Ask a retail-execution vendor for a coverage map and you’ll usually get a route map. Nobody is being cagey about it. In that model the two things are the same object.
Every platform in the field-execution category, and every merchandising agency sitting behind one, sells the same underlying thing: a person equipped to enter a store and record what is on the shelf. The software is the part you buy. The walking is the part that decides what you see. So when a brand evaluates one of these products as a data purchase, it is quietly buying a labor schedule with a reporting layer bolted on top, and the coverage it gets is a consequence of that schedule rather than of the technology.
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. Our model is the other one, so read the rest of this with that in mind. We’re also going to spend a full section on what field capture does better than we do, because a version of this argument that skips that part isn’t worth reading.
One boundary first. This post is about who collects retail data and what that does to your coverage. If what you need is the mechanics of running an execution program, meaning what to specify in writing, how to verify it, and what federal trade-practice rules let you pay for, that is a separate piece: retail execution tracking.
| Collection model | Who does the walking | What coverage equals |
|---|---|---|
| Your own field team | Your sales or merch reps | Your route list |
| Third-party merchandiser labor | An agency’s field workforce | The stores you paid to visit |
| Retailer staff or fixed shelf cameras | The retailer | Banners that deployed it |
| Remote storefront observation | Nobody in the aisle | What is observable from outside |
The store list is your account list, not a market
Field capture begins with an upload. Somebody puts accounts into the app, assigns them to reps, and the reps go. Every store that produces a row of data is a store that was already on that list, and the list came from your distributor’s authorized-account file or your own CRM.
Which means the coverage question is settled before any data gets collected. You see the accounts you are in. You do not see the accounts you are not in, and no configuration change fixes that, because nobody was routed to a store where you have no business to check on.
That bites hardest on the thing brands actually want, which is finding where they should be selling and are not. A store that never started carrying you produces no visit, no photo, no shelf check, no row. It does not show up as a zero. It does not show up. Sizing that population is a separate exercise entirely, and we walked through the reconciliation for it in void and out-of-stock reporting.
Competitive reading inherits the same shape. A rep standing in your account can photograph the whole shelf, competitors included, and that is genuinely useful intelligence. But it is a shelf in a store where you already have distribution. Ask what a competitor is doing across the several thousand accounts that carry them and not you, and this model has nothing at all to say, because there was never a reason to send anyone.
Cadence is a labor schedule, not a refresh rate
Second consequence. How current your data is has almost nothing to do with the vendor’s architecture and almost everything to do with a headcount decision somebody made last budget cycle.
A field visit costs two things you can price: paid time and vehicle miles. Both scale close to linearly with stores times visits. Double the frequency and you roughly double the bill, which is the opposite of how software cost behaves once it’s built. Get that straight before you negotiate anything.
Work it through. All of these inputs are assumptions and you should replace every one with your own: a fully loaded field cost of $38 an hour, 25 minutes inside the store, 35 minutes of driving between accounts, and 22 miles between accounts. For the vehicle side, the IRS business standard mileage rate is 76 cents per mile for travel on or after 1 July 2026 (IRS standard mileage rates), and GSA sets its privately owned automobile reimbursement to the same 76 cents from the same date (GSA POV mileage rates). Those are reimbursement standards rather than a measured cost of ownership, so treat 76 cents as a defensible stand-in, not a truth.
One visit is then an hour of paid time at $38 plus about $17 of vehicle cost. Call it $55.
Now scale it. Six hundred accounts visited monthly is 7,200 visits a year, so about $396,000 of field cost before anyone has paid for software. Visit the same 600 weekly and it is 31,200 visits, or roughly $1.7 million. Nothing about the app changed between those two numbers. The only variable was headcount.
How a route actually gets built
Tiering is the correct call, by the way. If somebody handed you 600 accounts and 14 reps, you would sort by value too. The problem is not that the sort is wrong. The problem is that your data freshness now varies by account in exactly the same pattern as your existing revenue, so the accounts you know least about are the accounts you already sell least in.
Gaps land where the route is expensive
Third consequence, and the one that never makes it into a deck. Skipped visits are not randomly distributed.
Give a rep 11 stops and nine hours and something gets dropped. What gets dropped is predictable: the single-store account 40 minutes off the highway, the one doing four cases a month, the one where the owner wants to relitigate last spring’s pricing every time, the one with nowhere to park a hand truck. Nothing about that is negligence. It is a person triaging a list that was longer than the day.
Then the bias compounds in a way that catches people out. Those same small independents are the stores that barely register in syndicated scan panels, for completely unrelated reasons involving licensing agreements and POS export capability rather than drive time. We covered that mechanism in where independent-store coverage goes blind. Two data sources, two different causes, one shared hole. A brand running both and still seeing nothing in the independent trade tends to conclude the trade is small. What it has actually done is measure the same absence twice.
Nobody publishes rep visit-compliance rates broken out by account tier. We went looking and did not find a figure we would stand behind. Vendors hold that data and have no commercial reason to release it, and the brands that hold it are not publishing either. So compute your own. Pull your target account list, pull the visit records against it, count the accounts with zero completed visits in the last 90 days, then sort that list by cases shipped. If the zeros cluster in your smallest accounts, you have measured your own coverage bias in an afternoon, for free, and you now know something no vendor was going to tell you.
What a rep in the aisle does that observation cannot
Most vendor-side writing on this topic skips the next bit. Skipping it is exactly why that writing never persuades anyone who has actually run a field team.
Only a person in the store can do these
Take 03 seriously, because it is the strongest argument in the category. Observation tells you a floor stack came down in week two of a four-week window. A rep who was there in week two rebuilds it. That is not a difference in data quality. It is the difference between knowing and fixing, and a brand losing three weeks of display exposure does not care which vendor had better resolution.
Then 05, dismissed as bureaucracy by people who were never measured on it. A national account manager is graded on coverage. Visits completed, accounts touched, tasks closed, exceptions cleared. GPS-verified visit records are the artifact that whole conversation runs on, internally and with the distributor, and they exist only because somebody’s phone was physically in the building. No observation product generates them, and not because of a feature gap. Observation doesn’t involve a person whose accountability is being tracked. There’s nobody to verify.
Planogram sign-off is similar. Facings, eye-level position, shelf-strip placement and whether the wrong SKU crept into your space are close-range judgments. Some of them need a tape measure. That is a person’s job and it is going to stay a person’s job.
The trade, stated plainly
So the honest axis here is coverage economics, not a ranking.
Two collection models, honest comparison
Field capture buys depth and the power to act, inside a store list you wrote. Remote observation buys breadth and repetition across a footprint nobody selected, and it cannot lay a hand on the shelf. Those are answers to two different questions. Plenty of brands ask both at once without noticing they are two.
Where we’d draw the line: put field labor where the money is and where somebody has to physically do something. Use observation for the accounts nobody is going to visit this quarter, which is where the unpleasant surprises live, and as an audit on the accounts that supposedly are being visited. Our Marketing Monitoring module is the second job, not the first. We have told plenty of prospects they need a rep in Ohio more than they need a subscription from us, and meant it.
What we’d recommend
Do the arithmetic before you take a single meeting, because it settles most of the argument on its own.
Count your target accounts. Count how many of them have a completed visit in the last 90 days. Divide the field program’s cost by the visits that actually happened, not the visits that were scheduled, and you will have a real cost per observation for the first time. Then look at the accounts with zero visits and add up the cases they ship. That is the population you are flying blind on, priced in your own units.
If the answer is that your visited accounts are well covered and the unvisited ones ship almost nothing, stop. Buy nothing. Spend the money on trade programs and use the field team you have.
Two more disqualifiers, so nobody wastes budget. If your business lives in 40 chain accounts and you already receive direct data feeds from those retailers, you do not need either kind of product. You need somebody to actually read the feed you are paying for. And if you have no field team at all and fewer than a couple of hundred accounts, hiring one good rep beats buying software of any kind, because at that scale a person with a phone is cheaper than a subscription and better than either model on its own.
Past those lines, the interesting decision is which gap is bigger: the accounts nobody visits, or the accounts nobody can act on. Grading distributor performance on what you can observe rather than what gets reported is the same discipline we described in distributor scorecards, and the same evidence problem shows up there.
Tell us your target account count, how many got a completed visit last quarter, and your cost per visit. We’ll give you a straight answer on which of the two gaps you have and whether we are the right fix for it. We do this for beverage-alcohol brands constantly, and the genuinely useful half of that conversation is usually the half where we tell you what not to buy.