- Your phone provider's call log already has the data — pull a full week including a Friday and Saturday.
- Missed calls cluster at exactly the hours you're most profitable and least able to answer.
- Multiply missed calls by a conservative conversion rate and your average ticket. Annualise it.
- Compare that number to the cost of covering the phone before deciding it's not a problem.
Missed calls are the easiest restaurant leak to ignore, because nothing about them shows up in your POS. There's no line item, no refund, no complaint. The only trace is a number in a call log that most operators never open.
Step one: get the actual number
You don't need to estimate this. Nearly every phone provider — carrier or VoIP — exposes a call log showing answered, missed and abandoned calls with timestamps. Pull a full week, and make sure it includes at least one Friday and one Saturday.
Then sort by hour and look for two things:
- Missed calls during service. Typically clustered between about 5pm and 8pm, when your floor is fullest and your staff are least able to stop.
- Calls outside opening hours. These are invisible in every other system you run, and they're pure loss — nobody was ever going to answer.
Most operators are surprised by the second category more than the first.
Step two: turn calls into dollars
A missed call isn't automatically a lost order. Some callers ring back, some were asking about hours. So use a deliberately conservative conversion assumption rather than an optimistic one — the point is a number you can defend, not the biggest number you can produce.
| Input | Where it comes from | Example |
|---|---|---|
| Missed calls per week | Your phone provider's call log | 25 |
| Share that would have ordered or booked | Your own conservative estimate | 40% |
| Average ticket | Your POS | $38 |
With those example inputs: 25 × 0.40 = 10 lost orders a week. At $38 each that's $380 a week, about $1,650 a month, or roughly $19,800 a year in revenue that never entered the building.
Those inputs are illustrative — substitute your own. The exercise only means something with your call log and your average ticket. Run it at a 25% conversion assumption too, so you know the floor as well as the middle.
Note that this is revenue, not profit. Apply your own food and labour costs to get to contribution. Even halved, the number is usually larger than operators expect.
Step three: compare it to the cost of fixing it
Now the number is useful, because you can put it against real options:
| Option | Roughly what it costs | What it doesn't solve |
|---|---|---|
| Do nothing | $0 | The leak continues, and it's largest on your busiest nights. |
| Voicemail | $0 | Most takeout callers won't leave one. A callback arrives after they've eaten. |
| A dedicated phone staffer | An hourly wage across every open hour | Still nothing after close, and one person handles one call at a time. |
| A traditional answering service | Monthly retainer plus per-minute charges | Operators take messages — they can't price an order or quote a pickup time. |
| An AI phone assistant | A flat monthly plan plus a per-completed-order fee | Won't replace a human on genuinely unusual calls — those should escalate. |
The part that compounds
There's a second cost that doesn't appear in the arithmetic. A customer who calls twice and gets no answer twice stops calling. They don't complain — they just quietly become someone else's regular, or they default to a delivery app where you'll now pay commission on orders you used to get for free.
That's the real argument for covering the phone: not the single order you lose tonight, but the customer you stop acquiring at all.
If you want to hear what automated coverage actually sounds like before deciding, there are unedited sample calls on the homepage — a reservation call and a pickup order call, start to finish.