It's four in the afternoon on a Tuesday and you have about six minutes of your owner's attention before the delivery truck shows up. You have been reading about voice AI for a month and you think it would fix the phone. You open with how much it could grow takeout revenue.
You lost. Not because the idea is bad, but because you asked an owner to approve a new expense against a projection, and owners have been approving expenses against projections for years and have a folder of receipts to show for it.
The version that works is shorter and less exciting. You are not proposing a new cost. You are proposing to move one.
The line items this actually competes with
Almost every restaurant already pays something for the phone. It's rarely called that, which is why it's easy to miss.
An answering service, if you use one for after-hours or overflow, is the cleanest comparison because it's a monthly invoice with a number on it. Message-taking services commonly run in the low hundreds per month and they do not take orders, which makes the comparison favorable before you have said anything about the technology. That specific trade is laid out in AI phone ordering versus answering services.
A dedicated phone person, or the fraction of a person you staff during dinner specifically because the phone rings, is the next comparison. Add up the hours honestly. If someone comes in an hour early on Friday and Saturday to handle call volume, that's a real weekly wage number.
An extra line, a call-overflow feature from your carrier, or an old auto-attendant with a maintenance contract nobody has looked at since it was installed. Pull the phone bill. There is often something on it.
And the least visible one, which is a manager stepping away from the pass or the host stand to answer. That is not a line item and you should not pretend it is, but it is the thing your owner has personally done and been annoyed by.
Why "it will increase sales" is the weakest thing you can lead with
Owners discount revenue projections, and they're right to. Every vendor who has ever walked in the door brought one.
The problem is not that the revenue claim is false. Recovered calls are real orders and the arithmetic is sound. The problem is that it arrives in the same shape as every claim the owner has already learned to distrust, so it gets the same discount, and by the time they've discounted it the number no longer clears the cost.
A swap does not get discounted the same way, because the owner can verify half of it by opening a drawer.
So lead with the swap and let the recovered orders be the second thing you say, framed as the part you're less sure about. Understating the upside is more persuasive here than pressing it, and it's honest, because you genuinely do not know yet how many of those missed calls would have converted.
Doing the arithmetic where they can see it
Bring four numbers on one page. No deck.
- The monthly cost of whatever you're proposing to replace, taken from an actual invoice or an actual schedule rather than an estimate.
- Your average phone ticket, pulled from the POS and separated from walk-in and online orders.
- Missed calls during your two busiest hours over one full week, from your phone provider. The counting method is in how many calls does your restaurant miss.
- The monthly cost of the thing you want, which starts at $250 for X1 Voice and is on the pricing page where the owner can check it themselves.
Then do one multiplication out loud: missed calls times average ticket, per week. Stop there. Do not annualize it, do not apply a conversion rate, do not add a lifetime value multiplier. Every additional step makes the number bigger and the case weaker, because each step is an assumption the owner now gets to argue with instead of arguing with the decision.
If your restaurant misses fifteen calls a week during peak and your average phone ticket is thirty dollars, you have a weekly figure that speaks for itself against a monthly cost. If the numbers do not work at your volume, they do not work, and you should say so rather than stretching the assumptions until they do. A small operation with a handful of calls a week genuinely should not buy this.
When there is no bill to swap
Some restaurants have no answering service, no extra phone staffing, and no line on the carrier bill. The phone just rings and sometimes nobody gets it.
That's a harder pitch, because the cost you're pointing at has never appeared anywhere the owner can see. Nobody sends you an invoice for the customer who called, got a busy signal, and ordered from the place down the street. That invisibility is the whole subject of the real cost of a missed phone call.
In that case, make the loss visible before you ask for money. Have someone sit with the phone log for one week and write down every call that rang out, with the time. Hand the owner the list. A page of specific missed calls at 6:40, 6:52, 7:15 on a Friday does something a percentage does not, which is make the owner remember that Friday.
The three questions you should answer before they're asked
Owners ask the same three things, and answering them unprompted is worth more than any argument you make.
What happens if it doesn't work. Say plainly what the exit looks like, what you keep, and how the phone number comes back to you. An answer that admits the possibility of failure is more credible than one that doesn't.
Who has to do work. Be honest that setup is typically under 24 hours but that someone has to verify the menu, the hours, and the delivery rules, and that a manager needs ten minutes a day for the first two weeks. Owners tolerate work. They do not tolerate discovering it later.
How will we know. This is where a dated trial earns its keep. A 30-day evaluation plan with a written decision rule converts an open-ended commitment into a bounded one, and bounded commitments get approved.
Ask for a decision on a date
The failure mode is not rejection. It's a yes-in-principle that never becomes a start date, and then it's November and nobody changes anything in November.
Close by naming a date and what happens on it. "I'd like to start a 30-day trial the week of the ninth. If you want to think about it, can we decide by Friday?" Then, if the answer on Friday is no, ask which of the four numbers on your page would have to be different. Sometimes the answer is a number you can go get. Sometimes the answer tells you the owner was never going to approve it, which is also worth knowing on a Friday rather than in March.