Your First Two Weeks

The first real call is the one that tells you. Not the demo, not the recording — the one at 6:12 on a Tuesday, when a caller asks whether the special is still on and the right answer comes back.

Getting there is not a leap of faith. It is a short list of steps spread across the fourteen days of the trial, and most of them take less time than a shift change. Done in this order, the problems show up on a Tuesday afternoon instead of a Friday night.

TL;DR

• Day 1 is the menu: import it from your POS or a menu page, then read every price and modifier back to yourself.
• Day 2 is three test orders and one $0.00 test ticket, which prints without ever marking an order paid.
• Day 3 is hours, wait times, and the 86 board, because those are the answers a caller checks first.
• Day 4 is limits, payment, and the 0 key, so a big order and a caller who wants a person both land right.
• Days 5 to 7 put it in front of real callers; week two is one recording, one fix, and a decision by day 14.

Day 1: the menu, then a second look at it

The menu decides almost everything. If the phone reads a price you changed two months ago, nothing later in the fortnight matters.

Pick one source: a catalog from Toast, Square, Clover, or Olo, or a menu page from Toast, DoorDash, Uber Eats, Clover online ordering, Slice, or a link you paste in. One source per shop, so there is one answer to how much the large is.

Then look at what arrived. Three things are worth the second look.

• Prices. A marketplace menu can carry a markup percent, so the phone quotes the delivery price rather than your counter price. Set the markup to zero if you want one price everywhere.
• Choice groups. Nested groups, half-and-half, and sizes are the part that goes wrong first. Find your most-ordered item with the most modifiers and read it on the page the way a caller says it out loud.
• The names. The phone needs the words callers use, not the words on the POS button. If everyone asks for a large pep, add that as one of the ways a caller says a large pepperoni.

Your edits survive the daily refresh, a change from the source lands on the fields you did not touch, and a deletion from the source deletes. That is the deal, and it is why the second look belongs on day 1 rather than day 10. The menu sync piece goes deeper on how the merge behaves.

Day 2: three test orders and one test ticket

Place three orders, in this order, on a real call.

1. The simplest item on the menu. No modifiers — you are checking that the read-back is right and that the ticket prints.
2. The worst item on the menu. The one with the size, the crust, the side, and the note, because that is where a modifier gets dropped.
3. An order over your limit. Ten items in total and 8 of one item are the defaults, so order eleven of something and hear what the caller is told.

After each one, walk over and look at the ticket. Then run the test connection: it prints one $0.00 ticket and never marks an order paid. That is the check you can repeat every morning without touching the till. If the ticket does not land where the kitchen looks, the Toast path shows what should have happened.

Day 3: hours, wait times, and the 86 board

Three answers a caller checks before ordering anything.

Hours. The line answers after close and on holidays, so your hours do two jobs: they tell the caller when you are open, and they stop the agent offering a pickup time you cannot make. Sync the Google listing, correct the grid, then add the holiday exceptions and the last-call cutoff. The exceptions matter most, because a holiday is when a wrong answer is most expensive.

Wait time. You set a baseline, or you set nothing. If you set nothing, the agent says nothing about the wait. There is an optional per-ticket figure on top of the baseline and a cap on it, so a full kitchen pushes the number up without letting it run away.

The 86 board. Mark the thing you ran out of on Saturday. Sold out is a toggle in the menu board, Square's sold-out flag and Clover's availability survive the refresh, and your 86 outranks a source that still says available. Then call and ask for it.

Day 3 takes about an hour, and it is the hour that stops a caller hearing a promise the kitchen cannot keep.

Day 4: limits, payment, and the 0 key

This is the day the money and the handoff get decided.

Limits. The defaults are 10 items in total and 8 of one item. You can change either, add an order-value limit, or set one to zero to remove it. Then decide what happens above the limit, because that is one setting and two different Fridays:

Either way the order is never dropped in silence, and you find out what it sounds like on a Tuesday afternoon rather than on the party order.

Payment. Pick one of two. With cards off, the order is confirmed on the call and paid when the caller collects. With cards on, a caller who has paid you before has that card charged while they are still talking, and a first-time caller gets a text link with the line held open. No card number is ever read out, and paying the link keeps the card for next time. Pay for one order yourself before you go live, and read how the charge works.

The 0 key. Set the handset number, then press 0 on a test call and see which phone rings. No handset means no handoff and the caller stays with the agent, which is a fine choice as long as it is a choice.

Days 5 to 7: put it in front of real callers

Start with the quiet hours. Lunch on a Wednesday, or the two hours before close, when a wrong answer costs you a table instead of a rush.

Do not switch the whole week at once. Forward the line for one service, then read what came in. Three things to look for:

• Calls that were answered and turned into tickets.
• Calls that were answered and went nowhere, which is usually a menu question with no answer behind it.
• Calls that rang out, if any, and where they went instead.

If the internet at the shop has a habit of dropping, this is also the week to run the ten-minute reliability check, because a printer that cannot print eats an order you already took.

Week two: listen once, fix one thing, decide

Listen to one whole call, from hello to the read-back. Not five. One, all the way through, with the ticket in front of you.

Then fix one thing. The usual culprit is a menu item whose name on the page is not the name a caller uses. The second is a wait time that was never set, so the agent says nothing when someone asks how long. Neither is a rebuild. Both are a five-minute edit.

Then decide. Fourteen days is long enough to know whether the phone is answering calls you used to lose and whether the tickets are landing where the kitchen looks. The plan is $99 for 750 connected minutes or $299 for 5,000, with no overage, and at the cap calls forward to your own phone.

FAQ

Final Thoughts

The trial is not a test of whether the software is clever. It is a test of whether your menu, your hours, and your 86 board are right, because the phone only repeats what you tell it.

Two weeks, six steps, one recording. If the first Friday sounds better than the last one did, you have your answer. Take a test call with your own menu and start with day 1.