The operator opened eighty creator profiles on Tuesday. She read each bio, checked the last three videos, and wrote the handle into a sheet if it looked right. By six in the evening the sheet had six names on it. Automated creator sourcing exists because that ratio is ordinary.
Eighty opened. Six kept. Everything below follows from that single number.
So this is for sellers running ten to thirty SKUs in the US or Southeast Asia, with one or two people on affiliate work. Monthly GMV sits between thirty and a hundred thousand dollars. There is no sourcing team, and the person doing the searching has four other jobs.
Across the automated creator sourcing runs we sit in on at DAMI, the ratio barely moves. Roughly one name kept for every thirteen profiles opened, whether the category is beauty or car parts.
Manual sourcing is not stupid. It is expensive in a way that never shows up on a report. The cost is a day of attention, spent producing a shortlist instead of a result.
What a day of searching actually buys
So break the day down, and the waste becomes visible. Two hours go into searching: typing category phrases, scrolling results, opening tabs. Four hours go into judging whether an account sells or only performs engagement. The rest goes into writing things down and re-finding profiles that were closed by accident.
Then the six names get contacted. Maybe two reply. One posts a video. That is the output of a day.
Then run the arithmetic across a month, and the real constraint appears. Twenty working days, one person, perhaps a hundred and twenty names a month if sourcing were the only job. It never is.
And that is the ceiling automated creator sourcing is measured against, not the price of a subscription. A tool that returns thirty names with a reason attached beats one that returns three hundred without. So the useful comparison is not databases, it is reasons.
The list runs dry before the SKU count does
Then there is a second problem, worse than the hours. Every store has a pool of creators who can move its products. In a large market that pool is deep. In a narrow category, or a smaller market, it is not.
After two or three rounds, the same names keep coming back. Teams notice it as a feeling, the sense of having seen a handle before. A spreadsheet would answer it, but the spreadsheet is organised by campaign, not by creator. So the same person gets contacted twice, months apart, with a different offer.
So it is a record problem, and it outlives the person who made the record. Our note on what happens to a creator list when an employee leaves covers the part that hurts most, which is the context walking out with them.
So the second requirement is not more names. It is a record that survives staff changes, because a list rebuilt from memory every quarter is a list that repeats itself. DAMI keeps that record at the creator level rather than the campaign level, so the question “did we already talk to this one” stops depending on who is asked.
Sourcing from the listing, not from a phrase
Yet most manual sourcing starts with a phrase. Beauty creator. Pet products. That is how search works, so it feels like the natural unit.
But the question a seller needs answered is different. It is not who talks about this category. It is who has sold something like this, to an audience that buys it.
And those two questions return different people. The first returns whoever uses the right words. The second returns whoever has a track record, which is a smaller and far more useful group.
Because of that, the input changes the shape of the shortlist. When the input is a listing rather than a phrase, the shortlist filters on evidence: has this creator sold in this category recently, does their audience convert, how many videos did they post for a similar product. That is the filter DAMI’s creator filtering runs on, and it is why one run returns different names for two listings in the same category.
| Keyword search | Product-based sourcing | |
|---|---|---|
| Starting point | A category phrase | A listing you already sell |
| What it filters on | Words in a bio | Category sales history |
| Typical day output | 6 to 12 usable names | A shortlist of dozens |
| Fails when | The pool is narrow | The product has no sales history yet |
The same reversal applies to competitive research. Working backwards from a competitor’s product to the creators already selling it is a sourcing method, not a data exercise. Our breakdown of competitor analysis starting from creator lists covers how that loop runs.
One caveat worth planning for. A product-based shortlist is only as good as the product’s sales history. On a listing with no orders yet, the filter has nothing to read. The honest move is to source from the closest product that does have history, and to accept that automated creator sourcing is only as smart as the listing you point it at.
What automation still cannot judge
One honest limit first. Automation cannot tell you whether a creator will be difficult. Some accounts sell well and are exhausting to work with: slow to reply, unwilling to follow a brief, prone to asking for a rate change after the video is live. DAMI will not surface that either, because no filter has ever read a message thread and felt tired.
One more limit. It cannot tell you whether the product suits the audience, only whether the audience bought something adjacent. A creator with strong kitchen sales may still be wrong for a specific kitchen gadget. That judgement stays with a person who knows the product.
And one practical limit. A run returning four hundred names has not solved the problem, it has moved it. The useful output of one run is a shortlist a single person can work through in a week, not a database to be admired.
So what does good look like? A workable shortlist has a reason next to each name: this one sold two similar items in the last sixty days, that one replied within a day on a previous campaign. Names without a reason get contacted last, or not at all.
That is the standard worth holding any run to. For the mechanics, see how product-based sourcing works in DAMI on the product page, or create your account and point it at one listing before rebuilding the whole process.
FAQ
Is automated sourcing just a bigger creator database?
No, and the difference matters. A bigger database makes searching slower, not faster. Sourcing is the filter layer that answers who has sold this kind of thing, which is narrower than who exists. DAMI’s creator resource management is where those answers accumulate, so a second run starts from what the first one learned.
How many creators should one run return?
Few enough that one person can work through them in a week. In practice that means dozens, not hundreds. If a run returns four hundred names, the filtering has not been done yet.
Does it work if my category is narrow?
Usually better than broad search does. In a narrow category the pool is small enough that a keyword search returns the same twenty accounts every time. A product-based filter finds the accounts that sold adjacent items, which is where the rest of the pool sits. That is the case where automated creator sourcing pays for itself fastest.
