The campaign report lands on the finance director's desk. Twelve creators, a healthy fee bill, and discount codes showing a return of about 1.6 times spend. Not terrible, not impressive, and certainly not enough to justify doubling the budget. What the report cannot show is the customers who watched a creator's video, did not bother with the code, and bought three days later after searching for the brand. Or, just as often, the customers who were going to buy anyway and simply used a code they found on a coupon site.
That is the core problem with measuring influencer marketing, and it has a practical answer. Stop asking how much a creator's code sold, and start asking what would have happened without them. Codes and tracked links remain essential, but they are a floor, not the full picture, and occasionally not even a reliable floor. Good measurement stacks several signals so the number you report is defensible, and saves the most rigorous method for the programs big enough to justify it.
The error runs in both directions
Discount codes and tracked links are the workhorses of influencer measurement, because they are cheap, simple and work at any size. But they only see last-click behaviour, and they miss a great deal. Someone who sees a post, screenshots the product and searches for it a week later is invisible to the code. So is anyone who buys in a shop, types your web address directly, or clicks a retargeting ad after the creator's content did the persuading. Social platforms do not always pass tracking parameters cleanly either, especially on mobile.
The less discussed error runs the other way. Codes can over-credit a creator. Customers who were already on their way to checkout apply any code they can find, and creator codes routinely leak onto coupon-sharing websites, where they are used by people who never saw the content at all. The creator looks more effective than they were, and you give away margin on sales you would have made regardless. It is the same leakage dynamic we covered in managing coupon partners without cannibalising full-price sales, and it is why a code report on its own can be wrong in either direction.
The measurement ladder
The answer is not one perfect method but a ladder, where each rung captures something the one below misses. Smaller programs climb the first two rungs; larger, ongoing programs should reach the top.
| Method | Captures | Misses |
|---|---|---|
| Discount codes | Direct, traceable purchases | Buyers who skip the code; leaked codes inflate it |
| Tracked links and UTMs | Clicks and post-level conversions | Anyone who buys later without clicking |
| Halo signals | Branded search, direct traffic, survey answers | Precise cause and effect |
| Incrementality test | Sales the campaign actually caused | Needs volume, planning and a clean calendar |
One metric deliberately absent from the ladder is earned media value, the estimate of what your impressions would have cost as paid advertising. It has no standard formula, different tools produce very different figures for the same campaign, and it measures exposure rather than any business result. Finance teams tend to discount it immediately, and reports that lean on it lose credibility.
Mangools
Simple and affordable keyword research suite loved by bloggers and freelancers
Best for: Keyword Research
Set up tracking before launch
Most measurement problems are created before a campaign starts, not after it ends. Give every creator a unique code and a unique link, and if they post across several platforms, a separate link per platform so you can see which format drove results. Agree a consistent UTM naming convention so creator traffic is grouped properly in analytics rather than scattered across a dozen spellings. Send traffic to a landing page that reflects the creator's audience and offer, not your generic homepage.
Two more steps make everything afterwards easier. Record a baseline for branded search volume, direct traffic and new-customer sales for several weeks before launch, because you cannot measure a lift without knowing the starting level. And put measurement into the agreement itself: which metrics count, how sales are credited and for how long, as covered in creator contracts, with success criteria set in the creator brief.
Capture the halo
The sales that codes miss still leave traces, and three are worth watching around every campaign. Branded search: a creator's audience that is interested but not ready to click will often search for you by name. Direct traffic: people type in your web address after hearing it mentioned. New-customer volume: if new customers rise while other activity is flat, something is driving them.
The most underrated tool is also the cheapest: a post-purchase survey asking new customers how they first heard about you. Include creator names or a free-text box alongside the usual options, and ask at the same moment you might invite a referral, when goodwill is highest. Self-reported answers are imperfect, but they routinely surface creator influence that no tracking link recorded, which is exactly the problem described in zero-click content. Survey responses also become useful first-party insight about how customers really find you, as part of a first-party data strategy.
The question Stop asking how much a creator's code sold. Start asking what would have happened without them. Codes are a floor, not the full picture, and a leaked code can even make a creator look better than they were.
Test for what the campaign caused
For larger or ongoing programs, the gold standard is an incrementality test, and the most practical version is a geographic holdout. You run the campaign in a set of test regions and keep comparable control regions free of it, then compare how sales moved in each over the same period. Because it compares regions rather than tracking individuals, it still works when cookies, device IDs and links fail.
Creators are activated in a set of test regions; a matched set of control regions sees no campaign.
→ Test regions: sales go from $100,000 to $130,000 during the campaign (+30%).
→ Control regions: sales go from $100,000 to $110,000 over the same weeks (+10%, the normal trend).
→ Without the campaign, test regions would likely have reached about $110,000.
→ Incremental revenue ≈ $130,000 − $110,000 = $20,000.
Campaign cost: $5,000 → incremental return ≈ 4.0×.
Discount codes alone credited $8,000 → an apparent return of just 1.6×.
Same campaign, very different story. In this example codes understated the effect. In others, leaked codes overstate it. Only the holdout tells you which.
Incrementality testing is not free or universal. It needs enough sales volume and enough comparable regions to produce a reliable result, and it can be distorted by other promotions running at the same time, so coordinate with the rest of the marketing calendar and design the test before launch rather than reconstructing it afterwards. For a single small campaign, disciplined codes, links and halo tracking are the honest limit of what you can measure, and that is fine as long as you present it as such. A simpler before-and-after comparison can give a directional read, but treat it with caution, because seasonality and other activity can easily masquerade as lift.
Judge creators on customers, not clicks
Sales are not all equal, and the most useful creator comparisons go beyond revenue. Look at the share of each creator's buyers who were new to the brand rather than existing customers using a discount. Look at repeat purchase: do the customers a creator brings come back, or do they buy once at a discount and disappear? That question is exactly what a cohort analysis answers, grouping customers by the creator who acquired them.
Look at discount dependency too: a creator whose buyers only ever purchase with a code is building a different kind of customer from one whose audience pays full price later. And count the value of the content itself when you have the rights to reuse it in ads or on your site. These measures often reorder a creator leaderboard entirely, and they explain why the right creators, found through careful vetting, are worth more than their code numbers suggest. Where creators sell directly through platform shops, as in creator-driven social commerce, in-platform attribution adds another useful signal.
Report it as a range
The most credible influencer reports do not claim false precision. Present directly tracked sales as the floor, supporting evidence from branded search, direct traffic and survey answers as the likely additional effect, and incremental lift from any holdout test as the strongest estimate of true impact. A range backed by several independent signals is far more persuasive to a sceptical finance team than a single suspiciously neat number.
This is the same measurement humility we argued for across marketing in why attribution is getting harder, and it fits naturally into an end-to-end funnel audit where influencer activity is judged alongside every other channel rather than in isolation.
The short version
Discount codes and tracked links are the backbone of influencer measurement, but they are a floor rather than the full picture, and the error runs both ways: they miss people who see a post and buy later through search or directly, and they over-credit creators when existing customers apply codes or codes leak onto coupon sites. Measure in layers. Set up tracking before launch with a unique code and link per creator and platform, consistent UTMs, a dedicated landing page and a pre-campaign baseline. Capture the halo through branded search, direct traffic, new-customer volume and a post-purchase "how did you hear about us" question. For larger programs, run a geographic holdout test to estimate the sales the campaign actually caused, since in a worked example codes showed a 1.6× return while the holdout showed about 4×. Skip earned media value as a return metric. Judge creators on the customers they bring: new-to-brand share, repeat purchase and discount dependency. Report results as a range supported by several signals, not a single precise number.
Want to know what your creators really sell?
We help brands set up measurement that proves influencer impact to finance.
Explore Performance Marketing →Frequently asked questions
How do you measure sales from influencer marketing?
Use layers rather than a single method. Start with direct tracking: give every creator a unique discount code and a uniquely tagged link, with consistent UTM parameters, and send traffic to a landing page built for that audience. That captures the sales that can be traced directly. Then capture what direct tracking misses by watching branded search volume, direct website traffic and new-customer numbers around each campaign, and by adding a post-purchase question asking new customers how they heard about you. Finally, for larger or ongoing programs, run an incrementality test, such as activating creators in some regions while holding others back as a control, so you can estimate how many sales the campaign actually caused rather than simply recorded. Treat code and link results as a minimum rather than a complete picture, and report influencer impact as a range supported by several signals instead of one precise figure.
Do discount codes accurately measure influencer ROI?
Not on their own, and the error can run in both directions. Codes undercount because many people see a creator's post, do not use the code, and buy later by searching for the brand, typing the web address directly, or visiting a shop, and none of those sales are credited to the creator. Platforms also do not always pass tracking parameters cleanly, especially on mobile. But codes can also overcount: some customers who were already going to buy simply apply a code they found, and codes regularly leak onto coupon-sharing websites, where they are used by people who never saw the creator at all. That second problem makes a creator look more effective than they were and quietly erodes your margin. Codes remain valuable because they are cheap, simple and work at any size, but they should be treated as one signal among several rather than the definitive measure of return.
What is incrementality testing for influencer campaigns?
Incrementality testing measures how many sales a campaign actually caused, compared with what would have happened anyway. The most practical form for influencer marketing is a geographic holdout: you run the campaign in a set of test regions while keeping a set of comparable control regions free of it, then compare how sales changed in each over the same period. The difference between the change in test regions and the change in control regions is the incremental effect, and dividing incremental revenue by campaign cost gives incremental return on ad spend. Because it compares regions rather than tracking individuals, it works even when cookies, device identifiers and tracking links fail. It does need enough volume and enough comparable regions to produce a reliable result, and other promotions running at the same time can distort it, so coordinate your calendar and plan the test before launch rather than trying to reconstruct it afterwards.
Is earned media value a good influencer metric?
Not as a measure of return. Earned media value estimates what the impressions and engagement from influencer content would have cost if you had bought them as advertising, but there is no standard formula, different tools produce very different figures for the same campaign, and it measures exposure rather than any business outcome. That makes it easy to inflate and hard to defend when finance teams ask what a campaign delivered. It can still offer a rough sense of reach for awareness campaigns, but it should never be presented as revenue or profit. Reports that lean on earned media value tend to lose credibility quickly. A more credible approach is to report directly tracked sales as a floor, supporting signals such as branded search and survey responses, incremental lift where you can test for it, and the quality of customers acquired, such as whether they were new to the brand and whether they came back to buy again.
