Field guide · 4 of 5 · about 20 minutes · Updated
How to measure influencer marketing ROI
Most campaign reports open with the biggest number they can find. This guide is about the smaller numbers underneath it, the ones that tell you whether anyone cared, and whether anything changed for the brand.
It covers what to measure for each goal, how to read a creator's insight screen line by line, the ratios worth working out, how ROI can and can't be proved, and how to spot numbers that were bought.
What brought you here?
The short answer
Decide the goal before anyone posts, pick two or three numbers that would prove it, then compare them with what they were before. Everything else in a report is decoration.
Each goal needs different numbers and different proof. Pick yours to see what to measure, where the number comes from, and the trap most reports fall into. For the content-to-reuse goal, our guide to UGC content explains how that kind of video works.
More people know you exist
- Measure
- Viewers (unique accounts), views, and the share who weren't already following the creator.
- Where it comes from
- The creator's insight screens, sent within a week of posting.
- Stronger proof
- A brand lift study, or branded searches compared with the weeks before.
- The trap
- Views count every play, including repeats. Ten thousand views can be six thousand people.
What to measure instead of likes
A like costs nothing. The numbers worth watching are the ones that cost the viewer something: a save, a share, a search.
Marketers know this and still default to reach. The World Federation of Advertisers found 94% of its members use reach as a primary KPI and 89% engagement, but only 23% use conversions or sales. In India, Kofluence's 2025 report found engagement was the KPI brands asked for most, at 31%.
Reach and engagement aren't wrong. They're just the start. The useful question is which actions show someone moved closer to buying, and those are rarer and smaller. Sort the eight cards below into the right bucket.
Card 1 of 8
Likes on the creator's postUseful0
Shows someone moved closer to buyingWeak signal0
Costs the viewer nothingDepends0
Useful only in the right contextBig results, and the question each one leaves
- India · 2025Garnier Men and Anubhav Singh Bassi
Garnier turned the comedian's own clip mocking men's face wash into a campaign that won a Bronze Lion at Cannes. The published result was views: more than 400 million across Instagram and YouTube.
Net Influencer, Jun 2025 - UK and South Africa · 2025Vaseline Verified
Unilever's influencer-led series fact-checked skincare hacks people were already posting, and took the Health and Wellness Grand Prix at Cannes. The campaign reported a 43% sales increase and 87% positive sentiment.
Storyboard18, Jun 2025; PPC Land - India · Dec 2025Myntra's shopper-creators
Myntra said shoppers who make content on its platform drive 25 to 28% higher conversion rates, across more than 3.5 million shopper-creators, two-thirds of them Gen Z.
Storyboard18, Dec 2025, quoting Myntra's CMO
How to read a creator's insight screen
The insight screen is the platform's own record of how a post did, and it's the first thing to ask a creator for. Meta changed the names in November 2024, so older reports and newer ones use different words. Tap each line.
Reel insightsExample screen, not a real campaign
Views
48,210
How many times the reel was played, repeats included. Since November 2024 Meta has used Views in place of Plays and Impressions across Instagram and Facebook.
What to askHow many of those were from people who don't follow you?
Definition: Meta, Nov 2024Engagement rate calculator
Engagement rate sounds like one number. It's at least two, depending on what you divide by, and the ANA's guidelines point out that platforms calculate it differently. Change the numbers and watch the rates move.
Free tool · works in your browser
Engagement rate calculator
Pre-filled with the example screen above. Type in a creator's real numbers.
Engagement rate by views
6.8%3,301 interactions ÷ 48,210 views. The fairer default for reels.
Same post, two rates, 2.5× apart. Always say which one you're quoting.
- Save and share rate
- 2.2%The two actions that take some intent.
- Viewers vs followers
- 26.2%How far the post travelled past its audience.
- Plays per viewer
- 1.54Above 1 means people watched again.
There's no single good engagement rate. Compare a creator's post with their own last ten, and only compare creators of a similar size in the same category.
What 30,000 views actually tells you
A view number on its own answers almost nothing. Five questions turn it into something you can act on. Step through them.
Question 1 of 5
Views count plays, not people
- What it means
- Instagram counts a view every time the reel plays, repeats included. YouTube counts one the moment a video starts.
- Ask
- Ask for viewers (unique accounts) as well. That's the number of people.
- The trap
- Adding views across ten creators double-counts anyone who saw two of them.
ROI: what you can and can't attribute
Codes and links catch the people who clicked. Creators mostly work on the people who didn't click yet.
IAB's January 2026 report puts it plainly: promo codes, affiliate links and last-click tracking reward what closes the sale, not what influenced it. Creators do most of their work earlier, when someone first hears of a product or decides to trust it. So direct tracking undercounts them, and the fix is to add a method that measures the rest.
That's why marketers find it hard. In Linqia's 2026 survey, 79% said they struggle to measure influencer ROI, and 48% named attribution as the biggest gap. Seven methods, and what each one can actually prove:
| Method | What it proves | What it misses |
|---|---|---|
| Promo code | Sales from people who typed the code | Everyone influenced who didn't use it, and codes that leak to coupon sites |
| Tagged (UTM) link | Visits and sign-ups from people who tapped the link | Anyone who searched for you later, or tapped on another device |
| Affiliate link | Sales through that link, credited to the creator | Influence before the last click. Browser extensions can swap the link at checkout |
| Brand lift study | Ad recall, awareness and consideration, against a group that didn't see it | Sales. And it needs paid delivery on Meta or YouTube to set up |
| Search lift | Branded searches from people who saw the content | What those searchers did next |
| Holdout or geo test | What the creators added on top of what would have happened anyway | Needs enough cities or people to split, and a few weeks |
| Marketing mix modelling | Each channel's share of sales over months, creators included | Needs at least a year of weekly data. Built for big budgets |
Measured, and how
- India · Jul – Aug 2022Maruti Suzuki NEXA, Grand Vitara launch
Creators made 15-second paid-partnership videos and Stories, which Maruti then ran as ads beside its usual video ads. A Meta brand lift study compared the two setups: brand awareness lift was 4.5 points higher with the creator ads in the mix.
Instagram for Business case study, 2022 - India · 2020Nestlé Maggi, Masala-ae-Magic
Four regional home chefs on YouTube made 20 cooking videos over six months. Nestlé tracked awareness, branded searches and the add-to-cart ratio on its own site: 2.3x awareness, 50% more branded searches and a 1.6x higher add-to-cart ratio.
Think with Google, Oct 2020 - India · 2026Bi-luma launch
Local creators made Shorts and long-form videos for the dermatology brand's launch. A Google Search Lift study measured a 195% lift in brand and product searches, and view-through rate came in 18% above the category norm.
Google, APAC YouTube 2026 case studies - US · Sep 2020Dunkin' and Charli D'Amelio
Dunkin' named a drink after the TikTok creator. On launch day its app downloads rose 57% and cold coffee sales 20%, with cold brew up 45% the next day, the brand's VP told TMZ.
Tubefilter, Sep 2020 - US · Jul 2019Chipotle #GuacDance
A six-day TikTok challenge drew more than 250,000 video submissions and about 430 million video starts. Chipotle served over 802,000 sides of guacamole, its biggest guac day ever.
Chipotle press release, Aug 2019 - UK · Feb 2021Little Moons mochi
Unpaid TikTok creators filmed trips to buy the mochi ice cream. Retail data showed Tesco sales up 700% and Waitrose up 350% at the peak.
Grocery Gazette, Feb 2021
Brand lift and holdout tests, in plain words
Both work the same way. Split people (or cities) into two groups, show the creators to one and not the other, then compare. Meta andYouTube run brand lift studies by polling both groups on ad recall, awareness and consideration. A holdout or geo test compares sign-ups or sales instead.
One catch from Meta's own help pages: don't run a brand lift study while other campaigns for the same brand are live, because they reach the holdout group too and blur the result.
Experiments with creator content
- UK and Italy · 2026LG Electronics TV launch
LG split its campaign in two: brand assets only, against brand plus creator assets run as partnership ads, mostly on Shorts. Search lift was 202% in the UK and 245% in Italy, 3.6x higher than brand assets alone.
Google, APAC YouTube 2026 case studies - Thailand · 2026Dove with creators Brinkkty and Soundtiss
A two-arm brand lift test compared creator Shorts on their own with Shorts plus long-form. Shorts alone lifted ad recall by 14.02%; the mixed arm by 4.71%.
Google, APAC YouTube 2026 case studies - US · 2025Kantar, 40 YouTube campaigns
Kantar compared exposed and control groups across 40 campaigns in 11 sectors. Creator plus brand ads beat brand ads alone by 9% on aided awareness and 15% on ad recall among 18 to 49 year olds.
Kantar, commissioned by Google
The measurement standards nobody follows
Influencer measurement does have rulebooks. Most campaigns just don't use them, partly because each platform counts things its own way.
Four documents worth knowing
ANA influencer measurement guidelines
- Sort metrics by goal: awareness, engagement, conversion.
- Platforms calculate engagement rate and video views differently, so say which definition you used.
- Push platforms for more data, especially on organic posts.
IAB creator measurement landscape
- Every platform defines a view, an impression and engagement its own way.
- Calls for common definitions and independent checks.
- Wants invalid-traffic rules applied to creator content.
- Move past last click, towards incrementality.
ISBA creator economy framework
- Three ways to run creator marketing: bolt-on, channel or core.
- Only 9% of brands it surveyed treat it as core; 48% run it ad hoc.
- 59% could assess creator ROI only somewhat confidently.
Barcelona Principles 4.0
- Advertising value equivalents are invalid and shouldn't be used.
- Measure outcomes and impact, not the cost of equivalent ad space.
- EMV is the influencer version of the same idea.
How to spot fake numbers
Bought followers and bought views make every other metric meaningless, so check before you book, not after. In 2018 Unilever pledged never to work with influencers who buy followers, and asked platforms for more transparency.
Tick any you've noticed on a creator's profile
Still ask for insight screens. They're the cheapest check there is.
When the numbers lied, or got lost
- India · Aug 2020Badshah, "Pagal Hai"
During a wider probe into a racket selling fake followers and likes, Mumbai Police said the rapper had admitted buying 7.2 crore views to chase a 24-hour YouTube record. He denied it, and YouTube didn't recognise the record.
The Quint, Aug 2020 - US · May 2019Arii's clothing line
An influencer with 2.6 million Instagram followers couldn't sell the 36 shirts her manufacturer needed to start production, and the line was cancelled.
BuzzFeed News, May 2019 - Global · Dec 2024PayPal Honey and creator links
A YouTuber alleged the Honey browser extension swapped creators' affiliate links for its own at checkout, taking credit for sales those creators drove. PayPal said Honey follows industry rules, including last-click attribution.
Fortune, Dec 2024
What a campaign report should include
A good report answers two questions: did it work, and who do we book again? Tick what your last report had.
Each gap is a question your team will ask after the campaign, when the answer is hardest to find.
Myths about measuring influencer marketing
Tap a card to see the answer. Two of them aren't a clean yes or no.
From our roster
Every view figure on our case-study pages comes from the creators' own insight screens. Where we couldn't check a screen, the page shows creators and videos and leaves the views out. Switch to confirmed views below to see which campaigns have one.
Four with confirmed views
Replit82 videos · 20M+ views
Higgsfield151 videos · 1.25M+ views
Pollo AI19 videos · 802K+ views
Syntx AI10 creators · 405K+ viewsWords you'll hear in this business
Six terms you will meet in this guide, in plain words. Tap a card for an example. The full glossary has every term, from KOL to scope creep.
Questions people ask
Pick the goal first, choose two or three numbers that prove it, collect the creators' insight screens and your own data, and compare them with a baseline from before the campaign.
It depends on the goal. For awareness: viewers and views. For consideration: saves, shares, profile visits and link taps. For sales: code redemptions, tagged-link sign-ups and purchases.
Add likes, comments, saves and shares, then divide by views or by viewers. Many tools divide by followers instead, which gives a different number, so always say which you used.
Match sales or sign-ups to each creator's code or link for the direct part, then use a brand lift, search lift or holdout test to see what the creators added beyond the clicks.
There's no single good number. Compare a creator's post with their own last ten, and compare creators in the same category and size.
Earned media value: reach priced as if it were bought ad space. It's easy to report but doesn't measure sales or lift, and measurement bodies advise against it.
In November 2024 Meta replaced Plays and Impressions with Views. Accounts reached now appears as Viewers. Interactions add up likes, comments, saves and shares.
A test where the platform splits people into a group that sees the ads and a group that doesn't, then polls both on ad recall, awareness and consideration.
Look for sudden follower jumps, comments that fit any post, engagement that doesn't match views, and an audience in the wrong countries. Ask for the platform's own insight screens.
Collect insight screens within a week of posting. Check your own searches, visits and sales for two to four weeks after, against the weeks before.
Yes. Views count every play, repeats included. Viewers counts unique accounts.
From each creator's own insight screen. Where we couldn't check a screen, the case study shows creators and videos without a view figure.
Where to go next
What we do