were the client’s decision to make.