Guide
Influencer Selection
Brands
Scale the creators who sold, repeatedly, at an acceptable cost per order — in that order of priority. Repeat performance across separate posts is the single most predictive signal, because it separates genuine selling ability from one lucky post. Volume of sales matters less than consistency and unit economics: a small creator who converts reliably is usually worth more budget than a large one who produced a single spike. Engagement rate, follower count and how much you personally liked the content are all weak predictors and should not drive the decision.
A creator test produces a spreadsheet. The value is not in the spreadsheet — it is in the decision about where the next round of budget goes, and that decision is made badly more often than it is made well.
This article is about reading the results. If you have not run the test yet, start with how to test many influencers without big upfront fees.
| Signal | What it tells you | Weight |
|---|---|---|
| Repeat performance across separate posts | Genuine selling ability rather than one good day | Decisive |
| Cost per order | Whether scaling this creator is affordable | Very high |
| Conversion rate per click | Whether the content pre-qualified the buyer | High |
| Total tracked sales | Raw contribution — but confounded by audience size | Medium |
| Return and cancellation rate | Whether the sales stick | Medium, and often ignored |
| Reliability and communication | What scaling will actually be like | Medium |
| Engagement rate | Audience responsiveness, not purchase intent | Low |
| Follower count | Potential reach only | Low |
| Whether you liked the content | Almost nothing | None |
A single strong post is genuinely ambiguous. It might be the creator's persuasive ability. It might equally be an algorithmic surge, a well-timed trend, a seasonal spike, or one enthusiastic comment thread. You cannot tell from one observation.
Two or three posts performing consistently is a different kind of evidence. The circumstances varied; the result did not. That is a property of the creator rather than of the moment, and it is the thing you are trying to buy more of.
This has a practical implication for test design: a test that only gives each creator one post cannot answer the most important question. If your results are all single posts, the correct next step is usually a second round with the top handful rather than immediately scaling any of them.
Raw sales totals will rank your largest creators first, which tells you mainly that they have larger audiences. Three normalisations make the comparison fair.
Cost per order. Total cost of the collaboration, including fee, commission and platform fee, divided by orders. This is the number that determines whether you can afford more of this creator. Related: blended CPA across a multi-creator campaign.
Conversion rate per click. Orders divided by clicks. This isolates persuasion from reach, and it is where smaller creators frequently beat larger ones — a tightly matched audience arriving pre-sold converts better than a broad audience arriving curious.
Sales per thousand followers. Crude, but it surfaces the creator who converted a small audience unusually well and would otherwise be invisible next to a large account's raw total.
An illustration, with hypothetical figures: a creator with 80,000 followers producing 40 orders looks better than one with 6,000 followers producing 14 — until you see that the second converted 9% of clicks against the first's 2%, and cost a third as much per order. The second is the one to scale. The first may still be worth keeping for reach.
Engagement rate. Measures whether an audience reacts, not whether it buys. High-engagement communities can be entirely non-commercial. See engagement rate, and what it does and does not tell you.
Follower count. A ceiling on reach, not a predictor of conversion. The size bands are discussed in nano, micro, macro or mega.
Content quality as you judge it. Marketing teams consistently prefer polished content, and direct-response performance frequently comes from the rougher, more specific, more argumentative post.
One enormous week. Check whether it coincided with your own promotion, a payday, a seasonal peak, or a discount that would have converted regardless.
Sales without margin. A creator driving volume on your worst-margin products, or into a high return rate, may be costing you money. See why a high return rate can make a deal unprofitable.
Sort every creator in the cohort into one of four groups, then act differently on each.
| Group | Pattern | Action |
|---|---|---|
| Scale | Sold repeatedly, cost per order at or below target | Raise the guarantee, increase frequency, buy content rights, discuss an ongoing arrangement |
| Retest | One strong result, or strong conversion on low volume | One more round on the same terms before deciding |
| Reframe | Good clicks, poor conversion | The problem may be the brief, the landing page or the offer — change one variable and retest |
| Stop | Little traffic and no sales across multiple posts | Close it out well; say thank you and be specific |
The Reframe row is the one most often misfiled as Stop. A creator who sent real traffic that did not convert has done their job — the failure is further down the funnel and is yours to fix.
Scaling a creator is not simply "more posts". Four distinct moves, roughly in order of how much they change:
Scaling on one result. The most common and most expensive error. Verify that it repeats.
Cutting everyone who did not sell immediately. Some categories need repeated exposure before the first purchase. Judge over the window you agreed.
Keeping a creator because the relationship is pleasant. Note it as a genuine benefit, then look at the numbers separately.
Scaling without raising the offer. Asking a proven performer for more work on the original terms is how you lose them to a competitor who read their numbers correctly.
Never revisiting the decision. Performance changes as audiences and algorithms move. Re-run the analysis periodically — see running a structured retrospective.
This section is Make Influence's own operational view, not an industry standard.
The view we hold most strongly here is that follower count is a proxy that should be discarded as soon as you have something better. Before a creator has worked with you, audience size is one of the few observable signals, so it gets used. After two or three tracked posts, you have direct evidence of whether they sell — and continuing to price on audience size at that point is choosing the worse information.
The practical consequence is that budget should migrate from "big accounts" towards "proven sellers" over the life of a programme. In our experience that reallocation is where most of the improvement in a brand's results comes from, more than from any individual campaign being run better.
We would also flag the decision most brands get wrong in the other direction: cutting a creator who sent good traffic that did not convert. That is diagnostic information about your funnel, and discarding the creator throws away the messenger.
How many posts before I can judge?
At least two, ideally three, spread far enough apart not to share the same moment. One is an anecdote.
What if a creator sells well but is difficult to work with?
Price it. Reliability has real administrative cost. If the performance is strong enough it may still be worth it; if it is marginal, it is not.
Should I tell creators they were not selected?
Yes, clearly and kindly. Specific feedback costs you nothing and makes them likelier to say yes next time. See ending a partnership without burning the relationship.
Can I use engagement at all?
As a tiebreaker between creators with similar sales performance, or as an early signal before sales data exists. Never as a substitute for it.
What about creators who drive sales I cannot track?
Real and easy to under-reward. Watch branded search and direct traffic around their posting dates — see measuring branded search lift.
Make Influence
Find creators with real audience data, run collaborations in one place, and see clicks and sales per creator while the campaign is live.
Book a demoCreate accountMake Influence
Apply to campaigns from brands that are actively looking, follow your own clicks and sales, and get paid without chasing invoices.
Create creator profileMore creator guidesMake Influence
Briefs, agreed terms, tracking links and results sit together — so brands and creators see the same numbers.
See how it worksBrowse the Academy
