Connecting your POS data to something useful

Your point of sale holds the only numbers in your business that are genuinely deterministic. Money changed hands, at a time, for specific items. Everything on the marketing side is either sampled, modelled, or blocked by a browser setting. When those two datasets disagree, the register is right.

Most owners never put them side by side, so the marketing conversation happens entirely inside tools that cannot see revenue.

What is the smallest useful join?

Date. That is the whole answer, and it is unglamorous enough that vendors rarely propose it.

Build one table. One row per day. Columns for net sales, transaction count, average ticket, and your dayparts. Next to those, columns for sessions, inquiries, calls, and ad spend from your marketing sources.

Ninety days of that will tell you things a dashboard costing thousands will not. Whether your busiest sales days follow your highest inquiry days, and by how many days. Whether the promotion that produced traffic produced tickets. Whether your slow daypart is a demand problem or a staffing one.

The lag between a marketing action and a transaction is the most valuable thing this table reveals, and almost nobody measures it. In a restaurant it might be same day. In a home service business it can be three weeks. If you do not know your lag, every campaign assessment you have ever done was measured over the wrong window.

What can a point of sale actually tell me that analytics cannot?

Four things, and all four change decisions.

Real revenue, not proxy revenue. Analytics conversions are events. The register is money. When your ad platform reports twelve conversions and the register shows four new customers, the register is the number to plan from.

Item and category mix. Marketing that drives volume in your lowest margin category is not the win the traffic report says it is. The mix shift is invisible outside the point of sale.

Repeat behaviour. How many of this month's transactions came from customers you already had. For most local businesses this is the strongest predictor of the next two quarters and the least reported number in the building.

Time of day and day of week. Demand shape. This is where marketing spend gets wasted most often, pushing volume into hours you were already full and ignoring the ones you were not. Maple's analysis of 1.2 million calls across more than 1,000 US restaurant and local business locations found 68% of calls arrive during the lunch and dinner rushes, 28% at lunch and 40% at dinner, and that 58% of reservations still start with a phone call. The phone peaks exactly when the floor is least able to answer it, and that collision only becomes visible when call data and daypart sales sit on the same row.

That last point connects straight to the operating side. The daily numbers an operator should see before service come from the same export, which means one report can serve both the floor and the marketing review.

Why not just compare myself to an industry average check?

Because for the segments most independents operate in, there is no industry average check. There is a number in circulation, and it has no source.

The claim is that the average quick service check runs somewhere between $8 and $12. Trace it and it dead ends. Every US quick service and fast casual company that files with the SEC discloses only the percentage change in average check, never a dollar level. Chipotle's FY2025 filing is representative: average check 1.2%, menu price increase 2.1%, check mix negative 0.9%. Wendy's, McDonald's, Dine Brands, El Pollo Loco, Portillo's, Texas Roadhouse and BJ's all do the same. Technomic and Circana own the dollar level data and it sits behind a paywall. The range on the content farms has no primary document behind it at all.

Where dollar levels genuinely exist, they exist for casual and fine dining, and they are worth knowing because the definition is unusually precise. Darden's FY2026 10-K, covering the fiscal year ended 31 May 2026, defines average check per person as total sales divided by number of entrées sold and reports $25.00 for Olive Garden, $19.50 for Cheddar's Scratch Kitchen, $28.50 for LongHorn Steakhouse and $107.00 for The Capital Grille. Cheesecake Factory reported approximately $31.79 for fiscal 2025. Those are audited, full population figures, and they are free.

Average party size is worse. A full text search of SEC filings for the phrase returns nothing. OpenTable states in its own 2026 methodology that it analysed party size for every active US restaurant on its platform across an eight month window, and then publishes no party size number. The platforms have it. None of them release it.

So pull yours. Toast, Square, Clover, Lightspeed and TouchBistro all report average check and covers per ticket natively. It takes five minutes, it is your actual business, and no benchmark argument can follow it.

What does the join look like with real arithmetic?

Here is one month off the table. Use your own figures.

The month. Net sales $96,000 across 3,200 transactions. Average ticket $30.00.

Split by channel, straight from the point of sale. Dine in $62,000. Direct, meaning your own phone and your own ordering page, $16,000. Delivery marketplace $18,000.

Now price the marketplace column. DoorDash publishes commission rates of 15% for Basic, 25% for Plus and 30% for Premier. Uber Eats runs 20%, 25% and 30% across Lite, Plus and Premium. Grubhub advertises 5% to 20% marketing commission, with delivery starting at a further 10% on top. All three converge at 30% for a restaurant on the top tier using the platform's drivers. At 30%, your $18,000 marketplace column paid $5,400 in commission.

Convert that to margin. The National Restaurant Association describes the pre-pandemic norm for a typical independent restaurant as roughly 5% pre-tax profit, which means one dollar of margin takes roughly twenty dollars of sales. On that arithmetic, $5,400 of commission is the margin equivalent of $108,000 in sales. That is more than the whole month.

Then size the decision. Move 20% of the marketplace volume onto your own channel and you shift $3,600 of sales, saving $1,080 in commission. At the same margin arithmetic, $1,080 is what $21,600 of additional sales would have produced. That is the case for owning the direct channel, and it came out of a spreadsheet with a date column.

Run it yourself. Your marketplace sales for last month, times your actual commission tier, divided by your own pre-tax margin. If you do not know your margin, the answer is your accountant, not a marketing tool.

Does the register also settle the food cost argument?

It settles the version of it that matters, which is scale rather than discipline.

The National Restaurant Association's 2025 Operations Data Abstract, built on data from more than 900 operators, cuts full service by volume. Restaurants at $2 million or more in annual sales ran food and non alcohol beverage costs at a median of 31.0% of sales and income before taxes at 4.3%. Restaurants under $2 million ran 33.7% and 1.1%.

That is 2.7 points more food cost and 3.2 points less pre-tax margin for the smaller operator, before anybody talks about purchasing skill. Scale buys purchasing power and absorbs fixed cost. It is also the strongest argument for why incremental covers matter more to a small independent than to a larger one, and why a marketing decision that fills a slow daypart is worth more than the traffic report makes it look.

How do I get the data out without a developer?

Almost every modern point of sale has a scheduled export or a reporting API. Start with the export.

Look for a daily sales summary report that can be emailed on a schedule or dropped into a folder. Set it to CSV. Set it to daily. Send it to a dedicated inbox or a cloud folder, and let ninety days accumulate before you try to analyse anything.

Then build the table in a spreadsheet. Not a warehouse, not a business intelligence platform. A spreadsheet with a date column and a manual paste once a week is enough to answer the questions above, and it is enough for a surprisingly long time.

Two warnings from doing this badly. Time zones and business day boundaries will not match between your point of sale and your analytics, so a restaurant closing at 1am will have transactions land on the wrong day unless you define the business day explicitly. And refunds, comps, and voids need a consistent treatment, decided once and written down.

If your marketing numbers still look wrong against the register after that, the cause is usually not the register. Website analytics and ad platforms disagree with each other by design, and neither was built to reconcile to a general ledger.

When is a real integration worth it?

When the manual join has already proved its value and the volume of it is costing you real hours.

Reasonable triggers: more than three locations, more than one point of sale system, a loyalty program you actually want to segment, or a marketing spend large enough that a two percent improvement pays for the build.

Bad triggers: a vendor demo, a feeling that you should be more data driven, or a competitor mentioning their stack.

When you do build, insist on one thing above all: the raw daily export keeps running in parallel and lands somewhere you control. Integrations break, vendors get replaced, and platforms end support. Owning the underlying files is the same discipline that makes server log analysis the only fully deterministic first-party measurement available on the website side. Whoever holds the raw data holds the ability to check anybody's work.

What to do this week

Find the scheduled report feature in your point of sale. Set a daily sales summary to export as CSV, automatically, to a folder you own.

Create one spreadsheet with a date column. Paste the last ninety days if the system will give them to you. Add columns for sessions, inquiries, and spend from your marketing sources.

Pull your own average check and covers per ticket while you are in there, and write them on the front of the file. That single line ends every future conversation that starts with a benchmark somebody read somewhere.

Look for the lag. Line up your best inquiry weeks against your best revenue weeks and count the days between them. Write the number down and use it for every campaign assessment from now on.

If the table shows demand arriving but not converting, the problem is upstream of the register. Check what a first time visitor sees, because the top of your page has about five seconds to answer who you are and what you sell, and calls that never happen never reach the point of sale. Counting calls as conversions properly closes the other half of that gap.

Questions about your POS data?

Email me at eric@seod.com with the name of your point of sale system and one question you wish you could answer about your sales. I will tell you whether that system will export what you need, where the export lives, and the one join worth building first. I have pulled data out of most of the common restaurant and retail systems.

I answer these myself. If the honest answer is that your system will not give you what you want, I will tell you that instead of proposing a workaround.

More on measurement sits in the analytics library.