Most measurement advice for local businesses assumes you have more than one location: compare the markets where ads ran with the markets where they didn’t. With one location, there is no second market. Before-and-after comparisons are tempting, but they mix your ads up with weather, holidays, a competitor opening down the street and plain luck.
There are still two honest options. If you know where each customer lives — jobs, appointments, patients — your ZIP codes can stand in for markets. If you don’t, you can use time instead: switch ads on and off in a planned pattern and compare. This guide covers the second, the on/off test.
First: is the effect big enough to see?
Every location has good weeks and bad weeks. If your ads add fewer customers than that normal swing, no method can prove what they did. A restaurant serving 1,200 guests a week whose weeks routinely vary by 8% has a swing of about 100 guests. A $2,000 monthly budget at $25 per extra guest adds about 18 a week — real money, but invisible against that swing.
How an on/off test works
- Pick blocks of two weeks. One-week blocks are too short: ads take a few days to build up and a few days to fade.
- Alternate on and off, decided in advance. Write the schedule down before you start, and don’t move blocks because a week “looks slow.”
- Run at least eight weeks — twelve is better. More blocks average out the weeks that are odd for other reasons.
- Change nothing else. No new promotions, price changes or menu launches during the test. If something unavoidable happens, note the week.
- Compare average customers per week, on versus off. Leave out the first few days of each block so carry-over from the previous block doesn’t blur the difference.
| Weeks | Ads | Note |
|---|---|---|
| 1–2 | On | Normal budget and targeting |
| 3–4 | Off | Pause every paid campaign |
| 5–6 | On | |
| 7–8 | Off | |
| 9–10 | On | |
| 11–12 | Off | Compare on-weeks with off-weeks |
Reading the result honestly
Subtract the off-week average from the on-week average to get extra customers per week, then divide your weekly ad spend by that number to get a cost per extra customer. Then check it against the swing: if the difference is smaller than about twice your normal week-to-week variation, the test couldn’t tell — that is not the same as “the ads don’t work.”
Treat the answer as directional. Without a comparison location, a heat wave in one block or a holiday in another can still tilt the result. Repeating the test in a different season is the cheapest way to build confidence.
When a ZIP-level test is better
If you record a ZIP code for each job, appointment or patient, you can do better than on/off. Run ads in half your ZIP codes, hold back a matched other half, and compare them over the same weeks. Both groups live through the same weather and season, so the gap is the ads. Countlift is opening ZIP-level tests to a small early-access group of single-location service businesses.
Next step
Check your numbers before you plan the test.
The free check tells you whether your budget can produce a lift big enough to see, and whether a ZIP-level test would suit you better.
Written by The Countlift team at First Byte in The Woodlands, Texas. Questions or corrections? Email hello@countlift.com.