DOOH Attribution Without Device IDs: Privacy-Safe Ways to Measure In-Store Screen Ads

At a Glance

Online ads get credit through clicks and cookies. A screen above a convenience-store counter or on an ATM gets neither. The usual workaround is to follow phones: put a virtual fence around the screen and the store, and count devices that show up in both. That works at scale, but it depends on mobile-location data and still produces estimates, not purchases.

You don’t need device IDs to see whether in-store screens are working. The cleanest method compares stores that ran the campaign with matched stores that didn’t. On 6 October 2026, at the NACS Show, Love’s Media Group published a fresh example: a MarketDial test-and-control analysis that reported a 23% incremental unit lift for a beverage promotion (Love’s). Those are Love’s own figures, for a full multi-touchpoint campaign.

This guide covers four privacy-safe methods (holdout stores, sales lift, codes, and footfall counts), what each one can and can’t prove, and how a local advertiser can run a small version.

---

Who This Is For

Audience	Why it matters
Local brands and franchisees	You want to know whether a screen flight moved sales, without buying a location-data product.
Regional CPG and beverage brands	You may get a vendor lift report. This helps you read it critically.
Agencies planning place-based DOOH	You need to explain why a holdout test beats a dashboard of “attributed visits.”
Not for	Anyone looking for a guaranteed lift number. Results depend on product, offer, store, and timing.

Book named convenience-store and ATM venues on self-serve. For a short same-day play on one screen, bought by scanning that screen’s QR code, see Instant Ads.

---

Why Device-Based Attribution Became the Default

Programmatic DOOH grew up next to mobile advertising, so it borrowed mobile’s measurement. A useful, well-documented recent example is VIOOH’s white paper “The Proof is Physical,” written with Arla Foods (VIOOH).

How it worked, as VIOOH describes it: measurement specialists Spotzi and media owner JCDecaux ran Foot Traffic Attribution studies. They placed virtual boundaries around digital screens and target stores, tracked anonymous mobile signals from ad exposure to store visit within an attribution window of about five days, and compared that behavior with a pre-campaign control group.

VIOOH reports these results for Arla brands:

• Starbucks Tripleshot: 145,301 incremental store visits at a 5% lift (March–April launch); a 41% lift in a June–July follow-up.
• Castello: 438,999 incremental store visits (April); a Christmas activation that peaked at a 71% lift (November–December).
• Arla Protein: a 23% lift and 297,745 incremental store visits (March).

The trade site invidis adds context (invidis, 24 Sep 2026). These were nine campaigns in Norway during 2025. invidis also notes that the figures are attribution estimates, not counts of purchases, nor proof that a single exposure caused a visit. Seasonal demand and other advertising may have played a part. Arla ran its Apetina campaign alongside Snapchat, TikTok, and YouTube.

That is a fair summary of the method’s strengths and limits. It gives you scale and a time-to-visit view. But it depends on device signals, it measures visits rather than sales, and a pre-campaign baseline can’t fully separate the ad from the season. For a small advertiser buying a few screens, it is also usually out of reach.

Carrier data is now coming into DOOH too. Vistar Media announced on 5 October 2026 that T-Mobile audience segments built from consented carrier data are available in its DSP (Vistar). That is targeting: deciding which screens to buy. It does not tell you whether the campaign sold anything.

---

Method 1: Holdout Stores (Test vs Control)

The idea is simple. Pick stores to run the campaign (test). Pick comparable stores that don’t run it (control). Compare how sales of the advertised product change in each group. The difference is your incremental lift. Our matched-market testing guide walks through store matching in detail.

A 2026 example: Love’s Media Group and MarketDial

As stated in Love’s 6 October 2026 release (Love’s):

• Campaign: eight weeks, 26 February to 22 April 2026, for an unnamed “leading beverage brand,” promoting a three-for-$9 offer at participating Love’s locations.
• Media mix: exterior posters, in-store digital screens, pump screens, pump toppers, interior posters, restroom flyers, and counter mats.
• Measurement: a test-and-control methodology powered by MarketDial. Campaign stores were compared with matched holdout locations. The analysis included 51 control stores after four outliers were removed.
• Results: a 23% incremental unit lift at a 99% confidence level, and a 19.6% incremental sales lift.

What’s worth copying is the design, not the headline number:

1. Matched holdouts, not a “before” period alone. Control stores live through the same weather, prices, and calendar as test stores.
2. Outliers handled openly. The release says four were removed and states the count that remained. Ask any vendor report for the same.
3. A confidence level is reported. A lift without a confidence level, or without a store count, is hard to judge.
4. Units and sales are both reported. A multi-buy promotion can move units and revenue differently.

Caveats you should keep in mind. The figures are Love’s and its measurement partner’s. The lift applies to the whole mix, from pump toppers to counter mats, not to screens alone. It covers one brand, one promotion, and one retailer. None of it predicts what your campaign will do, and none of it is a Take 10 result.

Running a small holdout test yourself

You don’t need 51 control stores to learn something. You do need discipline:

• Choose test and control areas before you look at any results. Similar store types, similar traffic, similar neighborhoods.
• Collect a baseline. Several weeks of your own sales, bookings, or calls for both groups before launch.
• Change one thing. Don’t launch a new price or a radio spot in the test area at the same time.
• Run long enough to cover normal weekly swings. Then compare the change in test areas with the change in control areas.
• Treat small samples as directional. With a handful of stores, one busy weekend can swing the result. Repeat the test before you call it.

Self-serve booking helps here because you pick the listed venues and dates yourself. That is what lets you leave comparable areas unbooked as your control.

---

Method 2: Sales Lift From Your Own Data

Holdout tests need store-level data. If you are the business being advertised (a restaurant, a car wash, a clinic, a phone repair shop), you already have it:

• Point-of-sale totals for the advertised item or service.
• Transaction counts (number of tickets), which show traffic even if basket size moves.
• Bookings, calls, and form fills, by day and by location.

A simple on/off flight is the other option when you only have one location: run the screens for some weeks, pause for some weeks, and repeat. It’s weaker than a holdout comparison because seasons and local events change over time. Still, alternating on and off weeks controls for more than a single “before vs after” comparison does.

What this can’t do: it can’t tell you which customers saw the ad. That’s fine. You’re measuring whether the business moved, which is the question you care about.

---

Method 3: QR Codes and Screen-Only Promo Codes

Direct-response signals are the easiest to set up:

• A QR code on the screen creative, pointing to a landing page that only that campaign uses.
• A promo code or offer that appears only on the screens (“say SCREEN10 at the register”).
• A dedicated phone number or short URL.

These give you a clean count with no personal tracking beyond what the customer chooses to share. Our guides to QR codes in in-store advertising and how Instant Ads use on-screen QR codes cover creative and placement.

The limit is undercounting. Many people see a screen and buy later without scanning or saying a code. Treat redemptions as a floor, not the total. If a holdout test shows lift but codes show few redemptions, believe the holdout. The codes only measured the people who chose to respond directly.

A related signal: OOH TODAY, citing OAAA analysis of hundreds of campaigns, reports a median 14% lift in digital outcomes such as website visits and app downloads after OOH exposure (OOH TODAY, 10 Sep 2026). That article is a partner piece and does not describe the measurement method. Still, it’s a reminder to watch your branded search and direct web traffic in test areas, not just scans.

---

Method 4: Footfall Proxies That Don’t Need Phones

If your goal is visits rather than sales, there are aggregate ways to count them:

• Door counters or people-counting sensors that report totals, not identities.
• Transaction counts per hour or day (see Method 2).
• Staff-recorded “how did you hear about us?” answers, kept consistent across locations.
• Reservation or appointment volume by location.

Compare these across test and control areas, the same way you would compare sales. This is also how you get closer to the VIOOH-style question (“did more people come in?”) without following any device.

The trade-off: a counter tells you more people came in, not why. Pair it with a control group.

---

Choosing a Method: Quick Comparison

Method	What it measures	Device data needed?	Main limitation
Holdout stores (test vs control)	Incremental sales or units	No	Needs comparable stores and enough data
Sales lift / on-off flights	Change in your own sales or bookings	No	Seasonality and local events can blur results
QR / promo codes	Direct responses	No	Undercounts people who buy without scanning
Aggregate footfall counts	Visits	No	Counts visits, not purchases or causes
Mobile-location foot traffic attribution	Estimated incremental visits	Yes	Estimates, not purchases; depends on device signals

The strongest setup combines one causal method (a holdout comparison) with one or two direct signals (codes, branded search, transaction counts).

---

What Shoppers Remember: The OAAA / Harris Angle

Measurement shouldn’t push you toward only the cheapest metric to count. OAAA and The Harris Poll presented new research in a members-only webinar on 7 October 2026 about OOH across experiential and retail. As DailyDOOH summarized it, the Harris Poll study found real-world brand experiences strongly influence consumer behavior, “with 80% more likely to consider a brand after experiencing it in person” (DailyDOOH, 28 Sep 2026). The full OAAA resource is members-only, so we quote only the public summary.

That’s survey data about consideration, not a measured sales lift. It’s still a useful reminder: part of what a screen at the counter does is keep you in the shopper’s mind for a later trip. A short test may miss that, which is another reason to repeat tests rather than judge on one flight.

---

Privacy: Why This Approach Ages Well

Measurement built on aggregate store data avoids the questions that come with collecting or buying location trails. That matters as more states pass privacy laws (see state privacy laws for local advertisers) and as shoppers grow wary of in-store tracking (see retail media privacy on in-store screens).

It’s also simpler. You already own your sales data. A holdout test doesn’t require a new data contract. This isn’t legal advice. If you collect data at the register or on a landing page, follow the rules that apply to you.

---

What This Article Is Not

• Not a Take 10 result. The Love’s, VIOOH/Arla, and OAAA/Harris figures are theirs.
• Not a lift forecast. One beverage promotion at travel stops, or nine Norwegian grocery campaigns, won’t predict your campaign.
• Not a Take 10 CPM, rate card, or venue count.
• Not legal advice on privacy law.

---

Sources

1. Love’s Travel Stops — Love’s Media Group campaign delivers 23% incremental unit lift for leading beverage brand (Las Vegas, 6 Oct 2026; MarketDial test-and-control; 51 control stores after four outliers; campaign 26 Feb–22 Apr 2026).
2. VIOOH — The Proof is Physical: How Arla Foods Delivered Performance with Programmatic DOOH (Foot Traffic Attribution with Spotzi and JCDecaux).
3. invidis — Viooh Study Reports Store-Visit Gains From Arla’s DooH Campaigns (24 Sep 2026; Norway, 2025; caveats).
4. DailyDOOH — OOH Across Experiential + Retail @YourOAAA Webinar (28 Sep 2026; OAAA / Harris Poll summary). Full OAAA resource: members-only.
5. OOH TODAY — The Physical World Can Drive Digital Action (10 Sep 2026; OAAA digital-outcome figures; partner content).
6. Vistar Media — Vistar Media brings T-Mobile audiences to digital out-of-home (5 Oct 2026; targeting context only).

---

Next Steps

Pick your outcome first: sales, visits, or direct responses. Then pick your control: comparable stores or areas where your ad won’t run. Then book.

On self-serve, you choose the convenience-store and ATM venues and the dates, so you can leave comparable areas unbooked as your control. Start in the Take 10 Media Ad Portal; your account and video still go through review before the campaign goes live. For a same-day promo on a single screen, Instant Ads are the quicker option.

Related reading: matched-market testing for in-store ads · how DOOH drives in-store sales · QR codes in in-store advertising · NACS and c-store retail media.