Choose Maple Voice ($150 a month, or $85 billed yearly) if you only want the phone answered, with hours, FAQs and transfers to staff. Choose Maple Pro ($350 a month, or $220 billed yearly) when callers also place orders or book tables, because only Pro adds ordering, reservations and point-of-sale links.
The gap between the plans is $200 a month on monthly billing, or $135 on yearly. Whether that pays depends on how many calls turn into orders you would otherwise lose. Maple's homepage and plan page are the main results for this search, and they describe the product rather than the decision. A pizzeria that misses 15 order calls a week and a café fielding "are you open?" calls need different plans, and a one-week call tally settles which you are.
What each plan includes, from Maple's pricing page
This comes from Maple's pricing page, read in October 2026. It is the vendor's own list.
| Maple Voice | Maple Pro | Maple Enterprise | |
|---|---|---|---|
| Monthly billing | $150 a month | $350 a month | Custom |
| Billed annually | $85 a month ($1,020 a year) | $220 a month ($2,640 a year) | Custom |
| Answers the phone 24/7 | Yes | Yes | Yes |
| Hours and FAQ answers | Yes | Yes | Yes |
| Transfers a call to staff | Yes | Yes | Yes |
| Call analytics, multilingual voice | Yes | Yes | Yes |
| Phone orders into your POS | No | Yes | Yes |
| Book, change, cancel reservations | No | Yes | Yes |
| Upselling, card payment by phone, confirmation texts | No | Yes | Yes |
| Multi-location controls, dedicated manager | No | No | Yes |
Two details are easy to miss. Maple says it has no per-minute fees, so the monthly price is not a meter. And drive-thru and kiosk are separate early-access products priced per unit, which sit outside both plans.
The decision in one question: what do callers want?
The plan names suggest a quality ladder. It is not one. Voice and Pro answer calls equally; the difference is what happens after "I'd like to order". With Voice, the caller hears the information or is transferred to a person. With Pro, the AI takes the order and writes it into your point of sale. So the question is not how good you want the AI to be. It is what the next step of a typical call is.
Keep a tally sheet for one week with five columns: hours or address question, menu question, takeaway order, table booking, other. Mark each call, and add a sixth column for "missed or abandoned". Two things then fall out:
- If most calls are the first two types, Voice fits.
- If a good share are orders or bookings, and some of those are missed in the rush, Pro has something to earn.
The tally takes ten minutes a day. It beats any estimate, because owners tend to remember the angry call and forget the 30 calm ones.
Break-even sums at three kinds of restaurant
Maple's own page suggests dividing the plan price by what you keep from an extra order after its variable costs. Its example: at $10 kept per extra order, $350 needs 35 extra orders a month to cover itself. Here are three illustrations with that method. All figures are invented for the sum, not forecasts.
A takeaway-heavy pizzeria
Say a small pizzeria gets about 30 phone calls on a Friday and Saturday evening, and the staff miss around 8 of them while hands are in dough. Suppose half of the missed calls are people who would have ordered and then ordered elsewhere. Four orders a night, two nights a week, about 35 a month, at roughly $9 kept per order after food and packaging.
- Recovered margin: 35 x $9 = $315 a month.
- Pro monthly: $350, a loss of $35. Pro yearly: $220, a gain of $95.
- Voice does nothing for this problem, because it cannot take the order. It would answer and tell people to call back.
Result: on monthly billing it is marginal, yearly it is positive, and the real gain depends on how many missed calls are real orders. That is exactly what the tally sheet tells you.
A café that gets "are you open?" calls
A café gets about 40 calls a week, nearly all asking opening hours, whether a dish contains nuts, or whether a pushchair fits. Orders are walk-in. No call is worth a sale. Voice is the match at $85 to $150 a month, but a cheaper fix may exist: updating hours on the online listing and adding an answer machine line stops many of these calls. Voice is worth it only if the calls interrupt service. A simple test: if someone is pulled off the coffee machine to answer more than five calls a day, it earns its fee.
A table-service bistro on a booking system
A bistro takes most bookings through an online system and gets calls for last-minute tables and large groups. Maple lists reservations on OpenTable under Pro. If the bistro uses a different booking system, ask whether that system is supported before taking Pro for reservations, because "every POS integration we support" is not a promise for yours. For this bistro, Voice that transfers the group enquiries to the manager may be the cheaper and safer step.
Monthly or yearly: the money, and the risk
Yearly billing is cheaper per month. Maple's toggle advertises "save 43%", which is the Voice figure; on Pro the yearly price works out at about 37% off. The arithmetic is simple:
- Voice: $150 x 12 = $1,800 monthly, against $1,020 yearly. Saving: $780.
- Pro: $350 x 12 = $4,200 monthly, against $2,640 yearly. Saving: $1,560.
The price of the saving is commitment. If the AI mishandles your menu and you leave in month three, a yearly plan may mean paying for months you do not use. I would not choose yearly until two things are settled in writing: what the cancellation and refund terms are, and what a trial or pilot period looks like. Ask for a short monthly start, then switch to yearly once the test calls pass. If Maple will not do that, price it as a yearly commitment and tell yourself it is $2,640 you must earn back.
Compare the same terms when you look at other vendors. A cheaper monthly price with a $299 setup fee, or a plan that bills per minute, is a different cost even if the headline looks lower. The general picture is in AI phone answering for restaurants: costs and payback.
A test-call script to run before you pay
Ask for a demo or pilot line, then call it ten times with your own menu loaded. Use a mix of ordinary and awkward requests. Write down each result as pass, partial or fail. A realistic script:
1. "What time do you close on Sunday?" (hours from your listing)
2. "Do you have anything gluten-free?" (should not guess; should transfer or give your written statement)
3. "Can I get a large margherita and a Caesar salad, no croutons?" (modifier)
4. "Actually, make that two large margheritas." (change mid-call)
5. "I'll pay when I collect." (payment choice)
6. "Can I speak to the manager?" (transfer works, rings a real phone)
7. "Table for six tomorrow at 7." (booking, or a clean hand-off on Voice)
8. A call in a second language your customers use.
9. A call with background noise, from a car or a loud room.
10. A caller who says nothing for five seconds, then asks something unrelated.Then look at what reached the kitchen for the ordering calls, not just what the AI said on the phone. Maple's own advice is similar: try an order with a required choice and an item change, inspect what lands in the kitchen, and test a transfer to staff. Item 2 is the most important. Allergen questions need an answer you wrote and approved, or a person. See whether an AI chatbot should answer allergen questions for the reasoning.
Reading a pilot: a filled-in scorecard
An illustrative pilot for the pizzeria, ten test calls plus a week of real calls:
| Check | Result | Action |
|---|---|---|
| Hours and address answers | 10 of 10 correct | None |
| Order with one modifier | 4 of 5 correct; one pizza arrived without the extra topping | Fix the modifier mapping in the menu import |
| Mid-call change | 2 of 3 handled | Raise with Maple; hold off yearly billing |
| Transfer to staff | 3 of 3 rang through | None |
| Allergen question | Transferred to staff | Correct behaviour |
| Real orders in week one | 31 taken, 2 needed a call back | Count the extra orders against the missed-call tally |
A scorecard like this shows where the money is and what to fix before signing for a year. A failure in the mid-call change is not fatal on a monthly plan; on a yearly plan it is a reason to negotiate first.
A week of calls, tallied: a filled-in sheet
Here is what the tally sheet might look like for a neighbourhood pizzeria after seven days. The numbers are invented to show the method.
| Call type | Calls | Answered by staff | Missed or abandoned |
|---|---|---|---|
| Hours or address | 41 | 37 | 4 |
| Menu or allergen question | 22 | 19 | 3 |
| Takeaway order | 96 | 81 | 15 |
| Table booking | 9 | 8 | 1 |
| Other (suppliers, wrong number) | 12 | 10 | 2 |
| Total | 180 | 155 | 25 |
Read it from the right. Of 25 missed calls, 15 were people trying to order. If 60% of them would have ordered and none rang back, that is 9 orders a week, or roughly 39 a month. Take that at $9 kept per order: about $350 of margin a month, which is exactly the price of Pro on monthly billing. On yearly billing at $220 it clears with $130 to spare. The 41 hours and address calls are real work for staff, but they are not a reason to pay for Pro; they are the Voice problem, and Voice would take them for $85 to $150.
Notice what the sheet protects you from. Without it, the owner might have concluded that 180 calls a week "must justify" an AI line. With it, the justifying number is 15, and the decision is between a plan that cannot take orders and a plan that is break-even on monthly billing.
What a good ordering call sounds like, and what you would fix
During the pilot, read the transcripts as well as the order tickets. This is an illustrative exchange with the notes a restaurant owner should make beside it.
AI: Thanks for calling. Would you like to order for collection or delivery?
Caller: Collection. A large pepperoni and a garlic bread.
AI: Large pepperoni and one garlic bread. Any extra toppings?
Caller: Can you do half mushroom?
AI: I can add mushrooms to the whole pizza for $1.50. Half and half is not
something I can set up on the phone. Would you like mushrooms on the whole
pizza, or shall I put you through to the team?
Caller: Put me through.
AI: Of course, one moment.This is a good result for a clear reason: the AI said what it could not do and handed over rather than guessing. What to fix: a half-and-half pizza is a common request, so either add it as a menu option or write a rule that sends it to staff straight away. Each time the AI says "I cannot", you learn a gap in the menu setup, and fixing those gaps in the first two weeks is where most of the quality gain comes from.
A bad result looks different. If the AI had answered "Yes, half and half is fine" and the ticket arrived with mushrooms on the whole pizza, the failure is silent: the customer is cross at pick-up, and nobody connects it to the phone system. That is why the scorecard above checks what reached the kitchen.
The upsell line, and whether to leave it on
Maple lists intelligent upselling under Pro. It is the part of the plan that pays for itself or irritates, depending on how it is set. Think in numbers. If 80 phone orders a month are offered a $3 drink and 15% say yes, that is 12 drinks, about $36 of revenue and perhaps $22 of margin. Useful but small. If the upsell is clumsy and one regular in twenty is put off, the loss on a repeat customer outweighs it. My approach is to turn it on for one item only, usually a drink or a dessert, watch the transcripts for two weeks, and turn it off if you hear a single call where the offer feels pushy.
Saving staff time versus winning orders
There are two different ways a phone AI pays back, and the plans split along them. Voice mostly saves time. Pro mostly wins orders. Put a value on each separately.
- Time saved (Voice and Pro). Say 60 calls a week are hours, address and simple menu questions at about two minutes each. That is 2 hours a week, 8 to 9 hours a month. At $15 an hour of staff time, about $130 of effort. That is not a cash saving unless you cut a shift, but it frees people during the rush.
- Orders won (Pro). The missed-call sum above, using margin kept rather than revenue.
- Errors avoided or created. A wrong order costs a remake, a refund and sometimes a customer. Count remakes before and after the pilot.
If the time-saved number is the only one that adds up, Voice is the honest choice. If the orders-won figure adds up, Pro is. If neither does, wait.
Five mistakes owners make with phone AI
- Loading an old menu. A phone assistant that quotes last season's prices will honour them in the customer's mind. Check prices, specials and sold-out items on the day you go live and after each change.
- Letting it answer allergen questions from the general menu. A written, approved allergen statement or a transfer is the safe route.
- Forgetting the busy-night fallback. Decide what happens when all staff lines are busy and the AI transfers: does the call ring out, go to voicemail or return to the AI?
- Skipping a spoken disclosure check. Some customers dislike talking to an AI without being told. Listen to how the greeting introduces itself and judge it as a customer would.
- Judging it on easy calls. Ten hours questions in a row prove nothing. The test is the order with a modifier, the change of mind and the muffled mobile line.
Other options exist if Maple does not fit your till or budget. The comparison of Yelp Host, Slang, Popmenu and Maple sets them side by side, and AI restaurant order taking shows the choice between an AI that takes the order and one that texts the caller a link.
When each plan is the wrong choice
- Voice for a takeaway business. The AI will answer, then it cannot take the order. You pay $85 to $150 a month and still miss the sale. Either upgrade to Pro or do nothing.
- Pro without a supported POS. If your till is not on the supported list, orders may have to be re-keyed by hand. Ask for a written statement of how an order reaches your kitchen.
- Either plan if the calls are rare. Under about ten calls a day, a decent voicemail greeting and a text-back message cost almost nothing.
- Enterprise for one site. It is priced for multi-unit groups.
- Drive-thru. Early access and per-unit pricing means a sales conversation. Plan on a pilot at one lane.
Before you sign: terms to get in writing
Voice AI sits between you and every customer who rings, so the contract matters more than the monthly price. Ask for:
- Whose phone number it is, and how it moves back to you if you leave. Protect your number, data and exit goes into this in detail.
- What happens to call recordings and transcripts: where they are stored, for how long, and who can hear them. Callers are often not told they are talking to an AI; check what disclosure Maple provides and whether you need to give a notice.
- The fallback if the AI is down: does the call forward to a staff phone?
- Cancellation terms for monthly and yearly billing, and any minimum term.
- Any charges beyond the plan: payment processing, texting, setup or integration fees.
If you want to lay these out against your own call tally and till setup before you commit, that is the sort of job an AI implementation consultation covers (contact). Otherwise: tally a week of calls, run the ten-call test, start monthly if you can, and move to yearly only when the scorecard is clean.
Maple Voice and Pro: questions restaurant owners ask
How much do Maple Voice and Maple Pro cost?
On Maple's pricing page in October 2026, Voice is $150 a month, or $85 a month billed annually. Pro is $350 a month, or $220 a month billed annually. Enterprise is custom for multi-location groups. Maple states there are no per-minute fees, but confirm payment-processing, setup and any third-party costs with them.
Can Maple take reservations on the cheaper plan?
No. Maple describes Voice as answering, hours and FAQ answers, and call transfer, not ordering or reservations. Pro adds phone orders into a supported point of sale, and reservations including OpenTable. If callers mostly want a table or a takeaway order, Voice will answer them and hand over, which may not be enough.
Does Maple work with my till?
Maple lists Toast, Square, Clover and Shift4 Dine among supported systems, with others available. Support differs by system and feature, so ask Maple to confirm your exact point of sale, your menu's modifiers and how an order reaches the kitchen before you buy.
Is drive-thru available?
Maple says drive-thru and kiosk are in early access and priced per unit, outside the Voice and Pro plans. Treat them as a conversation with sales, not a switch you can turn on from the plan page, and ask for references from a similar site.
Further reads
- Restaurant AI Voice Agent: 12 Test Calls to Run Before Go-Live — A fuller set of test calls to run before go-live.
- Square AI Voice Ordering for Restaurants: Set Up and Test (Beta) — Square's own voice ordering, if Square is your till.
- SevenRooms Voice AI: What It Does and Who It Suits — A booking-led alternative for table-service sites.
- Popmenu AI Answering: Does It Take Orders or Only Send Links? — Whether Popmenu's answering takes orders or sends links.
- AI Chatbot for Restaurants: Hours, Bookings and Hand-Offs to Staff — A chatbot route if phones are not your problem.
- Can AI Answer the Phone and Take Restaurant Reservations? — How reservations by AI voice really work.
- Is Yelp Host Worth It for a Small Restaurant? — An independent cost-and-listing verdict on Yelp Host: the price, the break-even sum, and what your Yelp page and POS must have in place.
- Slang.ai for Restaurants: Is It Worth It for an Independent? — Slang's Core and Premium tiers, the booking systems it connects to, private-events capture and a break-even sum for an independent.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Maple's pricing page (maple.inc), read in October 2026. Plan prices, billing terms and early-access products change; confirm on Maple's own page and in a written quote.