Ask a room of restaurant marketers what's wrong with third-party delivery and you'll hear a version of the same sentence: you get the order, they keep the customer.
That complaint is accurate, and as of a month ago it's also settled. New York City passed a law in 2021 requiring the delivery platforms to hand restaurants the names, phone numbers, email addresses, delivery addresses and order contents of the guests ordering from them. DoorDash, Grubhub and Uber Eats sued, and on August 5 the Second Circuit affirmed that the law is unconstitutional. Whatever you're going to know about your guests, you're going to have to earn.
So the industry has largely concluded that its guest-data problem is a third-party problem with a first-party solution: push more volume into your own app, own more of the relationship, and the picture fills in. That strategy isn’t wrong, it just aims at the smaller half of the problem.
The bigger leak is inside your four walls
Chipotle is a useful example here. Their digital presence was 38.3% of sales last quarter, the Rewards program has 23 million active members, and the app is genuinely good. But on the same earnings call, Scott Boatwright described what happens when those guests walk into a restaurant instead: "only about 20% of the transactions scan for Rewards compared to nearly 90% of owned digital transactions."
Four out of five guests standing at the counter of one of the most digitally sophisticated QSRs in the country complete their transaction without the brand knowing who they were. Not because a marketplace intercepted them, and not because anyone withheld anything. They just ordered lunch and left. The identity you lose at your own counter, on your own property, in front of your own staff, is a larger number than the identity any delivery platform is keeping from you.
Starbucks runs the most mature loyalty operation in American retail food, and its own investor dashboard puts Rewards member spend at 59% of tender dollars in US company-operated stores, with mobile order accounting for 33% of transactions.
Put differently: two-thirds of the transactions at the best-instrumented chain in the business still arrive without an app opening. If you’re assuming that identification approaches completeness once the app matures, the two brands furthest along that path are telling you otherwise.
Why your loyalty numbers flatter you
When only a fifth of your in-store transactions are identified you are, by default, tracking only your most frequent, most engaged, most already-loyal guests, because those are the people who bother to scan.
So when the quarterly deck compares loyalty members to non-members and reports the gap as what the program delivered, most of that gap is the sorting, not the program. You're measuring who signs up, not what signing up did to them. Twenty-three million enrolled members and a 20% scan rate are two different facts, and only one of them tells you what you can act on.
One number, and almost nobody computes it
The one you can act on is the scan rate. Take your transactions for a normal week, split them by channel and by daypart, and calculate what share of each one arrives attached to a known guest — app, web, kiosk, front counter, drive-thru lane, and every marketplace separately.
That's one query against data you already own, and it hands you an identification rate for every way a guest can reach your brand.
What comes back is usually uncomfortable in a specific way: identification is high exactly where volume is low, and low exactly where volume is high. That's the map that tells you what your personalization program can currently reach, which promotions you're able to measure at the guest level and which you aren't, and where a single operational change — how the loyalty prompt appears at the terminal, whether the drive-thru has any path to identification at all — would return more recognized guests than another quarter of app-download spend.
What it takes to close the gap
None of that requires resolving identity perfectly. It requires knowing, per channel, how blind you currently are.
The harder work comes after: recognizing the same guest across a counter transaction, an app order and a marketplace order without depending on them to remember to scan, and doing it in a franchise system where the systems producing those records were never designed to agree with each other.
That's real engineering and it isn't free. But it starts from a number you can produce this week, and every brand I've seen make progress on this started by admitting how much of their business was arriving anonymous.
That's the work OptiGraph℠ does — resolving and deduplicating the records a brand already generates across its channels into activation-ready profiles tied to verified identities, so that the app account, the loyalty ID and the marketplace order stop being three separate strangers. The part that matters most in a business where most transactions happen at a counter is the offline side of it: location and card-based attribution that can confirm a physical visit without waiting for the guest to remember to scan.
OptiReveal℠ then works in the other direction, overlaying that first-party base with observed response patterns like purchase intent and location behavior to surface the segments actually worth spending against.
None of it changes the fact that a guest can walk in, order lunch and leave without ever telling you who she was. What it changes is how many of those visits you can connect to a person anyway.
The question in the title of this post isn’t rhetorical. Most brands know their customers’ orders extremely well. Order counts, mix, ticket, daypart, channel — all of it clean, all of it reportable. The guest behind the order is a different dataset, and for the majority of transactions in this industry, it doesn't exist yet.
Knowing which majority, and where, is the first thing worth doing about it.
AiOpti will be at the QSR Evolution Conference in Atlanta, September 8–10. If you're working through what your brand can actually see about its guests, come find us.