You paid a creator, the work went live, and now someone wants to know whether it worked.
The answer depends almost entirely on a structural question that got settled before the shoot: whose channel did it run on? That single fact determines what evidence is available to you, and no amount of reporting effort afterwards can change it.
The four situations
1 · It ran on your channel
The best case. You have first-party analytics — views, retention, traffic sources, demographics, everything the platform gives an owner. Measured, tier 2 or better.
You can compare it against your own back catalogue, which is the only benchmark that means anything, and you can see the retention curve, which is the closest thing to attribution any dashboard offers for free.
2 · It ran on the creator's channel
You can see public numbers — views, and public engagement counts. You cannot see retention, traffic sources, or demographics unless the creator shares them.
This is the common case for influencer-style work and it's a real limitation. Public view counts tell you reach happened; they tell you very little about whether it landed.
Ask for a screenshot of the analytics as part of the deliverable, agreed in advance. Most creators will provide it. Note what that is, though: it's a reported number — you're trusting a screenshot, which is fine as long as you label it honestly rather than filing it alongside your measured data.
3 · Both
Increasingly common and genuinely useful, because you get the comparison. The same content on two channels, with the creator's audience against yours.
The disagreement is the interesting part. Strong on their channel and weak on yours usually means the creator's audience relationship was doing the work — which is a finding about what you actually bought.
4 · It ran as paid
Then you're in ad-platform reporting, which is a different measurement regime with its own numbers and its own caveats about honest reading.
What to measure, in order
Did it reach anyone? Views, on the same platform, against your own baseline. Not against a category benchmark — against what you normally do.
Did it hold them? Retention or watch-through where you can see it, engagement per view where you can't. A piece with high views and a low rate reached people who didn't care.
Did it do the job? Against the goal you set. If the goal was awareness in a specific audience, conversions are the wrong test — and running that test anyway is how good top-of-funnel work gets killed.
Was it worth the money? Cost per outcome, compared against your realistic alternatives — not against a theoretical perfect campaign.
The attribution problem, stated honestly
Someone will ask whether the video drove revenue. The honest answer is usually that you can't cleanly know, and the useful move is to say so plainly rather than produce a number that implies more than it contains.
Content sits early in most purchase decisions, gets consumed on platforms that don't pass identity, influences people who convert weeks later through a different channel, and works partly by making other channels more effective. Last-click attribution systematically undercounts it. Multi-touch models spread credit by rules someone chose.
Three defensible things you can do instead:
Measure the thing content directly causes. Reach in a defined audience, engagement, subscriptions, follow-on content consumption. These are real, measurable and honestly attributable.
Ask people. A "how did you hear about us?" field is unfashionable, imprecise, and frequently the best signal available for content influence.
Look for the correlation, and label it as one. If content output and pipeline move together over several quarters, that's worth noting — as a correlation, with the confounds named. It isn't proof, and presenting it as proof is the kind of thing that gets the whole content function distrusted when someone eventually checks.
Judging the creator versus judging the content
Two different questions that get collapsed constantly.
Did the creator do their job? Did they deliver what the brief asked, on time, at the agreed quality? This is entirely knowable and it's what a rating covers.
Did the content perform? Depends on the concept, the distribution, the timing and the audience — most of which the creator didn't control.
A creator can execute a flawed brief perfectly. The piece underperforms; they did their job. If you rate them down for it, you're rating your own brief, and you'll lose access to good people for a reason you've misdiagnosed.
Conversely a piece can perform well because the topic was hot, with mediocre execution. Notice that too.
What one piece can tell you
Not much, and this is worth saying because a single creator project usually gets treated as a verdict on creator work generally.
One piece is one data point in a wildly variable distribution. The three tests apply here as anywhere: is it outside your normal range, is it explained by something else, does it hold on more data?
If you're going to judge whether creator content works for you, judge a set of three or four. One is noise, and killing the approach on it is the most common way brands conclude something doesn't work when they never tested it.
Where Acumin fits
A booking can carry the live post once the work goes out, and the resulting figure is labelled by source: measured where Acumin fetched it from a platform it can read — YouTube today — and reported where the number was typed in by the brand. A verification mark appears only on measured numbers, never on reported ones.
That distinction is the whole discipline of this article, enforced structurally rather than by convention. The two never blend into a single confident figure.
The performance figure also travels: where a brand linked the live post, it can appear on the creator's referral card alongside the rating — which is how a creator ends up with portable proof of real-world results rather than a claim.
What Acumin doesn't do is attribute revenue to a piece of content. There's no model for that in the product, deliberately, because any number it produced would be an invention dressed as a measurement.
How to use this today
On your current creator project, answer one question in writing before the work goes live: how will we know if this worked?
If the answer requires data you won't have access to, fix that now — a line in the agreement about sharing analytics costs nothing today and is impossible to add later.
Related: Measured vs reported is the distinction this whole article rests on. Why booking-gated ratings mean something covers judging the creator rather than the content.