In the last 18 months three things changed at once. Quandoo announced its shutdown. The EU AI Act started enforcing transparency rules on voice agents. And OpenTable rolled out its own AI phone product, branded "Voice-KI", priced — like everything else at OpenTable — per cover.
If you have run a restaurant in DACH for any length of time, you know the per-cover model. You pay a marketplace €1.50–€2.50 every time a guest sits down. The marketplace takes credit for the booking — even if the guest had eaten with you twelve times before. Over a year, that fee adds up to a single big-line P&L item that nobody enjoys reading.
We built Tablario as the opposite of that. Plug-in modules, fixed monthly fees, no per-cover commissions, no marketplace economics. This post explains what that means concretely — and why the architecture decision is upstream of every other product decision we make.
What "sit on top, not replace" means
Most reservation platforms aim for the entire stack: their booking widget, their POS, their card terminal, their CRM, their email marketing, their marketplace, their app. The pitch is "we replace seven tools with one." The unspoken cost is that you also replace seven invoices with one — and that one invoice scales with your covers.
Tablario does the opposite. We assume you already have a POS you like, a card terminal that works, an accountant who reads your books, a Google Business Profile your guests find you on. We do not ask you to throw any of that away. What we add is the layer that nobody else covers cleanly:
- An AI phone that answers in real German (and 10 other languages), discloses itself per AI Act Art. 50, and writes the booking into a calendar.
- A public restaurant page that converts that calendar into bookings.
- A QR menu that captures orders without changing your POS.
- A guest CRM that stitches booking history with call history with WhatsApp history.
- A booking policy that holds a card without taking money.
Each is a plugin — independently turn-on-able, independently priced. You compose what you need.
The €14,400 question
Let's do the math an honest way. A 50-cover venue running 600 covers per month through a marketplace pays:
- OpenTable on the BookEverywhere plan: $1.50 per cover via your widget, $7.50 per cover via opentable.com. Assume an 80/20 split → average ~$2.70/cover → ~$1,620/month → ~$19,440/year.
- OpenTable Voice-KI is currently a free add-on for paying customers. The math is the same — you keep paying per cover.
- TheFork in Europe runs a similar economic model with €2/cover for premium placement.
A 50-cover venue on Tablario pays a fixed monthly fee for the plugins they use. The fee does not move with covers. Run 400 covers next month and you pay the same. Run 1,200 next month and you pay the same. The savings on a 600-cover venue are roughly €10,000–€14,000 per year depending on which marketplace you compare to.
That is not a marketing exaggeration. It is what happens when you stop paying a per-cover commission to a third party.
Why the architecture decision matters
A plugin-based, fixed-fee platform is not just a pricing choice. It forces engineering decisions that compound:
- Each plugin must be independently switch-off-able. That means clean module boundaries in code (we run a modular monolith on the backend, with strict per-module routes/services/repositories). When you turn off the QR plugin, the booking module does not notice.
- Each plugin must be priced on its own merit. That means we cannot smuggle a weak plugin past you by bundling it with a strong one. If the WhatsApp module is not useful, you do not buy it. We see fast which plugins land — and which we should reconsider.
- Each plugin must be replaceable by a partner. If you would rather use Mews for POS, fine — your Tablario booking module talks to Mews via a connector. If you'd rather use Klarna for deposits, fine — Stripe is the default but the contract is open.
A vertical platform cannot promise this. The whole point of vertical pricing is that the parts cannot be priced apart.
What "no marketplace economics" means
We are not building a consumer-facing app. There is no Tablario.com restaurant directory where guests browse and book. The Restaurant Page lives at tablario.com/r/your-restaurant, the URL is yours, the SEO accrues to you, the schema.org markup names your restaurant — not Tablario.
This is a deliberate trade. A consumer marketplace has higher gross margins than a SaaS — once you're the destination, you can charge for placement, lift the cover fee, and add a fee on guest-side transactions. The growth math at the seed and Series A stage favours marketplace.
But the growth math at the operator stage works the other way. A marketplace whose unit economics depend on cover fees is structurally incentivised to retain control of the guest. We are structurally incentivised to give it back to you, because our revenue does not depend on which channel the booking came from.
What you can compose today
The plugins live in three groups:
Always-on (included with the base):
- Public Restaurant Page
- Reservations (web widget, embed, link-in-bio + QR)
- Guest CRM
- AI Act-compliant audit trail
Guest-facing plugins (loose, lazy-loaded):
- AI Phone (DE/EN today; 9 more in Phase 2)
- WhatsApp inbound + outbound
- Booking Policy (card auth / deposit)
- Waitlist
Operational plugins (tight, real-time, with circuit breakers):
- POS connectors (Phase 2 — Lightspeed first, then Mews, then Apaleo)
- Payments at the table (Phase 2, KassenSichV-gated)
- KDS handoff (Phase 3)
You buy what you need. The base is enough to run a competent reservation operation today.
What this is not
- Not "everything is free." We charge. The fees are flat and visible — they don't grow with your covers, but they aren't zero.
- Not a replacement for relationships. A POS connector talks to your POS; it does not replace your POS. Your accountant still books your sales. Your card terminal still settles to your bank.
- Not infinitely customizable. We make opinionated choices on UX, on language tone, on what counts as compliance. The plugins are configurable; the architecture is not.
What to do about it
If you are paying per-cover today, run the math on a real month: covers × cover fee, twelve months. Compare it to a flat plan on Tablario. The number is usually not close.
If the comparison is close — for example, a 200-cover venue on a discounted plan — Tablario may not save you money in year one. That is fine. Your covers will still be yours, your guest data will still be yours, and the fee will not grow when you grow. We will be cheaper on the day you cross the threshold.
Plugin pricing is on /preise. Tablario's modular architecture is documented in our public roadmap. The cover-fee comparison numbers above are sourced from OpenTable's published BookEverywhere and Marketing-Boost rates as of Q1 2026; verify current pricing with OpenTable directly before signing.