Loopa reads the ticket, writes the reply and works out the compensation by your rules. In the scenarios you have approved it closes the case on its own — bonus included. Everything else arrives ready: read it and press send.
Loopa is a platform: venues connect their own channels and accounts to it. Cappi, our own chain, is client number one.
Same guest, same complaint, same pizza. The only difference is how many minutes he waited for an answer.
A typical ticket takes five steps: grasp the issue, find the rule, reply, credit compensation, set the status. Loopa does all five. It calls a human only when there is no other way — and hands over a case with the analysis done, not a link.
Tickets from connected channels land in one inbox. The system sees the content, the context and the type of situation.
For a typical case Loopa runs the allowed scenario itself — not a “handled” flag, but an actual resolution.
When a situation doesn't fit the rules, the case goes straight to the owner — with context, not a link.
Identified the case type, credited compensation within the rules and prepared the reply to the customer.
Loopa works inside the limits you set: scenarios and replies, the compensation cap, blocked topics and tone of voice. Every action is logged — what it did, when, and under which rule. The team sees not how much was collected, but what is still open.
The card shows what Loopa did itself, which rule it applied and where the team needs to step in.
Every reply you approve becomes Loopa's reference. It handles similar situations more accurately next time — and the share of tickets closed with no operator grows without new settings.
Analytics shows where the work still hits a human:
Action without measurement is faith. Measurement without action is a report. Loopa counts not how many reviews were collected, but whether guests are better off — from every closed case, with no separate surveys.
Cappi is our first client and our proving ground: we run on Loopa every day ourselves. We don't know what your numbers will be. The audit shows the starting point on your own tickets — and from there you watch it move every week.
Kitchen, delivery, packing, service.
Complaint, thanks, request, or our own initiative.
How irritated the guest is — that decides whether to call a human.
What to handle now and what can wait until morning.
Across 498 tickets people made 33 corrections. Each one is logged separately: what the model said and what it was changed to. The list of what it confuses most often sits open in your dashboard — not in our report, in yours.
Loopa is rarely compared to another service. It's compared to how things work right now. So the table has three real options, not an invented competitor.
| Who does it | A manager, by hand | A review collector | An outsourced operator | Loopa |
|---|---|---|---|---|
| Reads every channel | Google and Instagram, when there's time | Google and Facebook, sometimes TripAdvisor | Only what was forwarded to them | Every connected channel, around the clock |
| Replies to the guest | Whenever they get to it — often the next day | A template someone has to paste in by hand | From a script, during office hours | By your rule and tone of voice, within minutes |
| Credits compensation | Asks the owner in a messenger | Can't do it | Has no access to the points | On its own, within your cap |
| Assigns a task to the shift | Verbally, if they don't forget | Can't do it | Passes it on and stops there | A task with an owner and a deadline |
| Checks whether the guest returned | No | No | No | Comes back to the case after the number of days you seton the roadmap |
| What you see at the end of the month | Nothing but the rating on maps | The number of reviews collected | A call report | Closed cases, causes, and the share of work done without an operator |
We count tickets, not invitations sent. So a 500-ticket plan is 500 real tickets, not 500 invitations that fifty people answer.
Over your plan — we don't block your work and we don't charge extra. If it happens a second month in a row we'll suggest the next plan — but only with your consent; nothing changes automatically. −20% for annual billing. In Ukraine, invoiced in hryvnia at the NBU rate. Prices exclude VAT.
We connect the first channels in 2 hours. No card. You don't need POS access to start.
We reply the same day. You pay nothing until you see the result on your own tickets.
We'll be in touch today — we'll call or message you on Telegram at the number you left. Nothing else to do.
For a typical scenario you've approved, Loopa reads the ticket, replies to the guest, runs the allowed action and records the outcome — without your sign-off on every step. Unusual, expensive or sensitive cases go to a human. The full list of what it never does on its own is in the question about agent mistakes below.
A POS collects reviews and lets you configure loyalty rules — that's configuration. Loopa decides who to write to and what to say, does it, assigns a task to the location and verifies the outcome — that's action. We don't replace your POS and we don't fight it: we run on top of what you already have and read the channels a POS has no access to at all — Direct, the call centre. From your data the system can surface extra signals — cancelled orders, for instance. That's a separate layer, not a requirement for the product to work.
No. You can start with the channels where tickets get lost most often and connect the rest gradually.
The price is published above: from €49 a month for Start, covering a single venue or two locations, up to €349 for Chain. Billing is monthly and you can cancel in one click. There is no setup fee and no limit on users.
The plan depends on the number of locations and the ticket volume, not on how many people are on your team. The whole team gets access, including the floor staff.
Nothing happens: we don't block your work and we don't charge extra. You'll see it in the dashboard and we'll write to you. If it happens a second month in a row we'll suggest the next plan, and only with your consent. We never change or charge anything automatically.
A ticket is one case that has to be worked through: a complaint, a request, a rating with a comment, a negative review. One story with one guest, even if it contains several messages.
We don't count invitations sent, empty ratings with no text, or system events as tickets. So a 500-ticket plan is 500 real cases, not 500 invitations that fifty people answer.
Google Maps, Facebook and Instagram connect within 2 hours, and the first ticket closes in that same window. Those are the sources that switch on immediately; Telegram, TripAdvisor and Bolt Food are still in development. POS and call-centre integration takes longer and runs in parallel, without stopping your work.
Ready-made integrations are included in every plan: Google Maps, Instagram and Direct, Facebook, Telegram, TripAdvisor, QR in the venue, website, app, call centre. There's no separate charge for them.
We connect your own channel through an open API or webhooks — on the Business plan and above. If your system exposes the data, it's configuration, not development: usually a few working days and at no extra charge.
We charge separately only for what genuinely requires development on our side: a closed system with no API, file exchange, a non-standard protocol. In that case we name the timeline and the price before the work starts, not after.
Autonomy is on from day one, but not for everything. Loopa closes only the typical scenarios you've approved: late delivery, a cold dish, an incomplete order, a long wait in the venue, a thank-you, a booking or menu question. Anything outside that list goes to a human — with the analysis done, not a link.
Public replies — on Google Maps, TripAdvisor and in comments — sit in a 10-minute queue. The manager gets a notification and can edit or cancel the reply before anyone sees it. If nobody steps in, it publishes itself. In private channels (Direct, Telegram, website chat, call centre) the agent replies immediately.
There are topics it never takes, even if you allow it: poisoning, allergy, injury, a foreign object in a dish, a threat of a lawsuit or publication. Those go to a human instantly and with priority.
It never awards compensation beyond your cap. Autonomy can be paused with a single button — for one location or the whole chain.
Yes. The Start plan is built exactly for one venue or two or three locations: 200 tickets a month, every source, metrics and cancellation analysis. Automatic compensation switches on from Business, but closing tickets and replying to guests works from the start.
At the audit we'll show the numbers for your venue specifically. If it's clear the effect is small for now, we'll say so plainly.
The data is yours. We don't use it for other clients and never pass it to third parties. If you leave, you take a full export with you. Storage meets personal-data requirements; the details are in the contract.
No. We invite guests who have already told us they were happy to leave a review, and we do it within Google's rules. The rating grows here as a consequence of happy guests becoming audible and unhappy ones no longer being left without an answer. Inflation gives a short jump and a risk of platform sanctions; we have no need for it.