Software is the only thing a restaurant buys that nobody ever throws away. A broken oven gets removed; a fryer nobody uses gets sold. An app that twenty people opened last month sits there costing three hundred dollars a month for years, because switching it off feels like admitting something. This last topic is the five numbers that tell you which one you have, and the decision most owners never make.
📊 The five numbers of an app
| Number | Where it comes from | The example restaurant |
| Installs | The store dashboard, cumulative | 40 |
| Active in the last 30 days | The app dashboard — the only honest one | 16 |
| Orders per active user | Orders in the app ÷ active users | About 1 a month |
| Cost per active user | Monthly fee ÷ active users | $300 ÷ 16 = $18.75 |
| Share of sales | App sales ÷ total sales | $400 of $50,000 — 0.8% |
Only the second row is real. Installs are cumulative and only ever go up, which is why every provider's report leads with them; the number of humans who opened the thing in the last thirty days is the one that describes your business. Look at it monthly, next to the other numbers you already review — the reservations course put a half-hour weekly meeting in the diary, and this fits in it.
The fifth row is the one that ends arguments. An app moving 0.8% of sales is not a channel, it is a hobby with an invoice attached — and it is worth saying plainly that this is the ordinary outcome for a single restaurant, not a sign that something went wrong.
🔪 The rule for keeping it
Set the test before the emotions arrive, and apply it once a year on a fixed date.
The app stays if the sales it moves comfortably exceed what it costs to move them, counting the fee, the store accounts and the hours somebody spends on the menu. In the example restaurant the app produces $400 of sales a month against $300 of fee, which fails before you count anybody's time — it is spending seventy-five cents of every dollar it brings in, on a channel that is 0.8% of the business.
Two honest adjustments before you decide. First, count only the incremental sales: an order that would have arrived by phone anyway is not an app sale, and this is the same trap the reservations course found with booking platforms. Second, be fair about what else that money buys — three hundred dollars a month is a real marketing budget in the email or social courses, and the comparison is not "app or nothing", it is "app or the best alternative use".
And if it passes, keep it and say so out loud, because an app that is genuinely working deserves the maintenance attention topic 3 insisted on. The failure mode is not keeping a good app; it is keeping a dead one out of embarrassment.
🕯️ How to switch it off properly
If the answer is no, do it deliberately rather than by neglect — because the default, an app left in the store with an old menu, is the worst of every world.
Tell the people who used it, in the app and by email, with a date and a replacement: the mobile site, the wallet pass, the ordering page. Give the loyal ones something for the inconvenience, especially anyone holding points — those points are a promise you made, and the loyalty course is clear that breaking one costs more than honouring it. Export everything first: customers, contacts, order history, straight into the list the customer data course keeps. Then remove the app from the stores rather than leaving it downloadable, and cancel the subscription and the store accounts so they do not renew quietly.
Then take the money and put it somewhere with a measurement attached. That is the point of the whole exercise: the three hundred dollars was never wasted if the decision to stop was made on evidence, and it becomes an email programme, a wallet pass, a faster website or a month of the platform commission that actually brings new customers.
🧭 What you have built, and what closes here
You started with the only question that matters and found that it has three conditions, all of which have to be true (topic 1). You made your website fast on a phone, fixed the map listing, moved ordering and booking into the browser and put your loyalty card into the phone's wallet — which for most readers is where this course ends, with the money unspent (topic 2). If you went on, you chose between a custom build, a template and a listing with the year-two costs in front of you, and got the store accounts in your own name (topic 3). You removed the registration wall and made sure the four things people come for happen in sixty seconds (topic 4). You gave people one real reason to open it again and learned to send fewer notifications than you want to (topic 5). And you know the five numbers, one of which is real.
This also closes the operations run. The kitchen course fixed the flow behind the pass, the room course fixed what is in front of it, the reservations course filled that room on purpose, and this one stopped you spending money on a channel that would have moved under one percent of it. What comes next in the library leaves service entirely: what you buy and where it comes from, what you throw away, and the paperwork that arrives whether you were ready or not.