How we measure creator value.

Where this came from

We built matchmaking because a category filter failed in front of us. Putting a shortlist together for a luxury resortwear label, we filtered on womenswear — and got fashion creators with nothing whatsoever in common with the brand. Different customer, different price level, a completely different look. Every one of them correctly tagged, and not one of them right. A category is not a match, and that is the whole problem.

Two cards, compared

• Your brand card. We read your website, your product catalogue and your product photography, and write down what you actually sell: the products, who buys them, the price level you sit at, and how the brand looks and feels. You never fill in a form to get this — it comes from what you have already published.
• Every creator’s card. Built the same way, from what a creator actually publishes — their posts and how they describe themselves: the subject matter, how they treat it, who they are talking to. Their follower count is not in it, and neither are their brand deals — those are read fresh at the moment you search, so they can never go stale in a way we cannot correct.
• The comparison. When you search, we compare your card against the creators’ cards there and then. Nothing is decided in advance, and no one sits and reviews pairings — which is why it can rank a catalogue this size in the time a page takes to load.

What counts as a match depends on what you sell

How much the look matters is not the same for every brand, and treating it as a constant is how you get a beautiful shortlist that sells nothing. For a resortwear label the look carries most of the judgement — the aesthetic a creator projects is very close to the product itself. For a store selling something functional it barely matters, and what does matter is content style: whether this creator explains and demonstrates things to people who are deciding what to buy. The look counts for the brands whose look we can read, and is left out entirely for the rest. It is one reason two stores in the same category get genuinely different shortlists from us.

Where the match goes

Fit is one half of the ranking. It is combined with how good a creator is — a real, genuinely-engaged audience — into AcuScore, the single number the shortlist is ordered by. What a creator is likely to earn you is a separate question with its own answer: Real Reach against your own store economics. Fit decides who belongs on the list. The forecast decides what they are worth.

Why not followers?

A follower count is an account’s lifetime accumulation — people who tapped “follow” once, many of whom never see a given post and some of whom no longer exist. It is the influencer industry’s vanity metric: easy to inflate, weakly related to how many humans a sponsored post will reach, and the number most tools still rank on.

When we measured actual views against follower counts across a large sample of creators, the gap was not small or consistent — it widened sharply with size. Large accounts routinely reach a fraction of their followers per post; anchoring a forecast on followers over-states the audience of older, bigger accounts by an order of magnitude or more. Budget on followers and you systematically overpay the wrong creators.

Why the median of recent posts

A single viral hit (or a single flop) is not what your placement will do. The median of recent posts is the typical outcome — resistant to one spike skewing the picture, and current rather than a lifetime average. It answers the honest question: if you book this creator for one post, how many people realistically see it?

For multi-platform creators, Real Reach sums the per-platform medians. Every figure comes with the evidence behind it, so you always know how much recent posting the number is drawn from.

How to use it

Treat Real Reach as the top of the funnel — the audience a placement buys you. Everything downstream (clicks, customers, revenue, profit) is forecast from it against your store’s economics, and it’s one of the inputs to AcuScore. If you only change one habit: stop budgeting against followers.

Want the plain-English version, or the evidence? Read What is Real Reach? for the definition and how to measure it yourself, and the Real Reach Report for the numbers across hundreds of thousands of creators. If you still budget on engagement rate, this is a good engagement rate by follower tier — measured, not a benchmark table copied from a blog.

Two signals rank. Money joins when the price is real.

A shortlist decision is a single judgement, so AcuScore is a single number. Two independent signals build it:

• Fit — would this creator work for your store? We build a card for your brand from your own site and products, and a card for each creator from what they actually post, then compare the two — subject matter and visual style together — alongside any comparable paid collaborations they have done. Hard rules can veto a creator outright, and a creator never outranks the evidence behind them. The full explanation of how the two cards are built and compared is in matchmaking.
• Quality — is this a real, genuinely-engaged creator? Engagement intensity measured relative to the creator’s own platform. An authenticity term used to sit here; we removed its weight in July 2026 after measuring that it added noise rather than fraud protection, and it returns only when a signal we can stand behind replaces it. Deliberately size-independent: a sharp 40k-follower creator can out-score a sleepy 4M one, because reach is priced separately and shouldn’t double-count here.
• Revenue efficiency — shown, and deliberately not ranked. Beside each creator you also see how efficiently their audience could turn into revenue for you. Before anyone has quoted a price, every input to that estimate is ours — so letting it sort the page would let a creator out-rank a better-matched one purely because we guessed they were cheap. It informs your judgement; it does not order the list.

Why size-independent

Audience size is real, but it’s already captured by Real Reach and priced into the forecast. Letting it also inflate the quality score would just reward bigness twice and bury the efficient mid-tier creators who often pay back best. AcuScore separates “is this creator good and right for me?” from “how big and how much?” on purpose.

Profitability folds in

Once there’s a price on the table — a real quoted fee, not an estimate — AcuScore accounts for whether the deal actually pays back at that price, and a brilliant creator at an unaffordable fee stops out-ranking a well-priced strong one. That is the whole rule, and it is the reason the bar above does not rank: money moves the score exactly when the price is real, and never when it is modelled. The full money picture lives in the forecast.

The chain

Each number comes from the number before it — no black box:

1. Real Reach — the audience a placement reaches (median views of recent posts), not followers.
2. MER — the ratio the forecast is judged against (what a good MER actually is), stated per store rather than as an industry average.
3. → Clicks — the share of that audience that acts, by format and platform.
4. → Customers — clicks at your store’s conversion rate, adjusted for how well the audience actually fits what you sell.
5. → Revenue — customers at your AOV, including realistic repeat behaviour over a year.
6. → Profit — revenue minus your COGS, shipping, discount and the creator’s fee. The only number that decides whether the deal was worth it.

Against your economics — not benchmarks

The same creator is profitable for one store and a loss for another; the difference is the store, not the creator. Connect Shopify and we pull your AOV, conversion and repeat rate automatically; if you’re not on Shopify you enter them and industry benchmarks fill the gaps. Generic benchmarks are the fallback, never the basis.

Before, not after

Attribution tools (UTMs, post-purchase surveys, MMM) tell you what happened after you spent. That’s useful, but it can’t stop you wiring $5,000 to the wrong creator — the money’s already gone. Acurrate runs the same economics before you commit, so the bad deal never gets booked. Then it tracks the real result, so the model stays honest.

Honest by construction

Every forecast carries a confidence tier, so you always know how much evidence stands behind the number. We apply a deliberately conservative attribution buffer, because not every tracked sale is truly creator-driven and a forecast worth trusting errs downward. And every campaign you run logs forecast versus actual side by side, so the model is accountable to reality, not just to a marketing page.