Ninety days out, a new restaurant has a lease, a contractor, a POS quote, and a hiring plan. What it usually does not have is a phone number, and that single missing item is the one that quietly sets the deadline for half a dozen other things.
The number goes on the sign, the awning, the takeout menus, the vehicle wrap, the Google listing, the delivery app profiles, and the aggregator directories that will copy your listing whether you ask or not. Each has its own lead time and its own cost to redo. Get the number wrong at day ninety and you are paying for it at day fifteen.
Ninety to sixty days out: the number itself
Decide first whether you are inheriting a number or creating one.
If you bought an existing restaurant and the phone number came with it, that number has years of accumulated call history and it appears on menus in drawers all over your neighborhood. Keep it. Get the porting authorization signed at closing, and start the port early, because carrier transfers move on their own schedule and nothing about the process gets faster because you are opening Friday. The mechanics are in the phone number porting guide.
If you are creating a number, get a local area code. A toll-free or out-of-area number on a neighborhood restaurant reads as a call center to exactly the customers you want.
Then check what the number looks like from the outside before you print it. New numbers frequently arrive with a "Spam Likely" flag inherited from whoever held them last, and the fix takes weeks of registration work with the carriers. Test it by calling your own cell from it. The remediation path is covered in fixing "Spam Likely" caller ID, and the more permanent answer is branded calling.
Sixty to thirty days out: where the number lives
This is the phase people skip, and the one that produces the "why is nobody calling us" question in month two.
The number has to be identical everywhere it appears, in the same format, attached to the same address. Search engines and voice assistants treat inconsistent listings as evidence that they are not sure which business you are. Your Google Business Profile is the highest-value listing and takes the longest to establish, so file it as soon as you have a signed lease and a projected open date.
The list worth working through, once, in a single sitting:
- Google Business Profile, Apple Maps, and Bing Places, with the same number, address abbreviation, and business name on all three.
- Your own website header, footer, contact page, and the schema markup that feeds assistants.
- Every delivery marketplace profile, including the ones your operator set up for you.
- Printed menus, signage, receipts, and packaging, which are the versions you cannot change later.
- Your POS account and any SMS confirmations it sends, so the callback number on a text matches the number on the door.
Do this before printing anything. The reason is not tidiness. It is that a mismatch between your printed menu and your Google listing sends some percentage of callers to a dead number, and you will never find out which ones.
Thirty to seven days out: what happens when it rings
Now decide who or what answers, which is a real staffing decision rather than a technical one.
A new location is understaffed by design in its first month. You hire to projected volume, projected volume is a guess, and the guess is wrong in both directions on different days. The phone is the first thing that gets dropped when the room is full, and in a new restaurant the room being full is exactly when a first-time caller is deciding whether you are a place that answers.
If you are putting a voice agent on the line, build it against the real menu, not the design menu. Those differ more than you expect at opening, because your chef will cut items during the last month of training. Load real hours, real prep times, real delivery boundaries, and a handoff rule for anything the agent should not attempt. Our onboarding checklist is the long version, and setup on a new line is typically same-day rather than a project.
Two configuration items matter more at a new location than at an established one. Quoted pickup times should start pessimistic, because a kitchen in week one is slower than the same kitchen in week eight and a blown promise on a first order is expensive. And the FAQ layer should carry the questions specific to being new: where exactly you are, whether you are open yet, whether you deliver to a given neighborhood, and whether you are the same owners as the place that used to be there.
Soft opening: the week that tells you what you got wrong
Turn the phone on for friends and family. Let it take real orders from real people who will forgive you.
Then read every transcript. Not a sample, all of them, for the first week. It is the only week where the volume is low enough to do that and the information is dense enough to be worth it. You are looking for three things: questions nobody anticipated, menu items callers say differently than you wrote them, and quoted times that did not hold.
Fix those daily rather than weekly. A new location's call handling improves faster in its first ten days than in the following three months, because the errors are dense and obvious.
After opening: the number that tells you if any of this worked
Pull answered-call rate by hour for your first full week and compare it against your POS covers by hour. If your busiest hour is also your worst answer rate, the phone is losing business at precisely the moment you have the most demand, which is the standard failure and the one described in how many calls does your restaurant miss.
If you are opening store number two or five, do this once and template it. Per-location numbers, one shared configuration, per-location hours and menus, and a single place to see call data across all of them. The multi-site version is in voice AI for multi-location groups.
The test that tells you the setup is done: hand your phone to someone who has never seen the restaurant, ask them to order a specific item with a modification for pickup in an hour, and watch whether the ticket that lands in the kitchen is the one they intended. If it is, you are open. If it is not, you found the defect on a Tuesday instead of on a Saturday.