Popmenu AI Answering does not take food orders or reservations itself. Popmenu's own pages say it answers questions about hours, location, parking and allergens, and texts the caller a link to order or reserve. So it suits a venue whose guests will finish a booking on their phone; a takeaway living on phone orders needs more.
That one detail changes the maths for a restaurant. A system that takes the order captures it on the call. A system that sends a link only captures it if the caller opens the text and completes the form. The sections below show how to test it, what it costs you when callers drop out, and what else to check before relying on it.
What Popmenu says the product does
From Popmenu's AI phone answering page, read in October 2026, the product:
- answers calls around the clock with custom responses and a choice of voice;
- sends guests a reservation or ordering link during the call, and its FAQ says the AI texts links for orders and reservations;
- answers common questions about hours, location, parking and allergens, and can mention specials and events in a custom greeting;
- forwards calls from your existing number, so guests dial the same line;
- gives owners an app to watch priority calls, see peak hours and time saved, and sort calls into outcomes such as AI Answering Resolved, Voicemail, Spam Blocked and forwarded.
What the page does not say matters as much. It does not describe integrations with a point-of-sale system or a booking tool, how a call is handed to staff, call limits, or languages. I could not confirm any of those, so they are questions for Popmenu, not assumptions. The product also does not appear to collect payment on the call, because the link is what takes the guest to the order or booking page.
The price discrepancy to resolve before you sign
Popmenu's AI phone answering page says "AI phone answering starts at $149 per month" with custom quotes on request. A separate Popmenu demo page for AI Answering lists $349 a month and compares it to an always-on employee costing 47 cents an hour. I cannot tell from the pages whether those are different bundles or just different dates. Ask for a written quote naming what is included, the contract length, and whether the fee depends on call volume. Until then, plan with a range of $149 to $349 a month, and count the higher figure when you work out whether the system pays back.
There is a second line to price. Popmenu's main pricing page lists AI phone answering as a "popular add-on" to its restaurant website plans, not as a plan of its own: Starter is $159 a month prepaid annually or $179 monthly, Essentials $269 or $299, and Premier $449 or $499. Online ordering is listed as a separate add-on too. So a venue not already on Popmenu should ask three things: can AI Answering be bought without a website plan, which plan it needs, and whether the ordering link points to Popmenu's own ordering add-on or can point to the ordering page you already use. A rough illustrative total for a new customer on Starter with the add-on is $159 plus $149, about $308 a month on annual billing, before any ordering add-on.
What the caller actually experiences
A call to a link-sending system runs in stages, and each stage is a place to lose the guest:
- The AI answers and greets the caller, possibly with tonight's special.
- The caller asks to order or book. The AI says it will text a link.
- The caller must be on a phone that receives texts and must open the message.
- The link opens a page. If it is an online ordering page, the caller builds the basket, enters an address or collection time, and pays.
- If any of that is awkward, the caller either calls back, rings a competitor, or gives up.
Between step 2 and step 4 you have no say. You can't see the caller, you can't help with a question, and you don't know they dropped out. The only evidence is the gap between link texts sent and orders received. Popmenu's owner app lists call outcomes, so ask which report shows how many guests followed the link.
A test-call script that shows what happens
Ring the line yourself from a mobile and from a landline. Make about ten calls and score each. The calls below cover the situations a real service throws up.
1. "I'd like to order two large pizzas for collection." -> Link sent? How long?
2. Same call from a landline. -> Fallback? Transfer?
3. "Is there a table for four at 8 tonight?" -> Link or answer?
4. "Do you open on bank holidays?" -> Correct hours?
5. "Is the garlic bread nut-free? My daughter has an allergy." -> Does it transfer?
6. "I ordered 20 minutes ago and nothing has arrived." -> Transfer to staff?
7. "Can I change my order?" -> Sensible handling?
8. Ring at 3 am. -> Greeting, link, voicemail?
9. Ask in a noisy room, with an unusual name. -> Understood?
10. Hang up halfway and call back. -> New greeting or memory?Note the time between the call and the text arriving, the sender number, and whether the link opens a page that works on a small screen. A text that takes two minutes to arrive is a text that arrives after the caller has rung somewhere else.
An illustrative result for a neighbourhood restaurant: calls 1, 3, 4 and 7 pass; the text for call 1 arrives in 15 seconds; call 2 on a landline ends with "I've sent you a link" and nothing arrives; call 5 gets a confident allergen answer with no transfer. Those last two are the findings that matter, and a vendor should be asked to fix both before you go live.
The leak between link and order
Take a takeaway that gets 120 phone orders a week. Each is currently handled by staff, who miss around 15% at peak and lose them. Say the venue switches to link-sending and the AI answers every call. The question becomes how many callers finish the form.
Suppose 70% open the link and 80% of those complete the order. That is 56% finishing, or about 67 of 120. Staff used to capture 85%, about 102. Under those assumptions the AI loses 35 orders a week against the human baseline, though it removes the 18 missed calls. At a $30 ticket and 55% kept, each lost order is $16.50 of profit, so 35 lost orders is about $578 a week. It would take a very high completion rate to beat the old process.
Now try a restaurant with mostly reservation calls. Staff book 50 of 60 daytime callers and miss 10 at busy times, and another 20 callers a week ring after closing and reach voicemail, where perhaps 5 leave a usable message. Today that is 55 bookings from 80 callers. With a link, say 75% of all 80 follow through, which is 60 bookings, with no staff time on the phone and a text in the guest's hand at midnight. That venue gains a few bookings and a lot of time. The difference between the two venues is the product's design: a link suits a booking, where the caller is willing to pick a time, and it is harder on a food order, where callers often want a human to confirm toppings and times.
These percentages are invented. Substitute real ones: ask Popmenu or your online ordering provider for the click and completion rates, or count them yourself over a week.
Where it fits, by type of venue
| Venue | Main phone job | Does link-texting fit? | What to add |
|---|---|---|---|
| Table-service restaurant | Bookings, opening hours, allergen questions | Often yes, if the booking page is easy | Human transfer for allergies and groups |
| Takeaway or pizza shop | Phone orders for collection and delivery | Weak: each order needs a form | A system that writes orders into the till |
| Café with lunch pre-orders | Pre-orders and queries | Mixed: depends on the ordering page | A short URL on the menu and window |
| Venue with private hire | Enquiries for events | Poor: needs a conversation | Enquiry form plus a callback rule |
If your venue is the second row, look at systems built to take orders. Maple's Pro plan, for example, lists phone orders into a point-of-sale system at $220 a month on annual billing or $350 monthly. The comparison of AI order taking and link texting sorts the main restaurant options on this exact question, and the tutorial on monthly AI phone costs for a pizza shop turns it into a payback sum.
How the alternatives compare on the same question
The useful comparison is not price but what happens on the call. The table uses only what each vendor's page said when I read it in October 2026.
| Option | List price (USD) | Takes the food order on the call? | Reservations |
|---|---|---|---|
| Popmenu AI Answering | From $149 a month on one page, $349 on another; sold as an add-on to a Popmenu plan from $159 | No; texts an ordering link | Texts a reservation link |
| Maple Voice | $85 a month annual, $150 monthly | No; answers questions and transfers | Not included |
| Maple Pro | $220 a month annual, $350 monthly | Yes, into a POS such as Toast, Square, Clover or Shift4 Dine | Books, changes and cancels |
| Goodcall | $79, $129 or $299 a month by unique customers | Page does not mention orders | Not stated on the pricing page; ask |
| Voice platform build (Retell or Vapi) | About $0.08 to $0.13 a minute | Only if you build the order logic and till connection | Only if you build it |
Notice how little price tells you. Popmenu's lowest figure is below Maple Pro's annual price, yet Maple Pro takes the order and Popmenu hands the caller a link. For a venue where the phone is the shop window, paying more to capture the order on the call is often the cheaper route once you count the callers who drop out. The four-way comparison of Yelp Host, Slang, Popmenu and Maple covers the receptionist side in more depth.
Scoring a week of real calls
A test script shows what the system can do. A week of live calls shows what it does. Keep a simple log from the first day, even if it is a notes app on the counter. This illustrative week comes from a family-run restaurant with a busy takeaway side.
| Call type | Calls | Link sent | Order or booking seen | Note |
|---|---|---|---|---|
| Wanted to order food | 96 | 81 | 49 | 15 answered by staff directly |
| Wanted a table | 38 | 33 | 27 | 5 transferred |
| Question only | 70 | 0 | n/a | Answered; 6 were wrong or vague |
| Complaint or order problem | 12 | 0 | n/a | 10 transferred, 2 dropped |
The ordering row is where the decision sits. Of 96 callers, 81 got a link and 49 finished an order: a 60% completion from link to order, which is 51% of everyone who called to order. If your staff-only baseline captured 82 of 96 (85%), the system is behind on food orders and ahead on tables. The same week also shows 6 vague answers, which should become edits to the greeting and the menu notes the AI uses, and 2 dropped complaints, which should become a transfer rule fix. None of this appears if you only read a monthly total.
Text-back wording that rescues some of the drop-off
If you control any text sent after the call, make the link message do the work. The first draft below is what such messages usually look like; the second is what I would test. Check with Popmenu what wording you can set, because the page I read does not say.
Before: "Here is the link to order: [link]"
After: "Hi, thanks for calling [restaurant name]. Order for collection
or delivery here: [link]. Ready in about 25 minutes. If anything
looks wrong, call [phone] and a person will help."The second version names the venue (so the text doesn't read as spam), states the time (which answers the question callers usually have next) and gives a way out. If the vendor cannot change the wording, consider a sign at the counter and a line in your voicemail greeting that gives the phone number for orders.
A 30-day trial with a decision date
Run any link-texting product as a trial with an end date, not as a change you drift into. Before you start, write down the number you need to see, such as "at least 55% of link texts end in an order by day 21". Then:
- Week 1: test calls from the script; fix the allergen and landline findings. Go live only on quiet sessions.
- Week 2: route all calls to it during lunch, and keep a log of orders staff would have taken.
- Week 3: compare the link-to-order rate to your target, and read the calls that ended without an order.
- Week 4: decide: keep as is, restrict to after-hours and overflow, or switch to a system that takes the order.
The "overflow only" outcome is often the best one for a takeaway. Staff keep taking calls while they can, and the AI only catches the ones nobody has time for, which is where it adds orders instead of replacing them. The tutorial on AI answering and taking restaurant reservations shows how to set that rule for table bookings.
Allergen answers need a safety rule
The page says the AI answers questions about allergens. That is useful for "Do you have gluten-free pasta?" and risky for "Is this safe for a severe nut allergy?" The answer depends on your kitchen, your suppliers and what changed this morning. An AI repeating a menu note cannot know a supplier changed a recipe.
Write a rule and ask the vendor how to enforce it:
If a caller mentions an allergy, intolerance or medical diet:
1. Do not state that any dish is safe.
2. Say: "I'll put you through to a member of the team who can check."
3. Transfer the call. If nobody answers within 20 seconds, take a
callback number and send an alert to the manager's phone.Test it with the allergy call from the script above. If the system answers anyway, do not go live until it is fixed. If your venue serves customers in the EU, transparency rules on AI systems may also require telling callers they are speaking to an AI, so ask a qualified adviser what your wording needs to say.
Settings and fallbacks to have ready
- A working order page. The link is only as good as where it goes. Open it on a mid-range phone on mobile data and place a test order.
- A short spoken fallback. For landline callers, the greeting should be able to say your web address or put them through to staff.
- Forwarding that can be switched off. Popmenu forwards calls from your existing number; keep the ability to switch back during a rush or an outage.
- A weekly review. Read ten calls a week for the first month. Note which ended with a link, which with a transfer, and which with a drop.
- A sign-off checklist. Hours, special greeting text, allergen rule, holiday dates. Wrong opening hours in a greeting send people to a locked door.
Briefing the team so the link does not surprise them
Staff meet the consequences of a link-sending system at the counter. A guest arrives and says "I ordered on the phone" when they actually received a link and never finished the form. Nobody has the order, and the guest blames the restaurant. Give the team a one-line script and a rule: "If you have no ticket, I'm sorry, the link text didn't complete. I'll take it now and it'll be ready in 20 minutes." Then log the incident. A handful of these in a week is the evidence you need that the link step is leaking, and it is far better evidence than a dashboard total.
Brief the person who closes, too. Overnight calls will have left texts and voicemails, and someone has to check the call outcomes in the owner app each morning, call back the ones marked voicemail, and look at any that ended in a transfer nobody answered. Ten minutes a day is a fair estimate; count it in the cost.
Four mistakes this product invites
Reading "AI answering" as "AI ordering". The name suggests the call is handled end to end. It is not. Read the vendor's own FAQ before you plan a phone-ordering workflow around it.
Forgetting the link needs a home. A guest following a link to a clumsy page abandons the order. Fix the page before the phone.
Assuming a transfer exists. The page I read does not describe how calls reach staff. Test it with a ringing line and confirm the exact behaviour.
Choosing on the entry price. With a $149 and a $349 figure both on Popmenu's own pages, the entry price is not the real price until you have a written quote.
If you want someone to run these test calls and the link-leak sum with you on your own call log, that is part of what my AI implementation consultation covers; the contact page is where to start.
Popmenu AI Answering questions
Why do two Popmenu pages show different prices?
Popmenu's own AI phone answering page says it starts at $149 a month with custom quotes available, while a separate demo page I read lists $349 a month. The pages may describe different bundles or dates. Ask Popmenu for a written quote that names the exact inclusions and any per-call or contract terms.
Can it handle a caller with a landline?
A text cannot reach a landline, so a link-sending system needs a fallback for those callers, such as a transfer to staff or a spoken web address. Ask the vendor what happens when the caller's number cannot receive texts, and test it with a landline.
Is the AI safe to answer allergy questions?
Treat it as a pointer to your published allergen information, never as the final word. A wrong answer about a severe allergy is a safety risk. Set the system to transfer any call that mentions an allergy to a trained member of staff.
What should I use instead if I need orders taken on the call?
Look for systems that write the order into your point-of-sale system. Maple's Pro plan, for example, lists phone orders into a POS at $220 a month annual or $350 monthly. Compare options and test each with the same set of calls before committing.
Further reads
- Slang.ai for Restaurants: Is It Worth It for an Independent? — Weighs another restaurant phone AI for an independent venue.
- AI Phone Ordering for Takeaways: A Test Checklist for Menus and Postcodes — Tests phone ordering on numbered menus and delivery addresses.
- How to Set Up an AI Phone Line for Takeaway Orders — Sets up an AI line that takes takeaway orders.
- AI for Restaurant Order Management: One Queue for Every Channel — Brings phone, web and app orders into one kitchen queue.
- AI Phone Answering for Restaurants: What It Costs and Pays Back — Shows what restaurant phone AI costs and pays back.
- Is Yelp Host Worth It for a Small Restaurant? — Weighs another restaurant phone option for a small venue.
- AI Startups for Restaurants: Protect Your Number, Data and Exit — Protects your phone number and data if you leave a vendor.
- AI Bot for Restaurant Reservations: Chat Booking Setup and Cost — Set up a chat reservation bot step by step: channels, booking-system link, table rules, deposits and what it costs a month.
- Maple AI for Restaurants: Voice or Pro Plan, and Who Needs Which — Maple Voice answers and transfers calls; Pro adds orders and reservations. Compare the monthly cost, break-even orders and the test to run first.
- SevenRooms Voice AI: What It Does and Who It Suits — What SevenRooms Voice AI does, why you must ask for a quote, and how it compares with Yelp Host, Slang and Popmenu.
- AI Software for Restaurants: POS-Built AI or an Add-On Tool? — Compare Toast IQ, Square AI and Lightspeed AI with add-on tools, job by job, with list prices and a checked date.
- Restaurant AI Voice Agent: 12 Test Calls to Run Before Go-Live — Twelve scripted calls with pass marks, blocker or fix labels and a go or no-go rule, so you find the failures before a guest does.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Popmenu's AI phone answering page and its AI Answering demo page (read in October 2026), Maple's pricing page; the call volumes and conversion rates in the examples are illustrative.