Retail Execution Tracking: Verify Displays You Paid For

You can prove you paid. That’s the whole problem.

The invoice for a Q4 floor-stack programme is a document. It has a number on it, a date, an account count, a signature. The end cap, the case stack with the shelf talker taped to it, the four weeks of feature price: those are events that happened, or didn’t happen, in stores you have never walked into and have no contractual relationship with. What comes back to you is a report written by whoever was responsible for making it happen. Sometimes a photo. Often a spreadsheet with a compliance percentage and no photos at all. Occasionally nothing, and you find out in January when the numbers don’t move.

That asymmetry is the subject of this post. Not “trade spend is wasted,” which people say with a made-up percentage attached. We went looking for a credible published figure on display non-compliance in bev-alc and couldn’t find one we’d stand behind, so we won’t hand you one. The number that matters is your own unverified spend.

We’re gmware, a software and data engineering firm in Austin, TX, with delivery centers in Bangalore and Mohali, India. We build and run Shield Suite, our retail-intelligence platform for beverage-alcohol brands across 60,000+ storefronts, and Marketing Monitoring is the module that exists because of this exact gap. So we have a horse in this race. We’ll be clear about where our own approach stops working.

First, understand who you’re actually paying

This part is dry. It also explains why verification in bev-alc is structurally harder than in regular CPG.

Under the Federal Alcohol Administration Act, a supplier or wholesaler generally cannot induce a retailer by furnishing things of value or by “paying or crediting the retailer for any advertising, display, or distribution service” (27 U.S.C. 205(b)). TTB’s regulations are specific about what that covers. Reimbursing a retailer for setting up a display is paying for a display service (27 CFR 6.55). Renting display space at a retail establishment is also paying for a display service (27 CFR 6.56), and separately counts as acquiring an interest in the retailer’s property (27 CFR 6.35). The general rule adds that it doesn’t matter whether the service you got back was worth what you paid, and it explicitly reaches reimbursements for services the retailer bought from a third party (27 CFR 6.51).

There are exceptions, and they’re narrow. Product displays, meaning racks, bins, casks and shelving whose main job is holding product, may be given or sold to a retailer, capped at $300 per brand at any one time in any one establishment, carrying advertising matter permanently inscribed or securely affixed, with no pooling of dollar limits to beat the cap (27 CFR 6.83). The only condition you may attach is buying enough product to complete the display initially. Point-of-sale advertising materials get their own exception, same advertising-matter requirement, plus a flat ban on paying or crediting the retailer for using or distributing them (27 CFR 6.84).

Two more things worth knowing. A federal tied-house case also requires exclusion: the practice has to put retailer independence at risk and result in the retailer buying less of a competitor’s product (27 CFR 6.151). And for malt beverages specifically, the federal trade practice provisions only bite when the retailer’s state has a similar law, which TTB reads broadly and does not maintain a list of (TTB trade practice FAQs). Slotting fees are live policy territory rather than settled ground: TTB has said its regulations don’t expressly define them and asked the industry for comment on whether they should, alongside questions about category management and who provides shelf plans and schematics.

None of that is legal advice and we’re not lawyers. Check the current regulatory text, check your state, and route the specifics through counsel. TTB publishes its administrative cases, which is a bracing read if you’ve ever wondered whether any of this gets enforced.

Here’s why it matters for verification. Because you mostly can’t pay the store, your display money moves as distributor and broker programme funding: rep labour, POS production, promotional support, activation budgets. You are paying an intermediary to cause something to happen in a third party’s building. Nobody in that chain has a contractual duty to send you evidence, and the one with the best view of the shelf is the one being graded.

”Compliance” is too coarse to act on

Most execution reporting collapses into a single percentage. That percentage is useless for fixing anything, because six genuinely different failures roll up into it and each one has a different owner and a different remedy.

Work through them and the rollup starts looking useless. Existence failures usually mean the cases never got ordered or the rep never got there, which is a distributor conversation. Location failures mean the store built something, just not where you paid for, which is often a deal the chain never pushed down to store level. Window failures are the sneakiest: the display goes up late or comes down in week two when the store needs the floor, so a snapshot audit on day three records a clean pass on a programme that delivered half its exposure. Price failures are a different animal, and they’re why execution tracking and shelf price enforcement end up on the same dashboard. Assortment failures are usually a rep grabbing whatever was in the back room. Material failures almost never get reported, because nobody files a photo of their own torn shelf talker.

Decide which of the six you actually care about before you buy anything.

Why self-reported compliance runs optimistic

This is not an accusation of fraud. It’s arithmetic about who holds the camera.

Start with the incentive. The rep, broker or distributor merchandiser reporting execution is frequently the person accountable for delivering it. Even with total honesty on every individual report, the reporting is a judgement call at the margins: is a case stack two feet off the agreed spot compliant, is a display built on Wednesday for a Monday start compliant. Marginal calls made by an interested party drift in one direction.

Then the sampling. Field visits follow routes, and routes follow value: the bigger accounts, the closer ones, the ones with a relationship. So the stores you get evidence from are systematically the stores most likely to have executed. That’s convenience sampling. It doesn’t estimate your national compliance rate, it estimates compliance among stores good enough to be visited.

Then the photo problem. When a programme needs proof, the natural instinct is to submit the best available proof. One clean end cap, well lit, shot from the flattering angle, becomes the campaign’s evidence. It’s a real photo of a real display, and it tells you nothing about the other 399 accounts.

And the quiet one, which we see constantly in the data: missing records are not random. A store where nothing happened tends to produce no report at all, rather than a report saying “nothing happened.” Then the compliance percentage gets computed over stores that reported, and the denominator silently drops the failures. If your execution number is calculated on responses rather than on the full target list, it isn’t a compliance rate. Fix the denominator before you do anything else. It’s free and it usually moves the number more than any tooling will.

Four ways to find out what actually happened

Distributor and broker self-report is the default, and it’s free in the sense that you’re already paying for it. Keep it. Just stop calling it measurement. It’s a claim you test on a sample.

Field-team photo capture with image recognition is the mature commercial answer. Reps shoot store photos in an app and a model reads them, so compliance data falls out of a visit that was happening anyway. FORM, the company behind GoSpotCheck, shipped point-of-sale-material recognition that classifies displays and pulls campaign and price data from field images, and called out beer, wine and spirits as a category where big seasonal activations across thousands of accounts make the problem acute (FORM announcement, 27 July 2026, product write-up). Earlier in 2026, Gemspring combined FORM with the image-recognition business of Trax, which consolidates two of the bigger shelf-vision assets under one roof (Gemspring, 5 February 2026, trade coverage). Repsly plays the same game from the field-team-management side, pushing photo and survey tasks to reps and rolling results into promotion dashboards. Good technology, all of it. All of it inherits the route: an unvisited store stays invisible no matter how good the model is.

Retailer-side compliance reporting exists where a chain runs its own execution programme and shares the output, sometimes through a vendor portal, sometimes inside a retail-media deal. Where it exists it’s authoritative, since the retailer’s own staff built the thing. It’s also chain-defined, inconsistent between banners, and absent across most of the independent trade. So it covers the part of your footprint you already understand and skips the part you don’t, which is the same coverage cliff we ran into with competitor shelf tracking.

Paid third-party audits are the classic answer and still a reasonable one: commission an independent firm to walk a defined sample against a defined checklist. Clean, disinterested data at a known cost per store, with a small n and a long turnaround. Good for settling a dispute, poor for catching a window failure in week two.

Independent storefront observation at scale is the category our own Marketing Monitoring module sits in: execution observed across a wide footprint on a repeating cadence, store by store rather than as a regional average, so the accounts nobody was routed to stop being invisible. Its honest limit is that it’s outside-in. It’s not a rep in the aisle with a tape measure, so where compliance turns on something only visible at close range, buy a field visit or an audit for that check. What it gives you is coverage and repetition, which is exactly what routed visits can’t.

Turning a verified miss into money

Verification only pays if the programme was built to be verifiable before it ran. You can’t retrofit proof onto something agreed in an email as “good display support in the top 400 accounts,” and that’s where most brands lose the argument.

Now the arithmetic, inputs labelled as assumptions because we have no idea what yours are. Assume a Q4 floor-stack programme across 400 named accounts. Assume $150 per account in display support, POS production and billed rep labour carried through your distributor, so $60,000 of programme cost. Assume verification finds the stack present during the agreed window in 240 accounts. Then $24,000 bought no floor presence, before you count the volume it was meant to move. Run it with your own per-account cost. Until you can fill in the 240, the rest of the calculation doesn’t exist.

Then have the conversation with dates rather than adjectives. Roughly:

“We funded a floor stack in these 400 accounts, from 3 November through 1 December. We have store-level observation showing no stack present during that window in these 160. Here are the accounts and the dates. Which of them do you dispute, and what evidence do you have?”

That framing works because it’s falsifiable and because it hands them the chance to be right. Some of those 160 will come back with a legitimate explanation, and you want those: a verification programme that can’t be corrected won’t survive its first real fight. For the accounts nobody disputes, ask for a credit against the programme or a make-good in Q1 at the same account count, evidence standard already agreed. Then reallocate. Next quarter’s display money goes to the accounts that executed, and the rest earn their way back by executing once at their own cost. Same logic as a distributor scorecard: grade what you can observe, then move money toward it.

What verification still won’t tell you

It won’t tell you the display worked.

This is where execution reporting quietly oversells itself, so let’s be blunt. Proving a case stack existed for 28 days proves the case stack existed for 28 days. It doesn’t establish that volume in those stores was caused by the stack, because display stores are not a random draw. They’re your bigger, better-run, higher-velocity accounts, picked by people who expected them to sell. Compare display stores to non-display stores and you’ll measure your own account-selection skill about as much as the display.

A defensible lift number needs a comparison built in from the start: matched accounts that didn’t get the display but resemble the ones that did on velocity, banner and market, or the same accounts before, during and after with seasonality accounted for. More work than a compliance report, and the only version that survives contact with a finance team.

So hold two ideas at once. Verification is necessary, because without it every trade-spend argument comes down to whose anecdote is louder. It isn’t sufficient. We sell the observation layer, and we’re not going to pretend it’s a causal model.

What we’d recommend

Do the free thing first. Take your last completed display programme, get the original named target list, and calculate compliance over that full list rather than over the stores that reported. If the number drops a lot, you now know your reporting was measuring response rate, and you’ve learned something worth more than most software purchases.

Then decide which of the six observables you’re buying. If the gap is “our reps can’t cover the footprint,” a field-execution app with image recognition is the right spend and there are good ones. If the gap is “we have no idea what happens in the accounts nobody visits,” no rep tool fixes it, because the tool rides the route. That’s the case for independent storefront-level observation, and it’s what Marketing Monitoring does: verify execution against the agreed programme store by store, with on-the-floor evidence, so gap targeting names accounts instead of regions.

And the honest disqualifier. Running display programmes in 50 accounts? Don’t buy any of this. Call your reps, ask for timestamped photos, keep a spreadsheet. Verification infrastructure starts paying when the account count is past what a person can hold in their head and the trade-spend line is big enough that a 30% miss is real money. Below that you’d be buying a telescope to read your own handwriting.

If you’re above that line, tell us what your last programme looked like: how many accounts, what you paid per account, what came back as evidence. We’ll give you a straight answer on what we could verify, what we couldn’t, and what it would cost. We do this for beverage-alcohol brands every day, and the first thing we’d tell you is whichever part of it you don’t need.