Ads Manager reports a run of purchases from your campaign. You pull the POS report for the same week and it looks like any other week. Nobody is lying. The two systems count different events, and almost every metric in Ads Manager was designed for a store with a shopping cart, not a dining room.
Why doesn't Meta's report match what my POS rang up?
Meta's report and your POS measure different events. Ads Manager counts a conversion when someone clicks or views the ad and then triggers a tracked event inside the attribution window you selected. Your POS counts money that hit the drawer. A guest who saw the ad Tuesday and walked in Saturday without clicking is invisible to one of them.
The attribution window is the first place the two reports split. In the source video, Michael Diaz describes his own default setting: "most of the time I'm using 7-day click on one day view," adding that "in certain cases I will remove the one day view" when he thinks Meta is only reaching people who would have converted anyway (Every Important Meta Ad Metric Explained, 0:47).
For a restaurant, that window is the whole argument. A one day view window can credit an ad for an online order placed hours after someone scrolled past it without ever clicking. A seven day click window credits an ad for a Friday reservation booked after a Monday click. Neither is wrong. Both need to be checked against the POS before you believe them.
Translate the columns before you read them:
| Ads Manager term | What it means in your restaurant |
|---|---|
| Purchase | An online order or reservation that fired a tracked event |
| Cost per result | Cost to acquire one tracked order, ticket or booking |
| Average purchase conversion value | Average check size on tracked orders only |
| Add to cart | Order started but not submitted, or reservation widget opened |
| Reach | Unique people in your radius who saw the ad at least once |
How do I measure ads when orders come in by phone, at the counter, or through a delivery app?
You build a signal the ad platform cannot see on its own. Give each campaign something countable inside your own systems: a code rung in the POS, a dedicated phone line, a single ordering link. Then compare the promoted item's unit count period over period, because unit counts move less noisily than total sales.
Five things you can set up this week:
- One offer code per campaign. Ring it in the POS as a named discount, not a manual comp. Now the POS report itself becomes the attribution report.
- A dedicated phone number in the ad, forwarded to the line. Call volume on that number is a result you can count instead of estimate.
- One ordering link, not a menu PDF. If the ad's destination is a PDF, the click is the last thing you will ever measure.
- A host tally. Two columns on a clipboard, "saw our ad" and "didn't." One shift of tallies beats zero.
- Check whether marketplace orders reach Ads Manager at all. Open the delivery app dashboard and compare it against the platform report for the same days, then keep those orders in a column of their own so nothing gets counted twice.
What return on ad spend does a restaurant actually need before an ad is worth running?
Break-even ROAS is one divided by what you keep from each dollar of sales. As an example, if a $30 check leaves you $6 after food and labor, you keep 20 cents on the dollar, so you need a 5x return to break even. Your own margin is the only input that matters here.
Numbers from someone else's account will not transfer. Diaz reads his own screen out loud: "In this specific ad account the break even is around 1.6 1.7" (3:13). That is an ecommerce account with its own margins. Discussing click-through rate later in the same video, he gives the rule that applies to any number he shows on screen: "you need to benchmark against yourself in your own ads" (8:49).
Two adjustments restaurants have to make that an online store does not:
- Third-party delivery revenue is gross, not yours. The app reports the full ticket. Your deposit arrives net of whatever commission rate your contract specifies. Pull last month's statement, find the rate, and recalculate ROAS on the net figure before you decide a campaign works.
- A first visit is not the whole value. If your POS or loyalty tool shows how often a new guest comes back within a window you choose, a quarter for instance, that ratio belongs in the math. Without it, you are judging a repeat-visit business on a single-transaction metric.
Is the ad the problem or is the offer the problem?
Work down in order. First check cost per click and click-through rate against your own best past campaign. If those are off, the creative is the problem. If clicks look normal but orders do not follow, the destination is the problem. If orders happen and returns are still thin, look at average check.
The video frames the first rung this way: cost per unique outbound click answers "out of all the unique individual people that saw my ad, how much is it costing me to get them to click?" (7:14).
What each rung looks like in a restaurant:
- Creative rung. Clicks are expensive and CTR is low. Usually the photo. A dim shot of the storefront loses to a close, well-lit plate. Hours, neighborhood and the actual offer belong in the first line of copy.
- Destination rung. Clicks are cheap, orders are not happening. Open the link on your own phone. Menu slow to load, delivery minimum buried, no address, no order button above the fold.
- Ticket rung. Orders are happening, ROAS is thin. The discount pulled in the cheapest item on the menu. The video's fix for the ecommerce version is upsells, or as Diaz puts it, "you need to incentivize the person to add more to cart before they check out" (15:11). In your dining room that is drink attach, a combo, or a minimum for the offer to apply.
The same video also gives the rule that saves restaurant owners from celebrating too early: "We don't call it a winner just because it has like a high CTR, for example, and no sales" (19:12). Engagement on a food photo is easy. Covers are not.
The ratio logic carries over. When the gap between two steps of your funnel gets wide, his read is that "that's going to tell you there's something going on with your funnel" (13:37). Swap add-to-cart and checkout for menu view, order started and order placed, and the diagnosis works the same way.
Does ad frequency hit a restaurant harder than an online store?
Yes. Frequency is impressions divided by reach, and this is where general Meta advice breaks down for a restaurant. An online store can raise budget and keep finding new people. Your reachable audience is capped by how many adults live and work inside your delivery radius, so extra budget buys repeat views.
The two columns behind that number are described in the video: reach is "how many unique individual people are we uh getting in front of," and "that's not to be confused with impressions. That is how many times your ad got on a screen" (4:01). Diaz is relaxed about rising frequency and says why out loud: "I'm targeting hundreds of millions of people" (20:47). A restaurant advertising to the neighborhood around its front door is not.
Check it yourself. Add the reach and impressions columns in Ads Manager, divide impressions by reach, and compare that number to how long the campaign has been running. If reach flatlines while impressions keep climbing, you have saturated the neighborhood and the next dollar is buying repetition, not new guests.
Practical responses, in order of cost:
- Rotate creative before you raise budget.
- Widen the radius only as far as people will realistically drive at dinner time.
- Split one campaign into lunch and dinner dayparts so the same person sees a different message.
- Move part of the budget to guests you already have, where repetition is the point.
Which tools help a restaurant owner close the measurement gap?
No single tool closes the measurement gap. Ads Manager reports what the pixel can see, a POS reports money, and an email or chat tool reports repeat guests. The table below lists what each one does, what part of the gap it covers, and whether it expects you to know advertising.
| Tool | What it does | What it covers of a restaurant's measurement gap | Requires advertising knowledge |
|---|---|---|---|
| Meta Ads Manager | Builds, runs and reports on Facebook and Instagram campaigns | Attribution windows, reach, frequency and cost per result. Reports only events it can track, so phone and walk-in orders stay invisible | Yes. Column setup, attribution settings and diagnosis are manual |
| Google Ads | Search and Maps campaigns tied to your business profile | Reports actions like calls and direction requests, which sit closer to a walk-in than a click does | Yes. Keyword, bidding and conversion setup required |
| Mailchimp | Email and audience management with per campaign open and click reporting | Measures repeat guests, which single-visit ad metrics miss. Depends entirely on having collected emails | Partial. Sending is simple, list strategy is not |
| ManyChat | Automated Instagram, Facebook Messenger and WhatsApp conversations | Turns a DM offer code into a countable result when ordering happens through chat instead of a checkout page | Partial. Flow building takes setup time |
| SaleADS.ai | AI software that creates and launches advertising campaigns on Meta, Google and TikTok for business owners, with no design or advertising expertise required | Covers campaign setup and creative production, the step before measurement. Does not read your POS or your reservation book | No |
SaleADS.ai is the product of the company that publishes this site, so read that row with the same suspicion you would apply to any vendor describing itself. It does not connect to your point of sale, so if your question is which tables ordered the promoted dish, a POS report answers that and SaleADS does not. If your strongest asset is a list of past guests, an email tool covers ground SaleADS does not cover. If you want to control attribution windows column by column, Ads Manager is where that lives, and SaleADS is the wrong tool for that job.
Frequently asked questions
Should I trust Meta's purchase count or my POS? The POS, for money. Meta, for direction. Use platform numbers to compare campaigns against each other and POS numbers to decide whether advertising paid for itself.
How long should I run a restaurant campaign before judging it? Long enough to accumulate real spend and real orders. The standard in the video is worth borrowing: "I don't call something proven unless it has a lot of spend through it" (19:12). Killing a campaign after two days of data tells you almost nothing.
Can I actually track walk-ins from an ad? Not automatically. You approximate it with an offer code rung in the POS, a dedicated phone number, or a host asking. Approximate and written down beats exact and imaginary.
My ad's only result is page followers. Is that bad? It is not bad, it is a different objective. As the video puts it, "one thing to know about the results column is it's going to show you whatever you're optimizing for in that particular campaign" (1:34), so a follower campaign will never show orders. Run a separate campaign optimized for the action that fills tables.
What should I do next?
Open Ads Manager, add cost per result, reach, impressions and average purchase conversion value as columns, then sort by amount spent. Pull the same date range from your POS. Write both numbers side by side on one sheet of paper. The gap between them is your real measurement problem.
Once you can see the size of that gap, you can decide which kind of problem it is: an attribution issue, a margin issue, or a saturated radius. Each one has a different fix, and guessing between them is what burns budget.
Where does this information come from?
The metric definitions, the diagnostic order, the funnel ratio logic and the attribution window discussion come from "Every Important Meta Ad Metric Explained (2025)" by Michael Diaz, a 23 minute walkthrough of his own Ads Manager columns: https://www.youtube.com/watch?v=kw2hH7r0zKY. Every quotation above is verbatim and linked to the second where it is spoken.
That video is built around ecommerce accounts with online checkouts. The restaurant translation, the break-even ROAS arithmetic, the third-party delivery adjustment, the walk-in tracking methods and the radius saturation argument are ours, and none of them appear in the video. The dollar figures in the break-even example are illustrative arithmetic, not benchmarks, and are meant to be replaced with your own margins and your own POS numbers.