# Measuring influencer marketing ROI without inventing it > Codes, tracked links, whitelisted ads and post-purchase surveys each see a different slice of a sale. How to choose the method before launch, and how to report what it cannot see. - Tag: Benchmarks - Published: 24 September 2026 - Reading time: 8 min - By: Desk Influencer marketing return on investment is easy to calculate and hard to believe. Divide something by something and you have a number; whether it deserves to be believed depends on how the sale was seen in the first place. The method comes first, and the number is only as good as it is. ## Decide the method before the first post Attribution is built before a campaign, not after it. A tracked link has to be in the bio or the caption before the post goes live; a promo code has to be issued before the creator records it; a survey question has to be on the checkout page before the first sale. Add any of them afterwards and the first week of sales is already uncounted. Write down, in the brief, what counts as a sale attributed to a creator and which method will see it. Then write down what the method cannot see. That second list is the part most reports skip, and it is the part that stops a disagreement later. ## Codes, links, whitelisting, surveys A promo code travels. A creator says it on camera, a viewer types it at checkout, and the sale is credited to the creator. It counts people who were willing to use a code. A tracked link — a UTM-tagged URL — counts clicks that arrive through it and the sales that follow in the same browser session. Whitelisted ads, where the creator’s post runs as a paid ad from their handle, give the brand the platform’s own ad reporting: clicks, view-through, and conversions the ad platform can see. A post-purchase survey asks the buyer directly how they heard of the brand, and catches the people who saw a creator, did not click, and searched later. ## What each one misses Codes miss everyone who sees the post and buys without one, and they leak: a code posted to a coupon site is credited to a creator who never sent that buyer. Links miss anyone who watches, closes the app and later types the brand name into search. Whitelisted ads see only the ad’s own audience and say nothing about the organic post alongside it. Surveys capture memory, and memory is imprecise: a buyer who saw three things names whichever was most recent or most vivid. They also depend on response rates that vary by checkout design. No method is wrong. Each looks at a different slice of the same sale, and the honest report says which slice it is looking at. ## Reach is not sales Reach counts accounts that saw a post; sales count people who bought. Between them lie attention, intent and price, none of which reach measures. A campaign can have wide reach and thin sales, or narrow reach among exactly the right buyers. If a report leads with reach and closes with a cost per thousand, it is describing exposure. That is useful for awareness work and misleading for anything with a checkout. Be equally careful with engagement. Likes and comments are signs of an audience paying attention, not of an audience buying. Saves and shares tend to be more informative than likes, but they are still a proxy and should be labelled as one. ## What a weekly report should say For each creator and each method, the figure, how it was counted, and whether it is verified. Where a method has not yet reported, say so. Where a figure could not be cleared, the cell says “Metrics on request” rather than carrying a guess. And include a plain sentence on what the numbers cannot tell you this week. Our own format, the Monday sheet, carries reach, engagement rate, cost per acquisition, CPM and the next move; a sample is on /for-brands/sample-sheet. The report should end with a decision: what continues, what changes, what stops. A report that cannot recommend anything is a record, not a tool. ## When two methods disagree They will. A code reports one count, a link another, and the survey a third. Do not average them and do not choose the flattering one. Report each against what it sees, and read the gap as information: a large code count with few link sales suggests an audience that buys without clicking, and a survey that names creators the other methods did not see suggests influence that arrives late. Direct-to-consumer brands, where the checkout is the brand’s own and every method above is available, will find the fuller picture on /influencer-marketing-for-d2c-brands. For the definitions behind the words in a report, read /blog/what-we-count-as-reach. Where a figure has no record behind it, we do not publish it, and neither should a report that is meant to be trusted. ## More field notes. - [Planning festive-season influencer campaigns in India](https://vexo.club/blog/festive-season-influencer-planning) - [How to brief an influencer marketing agency in two paragraphs](https://vexo.club/blog/how-to-brief-an-influencer-marketing-agency) - [Retainer, commission or fixed fee: how influencer agencies charge](https://vexo.club/blog/influencer-agency-fee-models) --- Source: https://vexo.club/blog/measuring-influencer-marketing-roi