Views are the metric that is easiest to collect and least connected to money. A post can clear two million views and move nothing; a creator with a tenth of the reach can carry a product line. The fix the serious programs are converging on is to judge creators by the revenue their content actually drives — net of returns and refunds, set against the full cost of working with them, not just the posting fee. That is creator-driven net revenue, and the phrase hides three decisions that most teams get wrong. First, net versus attributed versus gross: attribution software happily credits a creator for sales it recorded a touchpoint on, before you subtract the returns, the discounts, and the overlapping credit that UTMs, promo codes and retargeting all claim at once. Second, the cost side: the honest denominator includes product seeding, usage-rights licensing and agency fees, not the line-item talent fee, so a creator whose audience returns half the order does not outrank one whose audience keeps it. Third, and hardest: attribution tells you which sales got credited, never whether they would have happened anyway — which is why net revenue has to be paired with incrementality testing to tell net-new demand from demand you already had. This guide defines the metric precisely, walks the net-versus-attributed-versus-gross distinction, builds the full-cost denominator, shows why engagement is a leading indicator and not a result, works the attribution-versus-incrementality ROI math with a concrete example, lays out a measurement stack you can defend in a boardroom, and ends on the constraint underneath all of it — that you cannot measure net revenue per creator or per format until you are producing enough tracked, consistent, owned content for the numbers to mean something.
Everyone agrees the point of a creator program is revenue. Almost nobody reports it that way, because views are free to collect and net revenue is work. So the deck shows reach, impressions and engagement rate, and the one question that matters — did this creator make us money after the returns, the discounts and the full cost of working with them — goes unanswered. Creator-driven net revenue is the discipline of answering it: judging each creator, and each content format, by the profit their content drove rather than the attention it earned. The phrase sounds simple and hides three decisions, each of which has an easy wrong answer: what "revenue" means, what "cost" means, and whether the revenue would have happened anyway. Get those three right and you can rank a roster by contribution. Get them wrong and you will confidently fund the creator who looks busiest.
The first decision is which revenue you are even counting. Attributed revenue is the figure an attribution tool hands you: it recorded a touchpoint — a UTM-tagged click, a promo code redemption, a shop-tag event — and credited the creator with the order total. That number is almost always too high, for two compounding reasons. It is gross: it has not had returns, refunds and discounts subtracted, so a creator whose audience returns half of what they buy looks identical to one whose audience keeps it. And it is double-counted: UTMs, promo codes, affiliate links and paid retargeting routinely claim the same sale at the same time, which inflates the roster-wide total above the real one. Net revenue is what is left after you subtract the returns and the discounts and resolve the overlapping credit. It is a smaller, harder, truer number, and it is the only one that maps to money you actually kept.
The practical rule: never compare creators on gross attributed revenue. A creator selling a low-return, full-price product will always look worse on gross-adjusted math than they are, and a creator whose discount code drives heavily refunded impulse buys will always look better. Net revenue closes both gaps. If your tooling only reports attributed order value, treat it as a ceiling and insist on the net figure before you make a renewal decision.
Revenue is only half of "net revenue per dollar of cost," and the denominator is where most programs quietly cheat. The talent fee is the obvious cost, but it is rarely the whole one. The honest denominator also includes product seeding — the real cost of goods you shipped for the creator to feature — usage-rights and licensing fees if you ran their content as paid ads, and any agency, platform or management fees that sat between you and the creator. A creator who posts for a modest fee but requires expensive seeding and broad paid usage can cost multiples of a flat-fee creator, and the per-post number will never show it.
The metric that falls out of this is net revenue per dollar of cost: attributed net revenue divided by fully-loaded cost. It is the single most useful cross-roster comparison because it normalises creators of wildly different sizes and deal structures onto one profitability axis. A nano creator with a small audience and a small net-revenue total can beat a macro creator on this ratio because their cost is a fraction of the macro deal — which is exactly the kind of counterintuitive, money-relevant finding that views-based reporting will never surface. Tie a 90-day retention or repeat-purchase check onto it where you can, so a creator whose buyers churn immediately does not read the same as one whose buyers stick.
None of this means you throw away views and engagement — it means you demote them to what they are. They are leading indicators: signals that an audience is interacting rather than passively scrolling, that the content fits the audience, that there is intent upstream of a purchase. A healthy save rate, meaningful comments and replies that go beyond an emoji are early evidence a creator might drive revenue. But they are inputs, not the output. The failure mode is treating the leading indicator as the result — celebrating a post that cleared millions of views and produced no measurable conversions, while under-funding a creator with a tenth of the reach who carries a product line. Weight engagement toward the interactions that correlate with intent, use it to forecast and to catch problems early, and keep net revenue as the number you are actually optimising. Our engagement rate definition covers how to weight those signals so they predict rather than flatter. For the broader pattern of judging creator deals by participation and outcomes rather than reach, see creator partnership participation and the shift from reach to trust.
Here is the decision that separates a measurable program from a self-congratulatory one. Attribution — even clean, net, deduplicated attribution — only tells you which sales got credited to a creator. It cannot tell you whether those sales would have happened without the creator at all. A creator can score beautifully on attributed net revenue while mostly intercepting demand you already had: catching buyers who were going to purchase anyway and tagging the last click with a promo code. That is recycled demand wearing the costume of new demand. The number that actually justifies the spend is incremental net revenue — the sales that would not have occurred without the creator — and attribution structurally cannot measure it.
The gap between the two is not academic; it changes the answer. Take a campaign that costs $50,000 and gets credited with $150,000 in net revenue at a 50% contribution margin. On the attributed basis, marketing ROI looks strong — you spent $50,000 to generate $150,000 in credited net revenue. Now run a lift study and find that only $100,000 of that revenue was genuinely incremental — the rest would have converted without the campaign. At a 50% margin, $100,000 of incremental revenue is $50,000 of incremental contribution, exactly equal to the $50,000 you spent: an incremental ROI of zero for the period. Same campaign, same spend, two opposite verdicts. The attributed view says "fund more"; the incremental view says "this barely washed its face." The figures are illustrative, but the shape is the point — attributed ROI is almost always the more flattering of the two, and it is the one most decks report.
Incrementality does not require exotic tooling. The two workhorse designs are a geo holdout — run the creator activity in some regions and deliberately withhold it in comparable ones, then compare sales — and an audience split, where exposed and control groups are matched and measured against each other. Both give you a causal estimate of net-new revenue rather than a credited one. You do not run them on every post; you run them periodically to calibrate how much of your attributed net revenue is real, then apply that discount to the attributed numbers you track continuously. The partner- and affiliate-measurement world has converged on exactly this distinction — attribution explains who touched a conversion, incrementality explains who caused it — and the IAB has repeatedly flagged fragmented, inconsistent measurement as the main barrier to evaluating creator spend next to other major media.
No single method answers every question, so the defensible programs layer them. Multi-touch attribution, run on net revenue with deduplicated credit, is the continuous day-to-day signal for optimising which creators and formats to lean into. Incrementality testing, run periodically via geo holdouts or audience splits, is the causal check that keeps attribution honest and tells you how much of the credited revenue is actually new. Marketing mix modelling, where you have the scale for it, is the top-down view for strategic budget allocation across channels. Underneath all three sits clean tracking hygiene: consistent UTM parameters on every creator link, promo codes mapped to specific creators, and shop-tag events wired to the right source, so the inputs to every model are trustworthy. The output you report upward is net revenue per dollar of fully-loaded cost, discounted by your measured incrementality rate — a number that holds up when finance asks what the program returned, because it already subtracted the returns, the full cost, and the demand you would have captured anyway. The companion data on where clicks stop converting is a useful reality check on how big the click-to-purchase gap really is.
Everything above assumes a problem you may not have yet: enough tracked content to measure. Net revenue per creator is a ratio, and a ratio built on three posts is noise. To rank a roster, compare formats, or run a geo holdout with any confidence, you need volume — many tracked shots on goal, posted consistently, each carrying the attribution plumbing — and you need it sustained long enough for the retention check to mean something. That is a production problem standing in front of the measurement problem, and it is where most creator programs stall before they ever get a clean number. It also exposes the cost side from a different direction: the denominator of net-revenue-per-dollar is dominated by what each post costs to produce and license, so content you own and can reproduce at near-zero marginal cost changes the math in your favour.
That is the leverage Kompozy adds to this specific metric. It is a full AI content generation and multi-platform publishing engine, so a persona you define once becomes a steady supply of owned creator content across eighteen formats — Persona Shorts and clipped video, carousels, quote graphics and image posts, blogs and text posts — produced in the volume that makes a net-revenue comparison statistically honest rather than anecdotal. Because that content is generated against a written Persona Brief, the brand reads the same across every post, which is what lets you attribute cleanly: you are comparing like with like, not confounding a format test with an inconsistent voice. And because autopilot fans each output across eight social platforms plus blog and email behind a per-post review gate, every piece can ship with its own UTM and tracked destination without someone hand-posting to each channel — so the attribution data arrives structured instead of reconstructed after the fact. The honest boundary: Kompozy does not run your incrementality tests, resolve your overlapping credit, or build your attribution model — those are yours. What it removes is the production wall that otherwise leaves you measuring net revenue on a sample too small to trust, and it lowers the cost denominator by making the owned half of your creator content reproducible. The metric points at profit; this is how you generate enough tracked, consistent content to compute it.
Creator-driven net revenue is the right thing to optimise and the hard thing to measure. Judge creators by the sales their content drove, net of returns and discounts, against the full cost of working with them — seeding, usage rights and agency fees included — not by views or engagement, which are leading indicators you demote to forecasting. Never report gross attributed revenue as if it were profit; resolve the double-counted credit and subtract the returns. And never trust attribution alone, because it credits sales it cannot prove it caused — pair it with periodic incrementality testing so you know how much of the credited net revenue is genuinely new. The binding constraint underneath the metric is not a cleverer dashboard; it is producing enough tracked, consistent, owned content for the net-revenue number to stop being noise.
It is a way of measuring a creator by the revenue their content actually drives — the sales you can attribute to their posts or allowlisted ads, net of returns, refunds and discounts — set against the full cost of working with them, rather than by views or engagement. The point is to judge creators on profit contribution instead of reach, so a high-view creator whose audience does not buy ranks below a smaller one whose audience does.
Attributed revenue is the sale total your attribution tool credited to the creator because it recorded a touchpoint — a UTM click, a promo code, a shop tag. Net revenue is what survives after you subtract returns, refunds and discounts, and after you resolve the fact that UTMs, codes and paid retargeting often claim the same sale at once. Attributed revenue almost always overstates; net revenue is the number that maps to actual money kept.
Because they measure attention, not purchase. A post can earn millions of views and produce no measurable conversions, while a lower-reach creator with a tightly matched audience drives sales. Views and engagement are useful as leading indicators — signals that an audience is interacting rather than scrolling past — but they are inputs to revenue, not a proxy for it. Judging a creator on views rewards the wrong behaviour.
Count the full cost, not the talent fee. That means product seeding (the cost of goods you send), usage-rights and licensing fees for running the content as ads, and any agency or management fees, on top of the posting payment. Dividing net revenue by that full cost gives net revenue per dollar of cost — the ratio that tells you which creators are actually profitable, not just which ones posted cheaply.
Because attribution tells you which sales got credited to a creator, never whether those sales would have happened without them. A creator can look highly profitable on attributed net revenue while mostly intercepting demand you already had. Incrementality testing — a geo holdout or an audience split that compares exposed groups to a control — estimates the net-new revenue the creator caused, which is the number that justifies the spend.
Creator-driven net revenue measures a creator by the sales attributable to their content net of returns, refunds and discounts, divided by the full cost of working with them — product seeding, usage-rights licensing and agency fees included — instead of by views or engagement. Because attribution credits sales a creator may not have caused, pair net-revenue attribution with incrementality testing, such as a geo holdout or audience split, to separate net-new demand from demand that would have converted anyway.
Get started → · ← All guides · Compare Kompozy vs other tools