Performance Max store visits and actual sales measure different things. Store visits are a modeled count of people who interacted with your ads and later walked into a location. Actual sales are transactions in your POS. The gap between them is mostly visits that would have happened without the ads, and the only way to size it is a test that compares locations where PMax ran with locations where it didn’t.
This matters because Performance Max optimizes toward whatever you tell it to count. Feed it store visits and it will find people likely to visit, and the cheapest people likely to visit are your existing customers. Below: how Google builds the number, why it diverges from your till, and a worked example of a pause test with the arithmetic.
What a Performance Max store visit actually is
Google is fairly open about the method in its help pages. Store visits use anonymous, aggregated statistics that are then extrapolated to estimate the wider customer base, and Google Ads creates the modeled numbers from current and past data on clicks, views and later visits. To check its detection, Google surveys select users from a panel of more than 10 million Google Opinion Rewards volunteers (Source: Google Ads Help, “About store visit conversions”, 2026).
There are thresholds, too. Your ads need enough clicks or impressions, and your locations enough foot traffic, to pass Google’s privacy thresholds, which vary by advertiser. You need active location assets (or affiliate location assets), and Business Profile locations must be verified. You also can’t bid on store visits and store sales at the same time (Source: Google Ads Help, 2026).
Two reporting quirks catch people out. Store visits are reported by the date of the ad click or impression, not the date of the visit, and figures for recent days change as the model collects more information. Store visits also sit in the “All conversions” column unless you set store visits as an account-default goal (Source: Google Ads Help, “Show store visit conversions”, 2026). So last week’s number will move, and it’s dated to when people saw the ad.
Performance Max inherited store goals from Local campaigns, which Google upgraded into Performance Max in 2022 (Source: Google Ads Help, 2022). If you ran Local campaigns before, your store-visit history carried over with them.
Why store visits run ahead of actual sales
- A visit isn’t a purchase. People walk in, look and leave. The gap is small for restaurants and large for furniture or auto dealers.
- Many visitors were coming anyway. Performance Max serves on Search (including searches for your brand name), YouTube, Discover, Gmail, Display, Search partners and Maps. A regular who searches your name on the way over and taps the ad is counted as a store visit. The ad changed nothing.
- The windows are generous. Click-attributed store visits use a 30-day window by default. Engaged views count visits within 3 days of watching at least 10 seconds of a video ad, and view-through counts visits within 1 day of seeing an ad (Source: Google Ads Help, “About store visit conversion window”, 2026). For a business people visit weekly, 30 days catches a lot of routine.
- You set the value. Store visit value is whatever you assign. A generous default value per visit turns a modeled count into an impressive ROAS that has never touched your POS.
None of this means Google is inflating the number on purpose. The model may estimate visits after ad interactions quite well. It just answers a different question from the one your finance team is asking. For more on why click-based credit drifts from what happened at the counter, see our guide to why click attribution misses in-store sales.
Store visits vs store sales conversions
Google also offers store sales measurement, which estimates in-store transactions and revenue influenced by ad interactions. It works by matching consented first-party transaction data you upload, or, in the US, through third-party partner data and receipts shared via Google Opinion Rewards. Uploaded conversions can take up to 7 days to appear, and the store sales window matches your store visits window, 30 days by default (Source: Google Ads Help, store sales measurement and onboarding guide, 2026).
Store sales are closer to the money than store visits, because they count transactions rather than footsteps. But they rest on the same logic: people who interacted with an ad and then bought. A customer who would have bought anyway still counts. If your accounts are allowlisted for it, use store sales instead of visits for bidding. Then measure what the campaign adds separately.
| Number | What it counts | Answers “did the ads cause it?” |
|---|---|---|
| Store visits | Modeled visits after an ad click, engaged view or view | No |
| Store sales | Estimated transactions or revenue after ad interactions | No |
| Extra transactions from a market test | Sales at locations where PMax ran, minus a twin of locations where it didn’t | Yes, with a range |
How to compare Performance Max with your actual sales
Performance Max can be geotargeted, so you can switch it off in some markets and leave it on in others. That’s the whole trick.
- Pull transactions by location, by week. From Toast, Square or your POS export. Use checks or transactions, not revenue, so basket size doesn’t add noise.
- Pick matched markets to pause. Around a sixth to a quarter of your markets, chosen by pairing similar markets and flipping a coin. Leave out markets with openings, remodels or local promotions.
- Keep brand Search running. If you have a standard Search campaign on your brand name, leave it on everywhere. You want to know what PMax adds, not whether people can find you when they search your name.
- Exclude the paused markets for six weeks. Use location exclusions in the PMax campaign, change nothing else, and log the dates.
- Compare with a twin. Build a weighted blend of the markets that kept PMax, fitted on the previous six months. The shortfall in the paused markets, against that twin, is what PMax was adding there.
Pausing feels riskier than it is. For six weeks in a sixth of your markets, the worst case is a small, temporary dip in places you chose. The alternative is spending the whole budget on a number you can’t check.
Worked example: Performance Max at a 60-location restaurant chain
This example is illustrative. A 60-restaurant casual-dining chain spends $60,000 a month on Performance Max for store goals. Google reports 18,000 store visits a month, or $3.33 per visit. The chain pauses PMax for six weeks in 10 matched markets holding 15 restaurants. Those markets carried a quarter of the spend ($15,000 a month, about $3,462 a week) and a quarter of the reported visits (4,500 a month, about 1,039 a week). Each restaurant averages 2,800 checks a week.
| Line | Value | Working |
|---|---|---|
| Checks per week in paused markets | 42,000 | 15 × 2,800 |
| Fall in checks vs twin while paused | −0.9% | 95% range −0.4% to −1.4% |
| Extra checks PMax was adding per week | 378 | 42,000 × 0.9% |
| Range of extra checks per week | 168 to 588 | 42,000 × 0.4% and × 1.4% |
| Store visits Google reported per week | 1,039 | 4,500 × 12 ÷ 52 |
| Extra checks as a share of reported visits | 36% | 378 ÷ 1,039 |
| PMax spend per week in those markets | $3,462 | $15,000 × 12 ÷ 52 |
| Cost per reported store visit | $3.33 | $60,000 ÷ 18,000 |
| Cost per extra check | $9.16 | $3,462 ÷ 378 |
| Range of cost per extra check | $5.89 to $20.61 | $3,462 ÷ 588 and ÷ 168 |
So is PMax worth it here? An average check of $58 at about a 35% contribution after food and hourly labor is worth $20.30. At $9.16 per extra check, PMax pays. On contribution, though, the dashboard implies $6.09 back for every $1 (18,000 × $20.30 ÷ $60,000), and the test says about $2.22 ($20.30 ÷ $9.16). Still a good channel, nearly three times as expensive as reported, and the top of the range ($20.61) sits right at break-even.
That’s the typical shape of this result. Performance Max usually isn’t worthless for physical locations. It just claims a lot of customers who were already on their way.
What to do with the gap between store visits and sales
- Use the ratio as a calibration factor. If tests show about 36 extra checks per 100 reported visits, divide Google’s cost per visit by 0.36 when you compare PMax with other channels. Re-test once or twice a year, because the ratio moves with your creative, your brand Search setup and seasonality.
- Lower the value per visit. If you assign a value to store visits, base it on what a reported visit is actually worth after the discount, not on your average check. Otherwise the bidding chases volume that isn’t there.
- Check the brand share. Google added full search terms reporting and a channel performance page to Performance Max in 2025 (Source: Google Ads blog, 2025). Use them to see how much spend goes to your brand name and to retargeting-heavy placements. A high brand share usually means a lower extra-customer ratio.
- Decide budgets by location, not by campaign. The pause test also shows which markets respond. Shift money toward markets with the lowest cost per extra check.
Common mistakes when comparing PMax store visits with sales
- Treating store visits as customers. They’re modeled visits after ad interactions. Not purchases, and not caused purchases.
- Reading last week’s number. Recent days get revised and are dated to the ad interaction. Wait until the conversion window has passed.
- Pausing brand Search along with PMax. Then you’re testing whether people can find your website, which isn’t the question.
- Comparing this month with last month. Seasonality, weather and menu changes swamp a PMax effect. Use a twin of markets that lived through the same weeks.
- Running a two-week pause. It will lie to you. Weekly noise at one location is bigger than most PMax effects; six weeks across several markets is the minimum.
- Leaving a big default value per visit. It inflates ROAS in the dashboard and pushes bidding toward people who were coming anyway.
When store visit data is still useful
Store visits aren’t useless. Inside one campaign, comparing asset groups, audiences or locations against each other, they’re a reasonable directional signal, because the same biases apply to every option. They also help Smart Bidding find people who visit stores at all, which beats optimizing for website clicks if your money is made in person.
Where store visits fail is the budget question: how much should go into PMax versus Meta, radio or nothing at all. That question needs extra customers at your locations. If you run fewer than about 15 locations, a market test gets noisy; the guide to incrementality testing for multi-location businesses covers what’s realistic at different sizes.
Questions people ask about PMax store visits
How does Google measure store visits in Performance Max?
Google observes visits in anonymous, aggregated data from some users, then extrapolates with a model built on current and past data about clicks, views and visits. It calibrates with surveys of its Google Opinion Rewards panel and only reports when you pass privacy thresholds for ad interactions and foot traffic.
Are Google Ads store visits accurate?
They can be a fair estimate of visits after ad interactions, and Google doesn’t publish an accuracy figure. The bigger problem is what they count: visits by people who interacted with an ad, including many who would have come anyway. Accuracy of the model isn’t the same as accuracy about what the ads caused.
Why are my store visits higher than my sales?
Visits aren’t purchases, many visitors were regulars or brand searchers already on their way, and the 30-day click window catches routine visits. The reported count is also extrapolated, dated to the ad interaction and revised over time, so it won’t reconcile line by line with your POS.
What is the difference between store visits and store sales in Google Ads?
Store visits estimate visits to your locations after ad interactions. Store sales estimate in-store transactions or revenue after ad interactions, using your uploaded first-party data or, in the US, partner data. You can bid on one or the other, not both. Neither measures what the ads added.
How do I measure the real impact of Performance Max on in-store sales?
Exclude PMax in a set of matched markets for about six weeks, keep brand Search and everything else unchanged, and compare transactions by location with a weighted twin of markets where PMax kept running. The shortfall is what PMax was adding, and dividing spend by it gives cost per extra customer.
Keep the store visit column. Just stop asking it to settle the budget. Your POS, read against a fair comparison, can do that.
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.