The clean-up job that was deleting page-one articles

When a site loses traffic, the usual suspects are a Google update, a competitor, or lost backlinks. On an affiliate website we worked on recently, part of the cause was the site itself: an automatic clean-up was deleting articles two days after they went live.

Nobody had chosen to delete good content. A piece of automation was doing it, quietly, every day.

What we saw

The first sign was in Google Search Console: pages that ranked well, then vanished from the data a few days later. When we matched those addresses against the site’s own records, the pattern was clear.

That last number matters. When a deleted page sends visitors to the home page, Google usually treats it as a dead page, not a move. The ranking the article had earned is simply lost.

Why problems like this go unnoticed

• Automation is quiet. Scheduled jobs run and report success, whatever they remove.
• Rules outlive their reasons. A setting that made sense for an older content strategy keeps running after the strategy changes.
• Nobody owns them. Editors watch what goes out; few people watch what comes down.
• The damage was spread thin. One lost article is noise. Hundreds a month is a trend that only shows up when you line up the data.

How we proved it

A suspicion is not a finding. We lined up the date each article was removed against its search impressions and clicks, page by page, and checked that the losses followed the deletions and not the other way round. Then we tried to break our own explanation: were these pages declining anyway? Was a Google update a better fit? Only when the explanation survived did it go into the weekly report, labelled proven.

It was not the only cause. A separate site update had hidden most article pages from Google, and that was the biggest single factor. Real collapses usually have more than one cause, and each one has to be dated on its own.

What we changed, one step at a time

1. Extended article lifetime from 2 days to 7. A small, reversible change we could ship in days and measure.
2. Switched the job off for good once the effect was clear.
3. Asked Google to re-check 22,951 pages it had dropped.
4. Replaced blanket forwarding with deliberate rules: 27,734 rules became 220, each sending visitors somewhere useful.
5. Archived instead of deleting. Pages with no value are taken out of Google’s view, but anything removed can be restored.

The audit we now run on every site

1. List every scheduled job (server tasks, CMS plugins, hosting tools) and what it deletes, unpublishes or redirects.
2. Give each one an owner and a reason it still exists today. No owner, no job.
3. Track the number of published pages week by week. A falling line with no editorial decision behind it is a red flag.
4. Check removed addresses against Search Console: were any of them still getting impressions?
5. Check where removed pages send visitors. A redirect to the home page is not a fix.