When every ad disappears at once.

An empty result is unusually good at suggesting a story.

The advertiser stopped spending. The campaign failed. The offer was pulled. A strategy reached its end.

Perhaps. First, check whether you still have a working view of the thing you are describing.

This is a research procedure for an illustrative situation: a collection that previously contained ads now appears empty. It is not a finding about a particular advertiser. The distinction matters because the same empty screen can sit at the end of several very different chains of events.

Preserve the disappearance

Before refreshing everything into a new state, record the result you received. Keep the source, the time, the advertiser identifier, the filters, and whether the request completed successfully. Save the last usable observation separately.

A successful request only establishes that the request returned. It does not, by itself, establish that the result covers everything you intended to inspect. A failed request establishes even less about the ads.

That is why a collection system needs to distinguish an empty result from an unavailable result. If both become the same blank list, the next person will have to reconstruct the difference after a conclusion has already escaped.

Check the view before explaining the change

Work from the cheapest explanation to the most demanding one.

Confirm that you are looking at the same advertiser and the same source. Compare location, status, date, and other relevant filters. Check whether the collection finished and whether pagination or another coverage limit could have cut it short. Look for a source change or an access failure.

Then inspect another comparable observation through the same path. If several unrelated collections become empty together, a shared observation problem deserves investigation. If only one changes, that narrows the question; it still does not answer it.

Where a public source is available, revisit it through its ordinary interface. Respect its access limits. Keep the new result and the original result, including any disagreement between them.

Put dates around what you know

“Last seen” means the last time your records contain an observation. It does not necessarily mean the last time an ad ran.

Likewise, the first successful check that no longer contains an ad is an observation boundary. Without further evidence, you cannot place the change at a precise moment inside the gap between checks. And if those checks were incomplete, even that interval needs qualification.

Write the actual dates in the research note. Describe what the source reported at each point. Keep the interval visible instead of converting it into a neat duration that the evidence cannot support.

This becomes especially important when comparing lifespans. A creative still visible at the end of your observation window has an unfinished record. Dropping it from a table of completed runs changes which ads the table describes.

Publish the smallest defensible update

If the source confirms a status change, report that status, the source, and the observation date. If a repeat check restores the missing records, correct the collection and preserve the incident. If uncertainty remains, say which observation is missing and what would resolve it.

None of those outcomes establishes why an advertiser made a decision or what it earned. That would require additional evidence.

An absence can be useful information. It can prompt a follow-up, expose a collection fault, or define the next research question. Give it that job before giving it a narrative.