How to Measure Influencer Marketing ROI Honestly

Influencer marketing ROI is gross profit from creator-driven sales divided by the full cost of running the program — not revenue over creator fees, which is the calculation that makes every program look brilliant until finance takes it apart. The honest version states which of three confidence levels the number came from and reports a range rather than a point.

Most disputes about whether creator marketing "works" are not disagreements about performance. They are two people using different numerators and different denominators and neither saying so.

Decide what you are dividing before you divide it

The denominator is where the lying happens, usually by omission. A complete program cost includes:

Cost	Commonly counted?	Notes
Creator fees	Always	The only line most decks show
Product COGS	Rarely	Gifted units are real money, at cost not RRP
Shipping and returns	Rarely	Includes the parcels to creators who never post
Usage rights and whitelisting fees	Sometimes	Often 20–50% on top of the base fee
Paid amplification	Sometimes	If you boosted the content, it belongs here
Agency or freelancer retainer	Usually
Tooling	Usually
Internal time	Almost never	One coordinator at 60% is a five-figure annual line

Omitting product cost and internal time typically understates a gifting-heavy program's spend by a third or more.

The numerator has one rule: use gross profit, not revenue, or the ratio is meaningless across categories. A 2x return on revenue is a loss at 35% margin and a win at 80%. If your leadership insists on ROAS, report both and label them — 2.1x ROAS / 1.1x on gross profit takes one extra line and prevents a year of confusion.

Three levels of confidence, and which to report

Level	What it measures	Typical use	Weakness
Tracked	Sales that carried a code or link	Per-creator decisions	Systematically undercounts
Modelled	Tracked, scaled by a survey or media-mix estimate	Channel-level budgeting	Only as good as the multiplier
Incremental	Lift versus a holdout	Deciding whether the channel exists	Needs scale and a quiet quarter

Run tracked as your operating number, modelled as your reporting number, and incremental once a year at most. Reporting only the tracked figure is the most common way a working program gets cut, because it is the one number guaranteed to be too low. The mechanics of capturing it — code formats, UTM conventions, what each method misses — are in our guide to tracking influencer sales.

The multiplier for the modelled figure comes from a post-purchase survey running continuously: if 22% of buyers say a creator, and your tracked codes account for 9% of orders, creator influence is running somewhere around 2.4× what your codes see. That multiplier is an estimate. Recalculate it quarterly and say in the deck that it is an estimate.

A worked example

Illustrative numbers, not benchmarks — the point is the shape of the calculation.

A quarter's program: 30 creators, mostly gifted with a paid tier on top.

Line	Amount
Creator fees	£18,000
Product at cost	£2,400
Shipping	£600
Usage rights add-ons	£1,500
Tooling and coordinator time	£1,500
Total cost	£24,000

Tracked revenue through codes and links: £41,000. Gross margin 62%, so tracked gross profit is £25,420.

• Tracked ROAS: 41,000 ÷ 24,000 = 1.7x
• Tracked return on gross profit: 25,420 ÷ 24,000 = 1.06x — £1,420 of profit on £24,000. Effectively break-even.
• Modelled, applying a 2.4× survey multiplier: £98,400 revenue, £61,000 gross profit, 2.5x.

The finding is not 1.06 and it is not 2.5. It is: this program is between break-even and comfortably profitable, and the spread is that wide because most of the effect is unattributable. That is a defensible sentence, and it points at the right next action — narrow the spread by improving capture, not by arguing about the multiplier.

One more cut worth doing before you present: strip out the creators who had no code and no link at all. A flat-fee deal with no attribution method is not a 0x result, it is unmeasured spend, and averaging it in as a zero drags a good creator's numbers down for a mistake the brand made.

Why EMV should not appear in your deck

Earned media value multiplies impressions by an invented CPM and calls the product money. It is not money. It cannot be spent, it does not appear in any ledger, and the multiplier is chosen by whoever wants the number to be large.

The specific harm is that EMV is unfalsifiable — a campaign that sold nothing still produces a large EMV, so the metric can never tell you to stop. If someone wants a value on reach, quote what buying equivalent impressions as paid media would have cost, call it equivalent media cost, and keep it out of the revenue line.

The window, the cohort and the distribution

Three habits that fix most reporting arguments:

• Fix the window at 30 days after posting and state it everywhere. Lifetime totals keep drifting upward and cannot be compared between creators or quarters. Seven days catches most short-form effect; 90 days is right for YouTube and for anything with a long consideration period.
• Split new customers from repeats. A code redeemed mostly by existing buyers is a discount to your own list, not acquisition. Report new-customer revenue as the headline and total as the footnote.
• Report the distribution, not the average. Creator results are heavily skewed: a minority of a roster usually produces most of the return, and a meaningful share produce roughly nothing. An average hides both. Show a ranked list, and set the target as "how many creators cleared payback" rather than "what did the program average".

Payback is the cleanest single test: what share of creators returned more gross profit than they cost, inside the window. If that share climbs quarter on quarter, your selection is improving, whatever the blended ratio does.

Getting the number without a reconstruction project

Most ROI exercises are archaeology. Fees live in an inbox, product cost in a shipping spreadsheet, codes in Shopify, usage rights in a signed PDF, and someone spends two days joining them by hand — which is why the calculation happens quarterly rather than weekly, and why unmeasured spend stays invisible until it is too late to fix.

CreatorCast keeps those pieces attached to the creator record as the deal happens: agreed fee and add-ons captured as structured deal terms, gifted product recorded against the same creator, the discount code editable after approval because sales arrive after content goes live, and deals with no attribution method held out of the ROI figure instead of averaged in as zeroes. The report becomes a read rather than a rebuild.

For the planning side of the same numbers, our influencer marketing budget guide covers how to allocate before the quarter starts.

Frequently asked questions

What is a good ROI for influencer marketing? There is no universal figure, but a useful internal bar is contribution-margin positive on tracked sales within 90 days, with the modelled number comfortably above it. Below-1x tracked with a strong modelled figure means your capture is broken, not necessarily the channel.

Should I use ROAS or ROI? ROAS for comparing against paid social, because that is how paid social is quoted. ROI on gross profit for any conversation involving finance. Report both, labelled, in the same table.

How do I measure gifting with no fee attached? Cost is product at cost plus shipping plus your time, not zero. Track it as its own program with its own payback test — gifting usually looks weak on tracked sales and strong on content volume, so also count the ad creative you got and what it would have cost to produce.

How long before an influencer program shows a return? Two to three months for consumer products with codes, longer where consideration is slow. Judging a program in week four mostly measures how fast your onboarding is.