How we count a creator campaign

Every agency reports a big number. Almost nobody says what it is made of. Here is exactly what we count, what we refuse to count, and the one habit that makes our totals smaller than they could be — and survivable in the second meeting.

A creator campaign produces a mess of evidence: screenshots, story frames that expire in a day, handles that were promised and never posted, a highlight reel that keeps growing for a fortnight. Turning that into a number is the moment where most reporting quietly goes wrong — not through dishonesty, but because there is no agreed unit and every choice flatters.

So we fixed four units and stuck to them across thirty-five cases. That is the only reason those cases can be compared to each other at all.

The four numbers

Creators activated

People who actually posted. Not people who were briefed, not people who confirmed, not people who attended. If nothing went live, they are not in this number — they are in the tier list further down.

Verified handles

Counted separately, never folded into the headline. It matters that Farzi Cafe’s twenty creators included thirteen verified handles, and it matters differently that The Weekend’s twenty included seven. Same headcount, different campaigns. A single blended number hides that; two numbers side by side show it.

Story frames captured

The unit of work that actually happened. Seventy frames at Sky Lantern, sixty-four at Tablespoon, eighteen at Flaunt. Frames are countable from the highlight, they don’t decay, and they are the closest honest proxy for volume of coverage.

Combined follower reach

The sum of the follower counts of the handles that posted, at the time they posted. Note what that is not: it is not impressions, it is not views, and it is not audience. It is the size of the rooms the message was carried into. We say “combined follower reach” in full on every case for exactly that reason, and we do not deduplicate overlapping audiences because we cannot see them — which is precisely why one creator per audience class matters more than headcount.

Reach is the size of the room, not the number of people listening.

The three tiers

The habit that costs us the most on paper. Every roster is reported in three tiers, counted apart:

• Highlighted — posted, captured, and live in the brand’s own highlight.
• Confirmed — posted, evidenced, but not carried into the highlight.
• Pending — committed and scheduled, nothing live yet.

Andrea’s Eatery is the reference case for this. Adding the three together makes a headline that is roughly a third larger and cannot be defended once a client goes and counts the highlight themselves. Keeping them apart makes a smaller number that stays true a month later, and turns the report into a working document instead of a victory lap.

What we don’t publish

Reach figures we cannot attribute to a named handle. Numbers from case PDFs older than the current archive, where the underlying capture is gone. Press citations without a scan in the folder. Anything from the 2022 brief that the source-of-truth document flags as outdated — there are three such figures and they are on none of these pages.

And where there is no case, there is no number. The performance room on the marketing page is the visible cost of that rule: a whole room of live client work with nothing published under it, because the write-up isn’t written. We would rather carry the empty room than fill it.

Why any of this is public

Because the method is the differentiator, not the numbers. Anyone can run twenty creators. What a client is actually buying is somebody who will hand them a total they can repeat to their board without a footnote.