measurement · attribution
How Do You Prove a Creator Campaign Worked When You Sell in Stores
In our record of 750,313 tracked sponsor deals, three powder brands carry 8,539 of them and seven canned and bottled brands carry 173. The gap is not about who has the better product, it is about which one ships with a discount code.
Key takeaways
- Three direct to consumer powder brands carry 8,539 tracked creator deals in our record. Seven canned and bottled brands carry 173.
- Two of those seven sold for a combined 3.75 billion dollars, Poppi at 1.95 billion and Alani Nu at 1.8 billion.
- A discount code measures the click, and in beverage most of the purchase happens later, in a store, with no click involved.
- The only methods that answer causation are a geo holdout and a matched market test, and both need a baseline set before launch.
Three powder brands carry 8,539 tracked creator deals in our record. Seven canned and bottled brands carry 173. Two of those seven sold last year for a combined 3.75 billion dollars.
That gap is not a story about which products people want. It is a story about which products can be measured.
A powder or a subscription ships direct with a code attached, so the finance team sees the return in the same week the video goes up. A can sold in a store produces no click, no code, and no customer record, so the identical spend looks like it did nothing at all. Budget follows the measurable thing, and creator budget has spent a decade following the code.
This is the practical version of what to do instead, if the thing you sell sits on a shelf.
Why is this so hard in the first place?
A. Because the code was never measuring the campaign, it was measuring the click.
A discount code answers one narrow question, which is how many people went straight from the video to your checkout. In a category people buy in a store, that is the small end of the response. Someone hears about a drink on a podcast on Tuesday, sees the can in a cooler on Saturday, and buys it because they recognise it. No code, no link, no way for the code to know.
So the honest problem is not that retail cannot be measured. It is that retail cannot be measured one person at a time, and every tool built for creator marketing was built to measure one person at a time.
What our own record shows
A. Deal volume tracks the discount code almost perfectly.
Our sponsor record holds 750,313 tracked deals across 46,440 channels. Sorting a set of beverage brands by how they reach the buyer produces a split that is hard to look away from.
| Brand | How it reaches the buyer | Tracked creator deals |
|---|---|---|
| AG1 and Athletic Greens | Powder, direct with a code | 3,089 |
| Gamer Supps | Powder, direct with a code | 2,846 |
| LMNT | Stick packs, direct with a code | 2,604 |
| Liquid Death | Cans, retail | 53 |
| Olipop | Cans, retail | 38 |
| BodyArmor | Bottles, retail | 29 |
| Celsius | Cans, retail | 21 |
| Alani Nu | Cans, retail | 16 |
| Gatorade | Bottles, retail | 12 |
| Poppi | Cans, retail | 4 |
Poppi has four tracked creator deals in a record of three quarters of a million. PepsiCo bought Poppi for 1.95 billion dollars, and Celsius bought Alani Nu for 1.8 billion, both in 2025. The money in this category is in the can, and the creator activity is almost all in the powder.
Source, PepsiCo's own announcement of the poppi acquisition, 17 March 2025 and reporting on the Alani Nu purchase. Deal counts are ours.
The five ways to measure it
A. Two of them answer the causation question and three of them are useful signals. Know which one you are buying.
1. Retailer specific codes and landing pages. Give each creator a link to a page for one retailer, or a code that only works there. Cheap, fast, and it starts working the day you set it up. What it cannot tell you is anything about the person who bought a week later in a store, which in this category is most of them. Treat it as a floor.
2. A geo holdout test. Split your markets in two, run creators in half, keep the other half quiet, and read scan data for both over the same weeks. This is the closest thing to a causal answer retail allows, because the held out markets tell you what would have happened anyway. It needs enough stores to be readable, a matched pair of market groups, and eight to twelve weeks of patience.
3. Retail media tie in. Run the creator's asset as paid creative inside the retailer's own advertising platform, so the retailer's closed loop reporting attributes the sales it can see. This gets you a number the retailer will stand behind. Two limits worth naming out loud, it only counts purchases at that retailer inside its own attribution window, and it measures the paid placement rather than the organic post the creator made.
4. Panel and scan data against a baseline. Buy the category read from a syndicated provider, take a pre campaign baseline, and watch your share of the category through the flight and after it. Good for context and for spotting a change big enough to see. It will not separate your campaign from your promotion, your new distribution, or your competitor's price cut.
5. A matched market in store test with the retailer. The retailer picks comparable stores, you run in some and not others, and everyone reads the same register data. Slowest and most political of the five, and the most convincing when it lands.
None of these is exotic. The reason they get skipped is that four of the five have to be arranged before the campaign, and the campaign is usually already booked by the time anyone asks how it will be measured. That is the single most common way a beverage brand ends a quarter with content it liked and no idea whether it sold anything. We set the measurement up as part of the deal rather than after it, which is a scheduling problem more than a technical one.
What to do before the campaign starts
A. Four things, and they take about a week.
- Write down the baseline. Units for the last thirteen weeks, by retailer where you can get it. A number you agree on before launch is worth more than a cleverer number argued about afterwards.
- Decide what a win is, in units. Not impressions, not engagement rate, not earned media value. If the campaign has to sell an extra number of cases to be worth repeating, say the number out loud first.
- Pick the holdout. Even a rough one. Two states you leave alone is better than nothing.
- Ask the retailer what they can report. The answer varies more than people expect, and it decides which of the five methods above is even available to you.
What these numbers do not show
A. More than I would like, and the direction is safer than the size.
- Our record is built from YouTube. It comes from what creators write in their video descriptions, so Instagram and TikTok are almost entirely missing. Canned brands skew to those platforms, so the 173 is certainly an undercount and the true gap is smaller than the table suggests.
- Brand names are messy. AG1 appears under two names in the data and LMNT under two, which is why those rows are summed. A brand that mostly gets mentioned by a nickname will be underrepresented.
- A tracked deal is not a paid deal. The record catches sponsorship language in a description, so it will include some organic mentions and miss some paid ones.
- None of this is a claim about sales. It is a count of creator activity. A brand with four tracked deals may be spending heavily somewhere this record cannot see.
What survives all of those caveats is the shape. Creator marketing grew up measuring clicks, the beverage money is in a format that produces no clicks, and the brands that solve that gap first get to spend against a number their finance team believes.
Where We Come In
The measurement question is really a sequencing question. Almost every method above is cheap if it is set up before the campaign and impossible if it is set up after, and the reason it gets set up after is that nobody owned it while the deal was being negotiated.
That is the part we take on. We build the read into the deal, meaning the baseline agreed before launch, a holdout where the market split allows one, and a total cost that includes the agency time and the content rights rather than the creator fee on its own. If you are earlier than that and still choosing between a code based program and a paid one, our comparison of influencer marketing against affiliate covers that fork.
If you already have a campaign in market and no way to read it, tell us what you sell and where, and we will tell you which of the five is actually open to you. Speak with us before the next flight is booked.
Reading loop
- Start here: How influencers make money, and what that means for your deal
- Same category, the vetting half: How to vet functional beverage creators
- Risk: The FTC disclosure playbook for brands in 2026
Frequently asked
How do you track influencer marketing without a discount code?
You measure markets rather than people. Split the target markets, run creators in half of them, hold the other half quiet, and read retail scan data in both over the same weeks. The difference between the two groups is the closest thing to a causal number that retail allows, and it needs a baseline taken before launch.
Why do creator deals concentrate on powders and subscriptions?
Because those products ship direct with a code attached, so the finance team can see the return the same week. A can bought in a store produces no click, no code and no customer record, so the same spend looks like it did nothing. Deal volume follows what is measurable, not what sells.
Can a retailer tell me if my creator campaign sold anything?
Partly. If you run the creator asset as paid creative inside the retailer's own ad platform, the retailer's closed loop reporting will attribute sales it can see. That covers purchases at that retailer inside its attribution window and it measures the paid placement rather than the organic post, so it is a floor on the answer, not the whole answer.
How long does a retail read on a creator campaign take?
Plan on a quarter. Scan data arrives weekly and moves slowly, shelf distribution changes underneath you, and a two week campaign inside a season of promotions is very hard to isolate. A short flight with no holdout will not produce an answer anyone should act on.
What is the cheapest honest measurement for a small brand in retail?
A retailer specific landing page plus a post purchase question at checkout on your own site, read against a pre campaign baseline. It will not prove causation, but it is directional, it costs almost nothing, and it beats quoting impressions as if they were sales.