Agencies prove local ad results to multi-location clients by reporting what the client’s own registers and booking systems show: how many extra customers each location got because of the ads, what each one cost, and how sure you are. “Extra” means customers above what matched locations, where spend didn’t change, saw over the same weeks.
Platform dashboards can’t do that, and your client’s CFO knows it. This guide covers what proof looks like to a multi-location brand, how to set the rules before a test, how to report a result that disappoints, and a reporting rhythm that keeps clients from asking “but is it working?” every other week.
Why platform reports don’t prove local results
Meta reports store visits. Google reports store visits and conversions. The CTV partner reports exposed-household visits. Each counts the customers it can tie to its own ads, and the same customer often shows up in all three. Add them up and you can claim more customers than the client served.
Multi-location clients see through this faster than most. Their operators know their locations’ numbers by heart. Finance sees revenue by store. When the agency deck says Meta drove 9,200 visits and the regional manager’s stores were flat, the deck loses.
The people who approve your retainer already prefer a test to a platform report. Give them one.
What proof looks like to a multi-location client
- Their data, not yours. Transactions, bookings, visits or jobs from the client’s own POS or scheduling system, by location, by day. If the count comes from a system they already trust, nobody argues with it.
- By location. Operators manage locations. A brand-level ROAS means nothing to the regional manager whose six stores didn’t move.
- Against a comparison. Extra customers are measured against locations where spend held steady over the same weeks. Before-and-after doesn’t survive the first question about seasonality.
- With a range. “4,800 extra, somewhere between 3,500 and 6,100” sounds less exciting than “4,800” and is far more convincing. It shows you know what you don’t know.
- With a decision. Every report ends with what you’ll change: which locations get more, which get less, which test runs next.
If you’re building the internal side of this, the multi-location reporting dashboard guide covers layout. For the finance conversation specifically, see how to prove marketing ROI to a CFO.
Agree the rules before the first test
Most agency–client fights about results are really fights about definitions, started after the numbers came in. Write these down and get a yes in writing before the media goes live.
| Agree on | Example | Why it matters |
|---|---|---|
| The count | Transactions per location per day from the POS | Stops anyone switching to revenue or leads after the fact |
| Test and comparison locations | 10 test locations in two metros; 30 steady | Prevents cherry-picking the winners later |
| Length | 8 weeks plus a 2-week tail | Short tests produce noise that someone will over-read |
| Smallest lift worth finding | 5% at test locations | Sets the spend needed so the test can answer |
| Decision rule | Scale if cost per extra customer < $45 and the range is clear of zero | Turns the readout into a decision, not a debate |
| What a null means | No measurable lift at this spend; rerun or reallocate | Saves the relationship when the result disappoints |
The “smallest lift worth finding” line is the one agencies skip and regret. If the client’s budget can only detect a 15% lift and the media will realistically produce 4%, the test will come back inconclusive and look like your failure. Minimum detectable effect explains how to check before you pitch.
Worked example: one quarter for a 40-location client (illustrative)
An agency runs Meta and Google for a 40-location quick-lube chain: $180,000 in the quarter, $4,500 per location. A car through the bay contributes about $55 after parts and labour. The platforms report 9,200 store visits (Meta) and 7,400 (Google). Comparing each location with a weighted blend of the chain’s steady-spend markets gives a different picture.
- Platform claim. 9,200 + 7,400 = 16,600 visits, at $10.84 each.
- Measured. 4,800 extra cars, 95% range 3,500 to 6,100. The platforms claimed about 3.5 times what the bays saw.
- Cost. $180,000 ÷ 4,800 = $37.50 per extra car, range $29.51 to $51.43.
- Return. 4,800 × $55 = $264,000. ROI = ($264,000 − $180,000) ÷ $180,000 = 47%, range 7% to 86%.
- By location. 28 locations show a clear lift. 8 show no measurable lift. 4 are too new or too noisy to read.
Now the part the client actually pays for. The 8 no-lift locations took $36,000 and produced about 160 extra cars between them. The top 10 produced 1,730 for $45,000, or $26 each. Moving half the no-lift budget ($18,000) to the top 10 gives up about 80 cars. Even assuming the top 10 get more expensive with more spend, at $30–$45 per extra car the $18,000 buys 400 to 600. Net: +320 to +520 cars a quarter, or $17,600 to $28,600 of contribution, without spending another dollar.
That paragraph is what wins renewals. Not 16,600 visits. Allocating budget across locations goes deeper on the mechanics.
How to report a result that shows no lift
Sooner or later a test comes back flat. How you report it decides whether the client trusts the next win.
- Say it first. “The radio test in Columbus showed no measurable lift.” Don’t bury it on slide 14.
- Say what it means. A narrow range around zero means the media didn’t add customers at that spend. A wide range means the test couldn’t tell. They lead to different decisions.
- Say what it cost. The spend, and what the client learned for it. A clear null on $30,000 can save $300,000 next year.
- Say what’s next. Reallocate, rerun with more weeks or locations, or change the creative or offer. Pick one and put a date on it.
Weekly, monthly, quarterly: a reporting rhythm clients read
Different people at a multi-location client need different things, at different speeds. Match the rhythm to the reader and you’ll stop rebuilding the same deck every Friday.
- Weekly, for the marketing lead. Three lines: what moved, what’s flagged, what you changed. Slack beats email for this.
- Monthly, for operators. Their locations, extra customers, and anything that needs their help: a location that’s full, a promo that clashed.
- Quarterly, for the decision-maker. One page of proof, one page of moves, one page of next tests. If the QBR deck is longer than that, the important page is lost in it.
Setting this up across clients
The practical problem for an agency is doing all of the above for eight or 15 clients without a person per client rebuilding spreadsheets. Here’s how Countlift for agencies handles it, so you can judge whether it fits.
- One workspace, a brand per client. Each client keeps its own locations, data, connections and results.
- Client logins limited to their brand. Invite a client’s owner or CFO as a viewer of their brand only. They never see your other clients.
- Proof Report PDF. Extra customers by market and channel, cost per extra customer and 95% ranges, ready for the QBR.
- Change briefs. Approve a budget move and hand your buyers a brief for Meta, Google and local buys. Ad connections are reporting-only; your team applies the changes.
- Weekly Slack and monthly email. Per-brand digests that reach the client without you building a deck.
One limit to know up front: reports and the app carry Countlift branding. There’s no white-label option today.
Common mistakes agencies make proving local results
- Leading with platform ROAS. It invites the one question you can’t answer: would those customers have come anyway?
- Agreeing definitions afterwards. Write down the count and the decision rule before the media goes live.
- Pitching tests that can’t answer. Check the smallest detectable lift against the budget first.
- Reporting only the brand total. Operators manage locations. Give them their locations.
- Hiding a null. Clients find out. Report it first, with a plan.
- Dropping the range. A point estimate without a range looks precise and isn’t.
Questions agencies ask about proving local ad results
How do agencies prove ROI to multi-location clients?
By measuring extra customers per location from the client’s own POS or booking data, compared with matched locations where spend held steady. Report cost per extra customer with a 95% range and tie it to the client’s value per customer. Platform-reported conversions can support the story, but they can’t carry it.
What should a QBR for a multi-location client include?
One page of proof (extra customers, cost per extra customer, range, results by location), one page of budget moves with the expected effect, and one page of next tests with dates. Put platform metrics in an appendix. Anything longer and the decision-maker loses the thread.
How do I explain incrementality to a client?
Skip the word. Ask: “How many more customers came in at the locations where we advertised than at similar locations where we didn’t, over the same weeks?” That’s the whole idea. Then show the comparison chart and the range.
What if a test shows no lift?
Say so first and plainly. Explain whether the range was narrow around zero, meaning the media didn’t add customers at that spend, or wide, meaning the test couldn’t tell. Then name the next step and a date: reallocate, rerun larger, or change the offer.
Can clients see their results without seeing other clients?
In Countlift, yes. Each client is a separate brand in your workspace, and a client login only sees the brands you choose. Your own team can see every brand, or you can limit them too.
Agencies that report this way have a different kind of QBR. Less defending the numbers, more deciding what to do with them. That’s a better meeting for everyone, and a much easier renewal.
Next step
Put your own locations through the comparison.
Your first location report is free at any size. Bring a CSV of weekly sales and spend, or connect your POS and ad accounts.
Written by The Countlift team at First Byte in The Woodlands, Texas. Questions or corrections? Email hello@countlift.com.