Sorting Spy Results by Run Length Instead of Impressions

For the first year I used ad spy tools, I sorted by impressions. It felt obvious — the biggest number at the top must be the winner. Then I noticed I kept studying ads that had one loud week and disappeared, while a quiet ad with a modest number kept showing up in my feed month after month. That’s when I flipped the sort.

Run length is the signal I trust most now. An ad that has been live for 90 days is telling me something an impression estimate can’t: somebody is still paying for it and hasn’t turned it off. That is real money making a real decision, week after week.

Impression numbers in these tools are estimates, and I treat them that way. They’re usually modeled from panel data or sampling, not pulled from the ad platform’s billing. A “200,000 impressions” label might be off by a wide margin, and it rarely tells you whether those impressions were cheap or expensive. I use them as a rough shape, not a scoreboard. Check the provider’s own documentation on how they estimate.

New ads and old ads answer different questions. A fresh ad tells me what a brand is testing right now — new angles, new hooks, seasonal pushes. An old ad tells me what survived. I keep both lists, but I read them differently. When I’m looking for a structure to adapt, I start with the long runners. When I’m looking for what’s changing in a niche, I look at what launched in the last two weeks.

Combining signals takes some care. Long run length plus a weak impression estimate can mean a small, steady budget rather than a winner. Short run length plus huge impressions can just mean a launch spend. I like ads that are both old and still getting steady volume, and I check the landing page before I trust the pairing.

Sorting by longevity changed how I work. I spend less time chasing spikes and more time studying things that lasted. I still use NativeAdSpy to browse the library, but the sort order is mine.

— Ryan Brooks