Let us say it up front: we are not neutral. Klarr sells a booking page on subscription, with no commission. So we have a direct interest in you finding commissions too expensive.
Read what follows with that in mind, and above all redo the sums with your own figures. That is the whole point of this article: you do not need to believe us, you need a method — and to be able to conclude that a commission is still the right choice for you.
What a commission costs, in euros
The marketplace model is straightforward: they bring you customers, they take a share of every cover booked through them. TheFork, the most used in France, sits between one and two euros per cover — rates taken from the publishers' public pages in September 2026, and they change.
Over a full year:
| Covers / month | Commission over the year |
|---|---|
| 200 | €2,400 to €4,800 |
| 400 | €4,800 to €9,600 |
| 800 | €9,600 to €19,200 |
That is, per month, €200 to €400 for the first line and €800 to €1,600 for the last. Take the one that looks like you. If you do not know your platform volume, it is in your restaurateur dashboard, and it is the first figure to go and find: plenty of restaurateurs discover they are paying the equivalent of a part-time salary.
Note the mechanism, because it is counter-intuitive: it is a cost that grows with your success. A restaurant that fills better pays more. A subscription does not move. The two curves cross at a precise point, and that point is what you are looking for.
The only question that decides
The amount alone says nothing. What counts is the share of those covers that would have come without the platform.
For a customer who did not know you, who was looking for a Japanese restaurant in the 11th and found you there: the commission is the price of an acquisition. It may be dear, it is not absurd.
For your Thursday regular, who knows your name, who would have come anyway, and who booked through the platform because that is the app on their phone: you are paying for a customer you already had.
Nobody can measure that share cover by cover. But you can get close, and three clues are enough to form an honest view:
- Your repeat bookings. The same name coming back three times a year through the platform is no longer an acquisition by the third.
- Your local standing. Type your restaurant's name into Google. If your listing comes up first with your hours and your reviews, people looking for you find you — and if they then go through an app, it is not the app that brought them.
- How long you have been there. A place open six years in the same neighbourhood does not have the same need for discovery as one opening on Tuesday.
The higher that share, the more the commission is funding visibility you had already paid for another way — in years of service, in quality, in word of mouth.
When a commission is still the right sum
This needs saying, because it is true and because the opposite would show: there are situations where paying a commission is the best investment of the month.
An empty room on a Tuesday in January earns nothing. A €45 cover leaving €18 of margin, less €2 of commission, still leaves €16. Sixteen euros beats zero, and no subscription will fill that room for you.
The commission model keeps all its sense if:
- you have just opened and nobody knows your name;
- you are in a tourist area, where tonight's customer will never come back and was not looking for you;
- you have structural quiet periods that no direct channel fills;
- you are aiming at foreign customers who search in an app before searching in Google.
In those cases the right arrangement is not to cut, it is to let the two coexist: the marketplace for what it does well — bringing you strangers — and a commission-free direct channel for everyone who is already looking for you. You reduce the variable part without losing the volume.
What we sell brings nobody. Klarr will not fill your Tuesday in January. It takes bookings commission-free from the people who do find you — that is a different thing, and you need to know which of the two problems is yours.
What to have in place before cutting back
The classic mistake is to cancel first and then wonder where to send customers. A direct channel is not declared into existence: it is prepared, and two things are enough.
A Google listing you look after. It is your first shopfront, before your website. Exact hours, recent photos, the right category, reviews you reply to. An incomplete listing loses places in local results, therefore direct traffic — and mechanically makes you more dependent on the apps. That is the circle to break first.
A booking link in your own name, pasted into that listing. Google then shows a button pointing to your page. It is free, immediate, and depends on nobody's approval.
An honest clarification while we are here, because the opposite is often written: that button is not the “Reserve with Google” programme, which lets people book without leaving the listing and requires a formal partnership. Klarr is not part of it. The link, on the other hand, works for everyone and works today.
What AI assistants say about you
A second door has opened over the last two years, and it does not look like the first.
When someone asks ChatGPT, Gemini or Perplexity “where can six of us eat in the 11th”, the assistant does not return a list of results: it chooses. It names two or three places, and the others do not exist in that answer. There is no second page.
What decides who gets named resembles what decides local search, but is more demanding on consistency:
- a complete, living Google listing, with recent reviews;
- mentions somewhere other than your own site — local press, blogs, guides;
- a site whose content a machine can read: hours, menu, type of cuisine, address;
- structured data, saying explicitly “this is a restaurant” and not “this is a page”;
- the same information everywhere — Google, TripAdvisor, Yelp, your site. An address differing by one line is enough to sow doubt.
The difference with ordinary search optimisation is that here you cannot know where you stand without asking. There is no ranking to consult. You have to put the question to the assistant, read its answer, and see whether your name is in it — or which competitors are in it instead.
That is exactly what Klarr's Visibility module does: it puts your questions to the assistants, keeps the exact answer with its date, and tracks it over time. Including when the answer is bad — that is when it earns its keep.
What it costs with us
For scale, since that is the subject of the article: the Bookings module is €29 excluding VAT per month, with no commission on covers or on private hire, and deposits go straight to your own Stripe account. Thirty days' trial, no card, cancel in one click.
To see how it actually works — a guest books, the request arrives, you handle it — we have filmed the whole journey and put the screens online: commission-free booking software, screen by screen.
Compare that line with the one you worked out above. If your platform volume is 200 covers a month of which half would have come anyway, you are paying roughly €100 to €200 a month for customers you already had. That is the sum to do, and it needs nothing but a multiplication.
If you want to see the four solutions on the market side by side, including what we cannot do, we have written a page for that: the comparison with TheFork, Zenchef and Guestonline — in French.
Where to start
In this order, because the order matters:
- Find your platform volume and multiply it. You now have an annual figure.
- Estimate the share of regulars in that volume. That is your reserve.
- Look after your Google listing and paste a booking link in your own name into it. There is no point cutting before that.
- Look at what the assistants answer about your neighbourhood. If you are not in it, the direct channel will stay narrow whatever you do.
- Then cut back, not before — and keep the marketplace for what it does well.
A last word on the figures in this article: other publishers' rates move, and this piece carries a date. If you see a gap with what you pay, your invoice is right, not us.
