How to tell a product is saturating before your CPM does
Ad costs tell you a product is crowded weeks after it happened. Here is what moves first, and how to read it while there is still margin left.
Most dropshippers find out a product is saturated the same way: the cost per thousand impressions climbs, the cost per acquisition follows, and a product that worked last month stops paying for itself. By then the decision was made weeks earlier, by other people, and you were the last to know.
The useful thing to understand is that CPM is a trailing indicator. It moves after enough advertisers have crowded into the same auction. Everything that caused it moved first, and most of it is visible.
Views are not the signal. The slope is.
A product with two million cumulative views looks like a winner. It might be a corpse. Cumulative view counts only tell you something happened; they do not tell you whether it is still happening.
What matters is the rate of change. A product going from 40,000 to 300,000 views in a week is in a different state from one that went from 1.8 million to 2 million in the same week, even though the second has far more views in total. The first is accelerating. The second has plateaued, and a plateau in organic attention usually means the creators who were posting about it have moved on.
When you look at a product, ask what the last seven days did, not what the last six months totalled.
Creator count moves before advertiser count
There is a reliable sequence to how a product spreads.
First a handful of creators post it organically, usually because a supplier or an affiliate programme seeded it. Then more creators pile in, because they can see it performing for the first group. Only then do media buyers notice, build a page, and start running paid traffic. The auction gets expensive last.
That ordering is your window. The gap between "many creators are posting this" and "many advertisers are bidding on this" is typically weeks, not months, and it is the entire opportunity. If you wait for proof in the form of competitors' ads, you have waited past the margin.
Ad count is the confirmation, not the discovery
This does not mean advertiser data is useless. It means you use it for a different job.
Live ad counts answer a specific question: how many people are already fighting for this audience. That is a cost question, not a demand question. Pair it with the velocity picture and each product falls into one of four states:
| State | Demand | Advertisers | What to do |
|---|---|---|---|
| Untapped | Rising | Almost none | Enter. Margin intact, creative space empty. |
| Scaling | Proven | Arriving | Enter only with a better page or a sharper angle. |
| Saturating | Flat | Climbing fast | Your CPM is about to rise. Needs a real differentiator. |
| Saturated | Flat or falling | Everywhere | Skip it, however good the view count looks. |
The mistake is treating the second and third states as the same thing. They look similar in a dashboard, and they are opposites in a P&L. One has demand growing into the competition; the other has competition growing into fixed demand.
What to check before you commit money
A short list that costs you ten minutes per product and saves you an ad budget.
Seven-day direction. Is attention accelerating, flat, or falling? Flat is a no, however big the total.
Creator spread. Is it one creator repeatedly posting it, or many creators independently? One creator is a seeded campaign and can vanish the week their deal ends. Many creators is a real wave.
Live advertiser count. Not whether ads exist, but how many, and whether that number climbed this week.
Supplier depth. If only two suppliers list it, lead times will slip the moment demand spikes, and the product dies of stockouts rather than of saturation.
Your own fulfilment maths. A product you cannot deliver inside your market's expectations is not a winner at any velocity.
The trap to watch for. A quiet supplier side and a loud auction can happen at once. Few resellers listing a product does not mean few advertisers bidding on it: it can simply mean a brand owns the product and is spending on it directly. Check both sides before reading quiet as early.
Why most tools get this wrong
Most product research tools are databases. They scrape once, rank by total volume, and show you the same list they showed everyone else who paid that month. A list sorted by cumulative views is a list of products that already peaked, sold to a thousand people simultaneously.
What you actually want is a measure of change, refreshed often enough to be current, cross-referenced against how crowded the auction already is. That is the difference between a catalogue and a signal.
HermesSync scores products on velocity rather than raw volume, checks live advertiser counts, and labels each one Untapped, Scaling, Saturating or Saturated. It then builds the Shopify product page for whatever you decide to run. There is a free plan with no card and no expiry.