Voice ordering for restaurants: what happens between "two burgers" and the cart

Two kinds of voice ordering

Most pages on this topic blur two products together. One is a guest speaking to your online ordering site, and the other is a caller speaking to your restaurant's phone number. On the site, the guest sees the menu and the cart while they talk. A caller sees nothing. The system has to read every choice back aloud, and a mistake costs a wasted trip to the counter. That gap decides what each version can safely do, so we built two features with separate limits.

• Storefront voice: the guest taps a microphone in the assistant and talks. The browser turns speech into text, and the text becomes cart lines.
• Phone voice: Sous answers the call, takes a pickup order, and sends it like any direct order. See AI phone ordering.
• Both end in the same place: an order on your own system, not a marketplace's.

A worked order, line by line

Take a guest who says: "two cheeseburgers, one with no onions, and fries." A text-only assistant would answer with a card the guest has to tap. Voice order mode builds the cart instead. The guest's sentence becomes three separate lines, because a line with a note never merges into a line without one.

• Line 1: one cheeseburger, no note.
• Line 2: one cheeseburger, note "no onions".
• Line 3: one order of fries.
• If the burger has a required choice, such as a bun, the assistant asks first and adds the line after the answer.
• Each line shows as an Added card with an Undo, and the cart badge updates as it happens.

What checks each spoken line

The risk with voice is a made-up item or a wrong price. Three checks run on the server before a line reaches the cart. Without them, a spoken order could not touch the cart at all, which is how the typed assistant works: it shows an order card and the guest taps Add. Voice mode adds these checks so the cart can fill itself.

• The quote check: each line must quote words the guest said in this conversation, or it is refused.
• The menu check: the quote must also match the item's name on your menu.
• The price check: the line is priced from your menu. A line with a required choice still open is refused until the guest answers.
• A voice turn is never served from a cache, so one guest's order cannot repeat for another.

What voice ordering never does

The limits matter more than the demo. Voice mode never starts checkout, quotes the cart, submits the order, or calls a payment route. The guest reviews the cart and pays with a tap, so a misheard word costs one Undo instead of a wrong order at the counter.

• No audio leaves the guest's phone. The browser owns the microphone and nothing is recorded or uploaded.
• No spoken replies. The assistant answers in text.
• It works in browsers with speech recognition, such as Chrome and Safari. Where a browser lacks it, such as Firefox, the microphone is hidden and typing still works.
• You can switch it off, and it costs nothing extra.

What your menu needs for voice to work

Voice ordering is only as good as the menu behind it. The assistant reads your live menu, so a sold-out item, a price change, or a new modifier shows up in the next conversation with no extra step. The part owners overlook is required choices. If a burger needs a bun and the menu marks the bun as required, the assistant asks "which bun?" before the line lands in the cart. If the menu leaves the choice optional, the assistant has no reason to ask. So the setup work is the same work that makes any online menu accurate. Nothing here is voice specific, and that is the point: the menu you already maintain is the menu the assistant reads.

• Mark every choice a kitchen needs as required: bun, protein, size, spice.
• Keep item names the way guests say them. "Cheeseburger" beats "The Classic No. 2" for a spoken match.
• Keep prices current, because every spoken line is priced from the menu at that moment.
• Connect your POS if you have one, so prices and hours match the register. See POS sync.

Six questions to ask any voice ordering vendor

Demos of voice ordering look alike, because the happy path is easy. The differences show up in the questions below. Ask them on a live store, not in slides, and ask what happens on a misheard word rather than a clean one.

• Who sets the price on a spoken line? The answer should be the menu, never the model or the guest.
• Can the assistant check out on its own? For a food order, the guest should confirm and pay.
• What happens when the guest says an item you do not sell? It should refuse, not substitute.
• Is audio uploaded or stored? Ask where the speech turns into text.
• Which browsers and phones does it work on, and what do the others see?
• Can you switch it off for a location, and what does it cost to run?

Try it on a live store

You can hear it take an order on the Royal Burger demo store. Two buttons start it. "Watch it" plays a scripted guest order, with the browser speaking each line while the words appear and the real assistant answers. "Try it" uses your own microphone. The demo orders nothing and takes no payment, and the lines it adds leave the cart when the demo ends or starts over.

When voice is not the answer

Voice is a convenience, not a replacement for the menu. A guest in a loud room, or one who just wants to scroll, will tap. A guest with a serious allergy should talk to a person, and the phone agent transfers any allergy or dietary safety question to your staff for that reason. Voice also does nothing for the guest who is not on your site yet, which is a search and listing problem. If your Google listing or ordering link sends guests elsewhere, fix that first with the free audit.

When the phone is the better door

Storefront voice helps a guest who is already on your site. It does nothing for the call that comes in during the rush. For that, the phone version answers in the first sentence that it is automated, takes a pickup order, and sends it to your kitchen. It pays at pickup only, passes allergy questions to your staff, and hands the call to a person on request. Voice is included in Growth at no extra charge, with 1,000 minutes a month per location. Read the phone ordering demo checklist before you judge any vendor's call.

Voice ordering FAQs

• Does voice ordering for restaurants need special hardware?: No. The storefront version uses the guest's own phone or computer and its built-in speech recognition. The phone version uses a number we provide for each location.
• Can a guest check out by voice?: No. The assistant adds lines to the cart. The guest taps to pay.
• What happens when it mishears?: The wrong line shows as an Added card, and the guest taps Undo. A line the guest never said is refused before it appears.
• Is audio stored?: On the storefront, no audio is sent or stored. On the phone, the call report counts calls and topics only, with no recordings or transcripts.

Next steps

Hear the storefront version on the demo store, then see how the phone line works on AI phone ordering. To see your own restaurant on Appetier, run the free audit.