Add up the hours: reservations retyped from voicemail, the same "do you have a table on Friday?" answered twenty times, opening hours updated in four different places. For most independent restaurants that is ten hours a week or more — a full shift, worked by the most expensive employee in the building: you. The frustrating part? Every one of those tasks was solved years ago.
This guide builds the stack that solves them — without gadget worship, and without losing the warmth that fills your room. A website that turns craving into a booking in under a minute. Guest data that works like a memory instead of a liability. Automation for everything repetitive, AI only where it genuinely helps, ordering channels that don't cheapen a fine-dining brand, and the five numbers that tell you the whole machine is running. One rule throughout: every tool pays rent or leaves. First up: the single page every guest checks before they trust you.
The short version
- Your website has one job — the booking: a button above the fold, the menu as text (not PDF), and three-second load on mobile.
- Guest data is your quietest asset: collected at booking, GDPR-clean, and turned into recognition and reactivation.
- Automate the repetitive 80% — confirmations, reminders, waitlists, review asks — and keep humans for hospitality.
- AI answers what doesn't need you: a phone and inbox assistant that books tables mid-service beats voicemail every night.
- Run the house on 5 numbers: occupancy, no-show rate, RevPASH, repeat share and prime cost — one dashboard, reviewed weekly.
-
Front door
A website with one job: turn craving into booking
A restaurant website converts when it answers four questions in seconds — what's the food like, can I book now, where are you, what does it cost — with a booking button above the fold, the menu as crawlable text, real photography and sub-3-second mobile load. Everything else is decoration.
Your website has one moment to live: someone, probably on a phone, probably tonight-minded, deciding whether the craving becomes a booking. Most restaurant sites lose that moment to an autoplay video, a PDF menu pinching open, and a "contact" page where a booking button should be.
The conversion anatomy
- Book, above the fold, on every page. The visitor who must hunt for the button is a visitor your neighbour converts. Link it straight into your reservation system — never a contact form that promises a reply "within 48 hours".
- Menu as text, not PDF. Text menus load instantly, work on phones, update in minutes — and they're how Google learns you serve turbot, which is how "turbot restaurant near me" finds you.
- Photography that tells the truth beautifully: six excellent photos (room at golden hour, three signature plates, faces, the door) beat sixty mediocre ones — the craft is in the marketing guide.
- Speed and basics: sub-3-second mobile load, hours and address in the footer of every page, no music, no splash screens. The full checklist is in designing a restaurant website.
If maintaining this yourself sounds like a second job — it is; that's why a managed restaurant website tied to your booking and menu data exists as a product.
Do this tonightOpen your site on your phone, on mobile data, and time three things: seconds to load, taps to a confirmed booking, and whether tonight's menu is readable without pinching. Each failure is bookings leaking — and each is fixable this week.
-
The asset
Guest data: the asset hiding in your reservation list
Every booking already captures name, contact, party size, dates and preferences. Structured into guest profiles — with GDPR-clean consent, purpose and retention rules — that data becomes recognition, reactivation and smarter forecasting. Scattered across notebooks and inboxes, it's a liability instead.
Big platforms spend fortunes to learn what you learn free every night: who dines, with how many, drinking what, celebrating what. The difference is they structure it. Most restaurants leave the same gold scattered across a reservation book, a WhatsApp thread and the maître d's memory — unusable, and one inspection away from a problem.
From data to advantage
- Recognition: guest profiles surface preferences and history at the next booking — the loyalty engine of the guest experience guide.
- Reactivation: "guests not seen in 90 days" becomes a monthly list and a personal note — consistently the highest-ROI message a restaurant sends.
- Forecasting: booking curves predict covers, covers predict prep and rosters — chapter 3 of the staffing guide runs on this.
GDPR as craft, not fear
The European rules, mapped for restaurants in guest data & GDPR, reduce to four habits: collect only what serves the guest (a birthday helps; a passport number doesn't), ask consent properly at booking (pre-ticked boxes are illegal; a clear marketing opt-in is fine), protect access (one system with logins, not an exported spreadsheet on three laptops), and honour deletion requests within a month. Done right, privacy is hospitality: "we remember your allergies, and only what you'd want us to" is a trust sentence, not a compliance cost.
Do this tonightCount where guest data lives in your house right now — book, inbox, phone, memory, spreadsheet. Each location past "one system" is both leaked value and GDPR exposure. Pick the system that becomes the single home, and set a migration date.
The 90-day list that out-earns every ad
Run one query monthly: guests with 2+ visits whose last visit is 90+ days ago. These are people who chose you twice and drifted — not rejected you. A personal two-line note ("the new season's menu landed, and the venison reminded us of your table") reactivates a remarkable share, at a cost of zero. Ad platforms charge fortunes for audiences a tenth this warm; your reservation history builds it free, forever.
-
Autopilot
Automate the repetitive 80% — keep humans for hospitality
Confirmations, reminders, waitlist refills, review requests, no-show follow-ups: rule-based and repetitive, they consume 10+ staff-hours a week done manually. Automating them is the highest-ROI move in restaurant tech — and it makes service better, because messages never forget and never get busy.
List everything your team did yesterday that a rule could describe: "when a booking lands, send confirmation", "24h before, send reminder", "when a table frees, message the waitlist", "day after a visit, ask for the review". None of it needs judgement, warmth or memory of last season's menu. All of it needs to happen every single time — which is precisely what humans under pressure can't guarantee and automation exists for.
The automation map
What runs itself in a well-set-up house Flow Trigger → action Weekly hours back Confirmations Booking → instant WhatsApp/email confirm 2–3 Reminders 24h before → one-tap confirm/cancel 1–2 (plus the no-shows it kills) Waitlist Cancellation → next match messaged 1–2, mid-service Review ask Day after → one warm message with link 1 Reactivation 90 days quiet → personal-feeling note 1, plus the revenue Occasions Anniversary near → invitation —, pure delight The one rule of tone
Automated must never feel automated. Write every template the way your best maître d' speaks — by name, in your house's voice, one purpose per message. Guests don't resent automation; they resent cold. (And they never see the difference between a warm template and a warm typist — except the template never sends at 1 AM with a typo.)
Do this tonightTally yesterday's repetitive messages — confirmations sent, reminders typed, waitlist calls made. Multiply by 360 days. That's the annual hour-cost of not automating, and your business case is now written.
-
The assistant
AI: the host that answers when you can't
Restaurant AI earns its place where unanswered demand dies: a phone assistant that books tables during service and after close, an inbox assistant drafting replies to routine questions, and campaign drafting for marketing. It handles the 80% that's routine and hands the 20% that's human to humans.
Count the calls your restaurant misses in one week: mid-service rushes, Sunday mornings, 22:40 after a date-night decision. Each unanswered ring is usually a booking attempting to happen — and a voicemail is where booking attempts go to die. This, not science fiction, is where AI in hospitality pays: the work was never about replacing your maître d'; it's about existing in the hours your maître d' doesn't.
Where AI earns its keep today
- The phone: an AI receptionist answers every call, checks live availability, books the table, answers "do you have vegetarian options?" — and hands anything unusual to a human with a summary. Restaurants switching one on discover how many bookings the busy tone was eating.
- The inbox: an AI inbox drafts replies to the twenty daily routine emails — allergies, group requests, parking — in your tone, for one-click human approval.
- The marketing desk: AI marketing drafts the monthly newsletter and campaign copy from your menu changes; you edit the warmth in, in minutes instead of evenings.
The boundary that keeps it hospitality
One principle decides every AI deployment: AI handles requests; humans handle relationships. The anniversary table's special wishes, the complaint, the regular who calls to chat — routed to people, always. Guests forgive a machine for being a machine; they never forgive a restaurant for making them feel processed. Drawn this way, the line means AI gives your floor more human minutes, not fewer.
Do this tonightCheck this week's missed-call count on your phone system (or count tomorrow's unanswered rings honestly). Multiply by your average ticket and a 50% booking intent. That's the monthly revenue sitting in your busy tone.
-
Channels
Ordering channels: digital where it serves the concept
For fine dining, ordering tech must pass one test: does it deepen or cheapen the experience? QR menus work as living information (wine lists, allergens, translations), not as waiter replacements; direct online ordering suits structured products like tasting boxes and gift cards; delivery platforms deserve hard margin math before any yes.
Hospitality tech debates get religious — "QR codes killed service!" — when the question is operational: which channel serves your concept's promise? A three-star tasting room and a bistro answer differently, and both can be right. The decision framework lives in digital ordering & QR and online ordering; here is the fine-dining read.
The channel test, applied
- QR as a living document, not a waiter: in fine dining, ordering stays human — but a QR wine list with tasting notes in four languages, live allergen filters, and the story behind tonight's menu adds theatre. Information digitises beautifully; hospitality doesn't.
- Direct online sales for structured products: gift cards, the chef's tasting box for home, wine from your cellar, prepaid event seats — products with fixed grammar sell perfectly online, fund cash flow, and carry zero service risk.
- Delivery platforms — do the math first: 25–30% commission on a concept built on plating and room is usually margin theatre. If takeaway fits at all, a focused own-channel product (Sunday roast boxes, the bistro line) at full margin beats renting your brand to an app. Your own app keeps regulars one tap away without the commission.
Whatever channels you choose, they must feed one system — the same guest profiles, the same availability, the same numbers (chapter 6). Channel sprawl with disconnected data is how houses end up managed by their tools.
Do this tonightApply the test to every digital channel you run: deepen or cheapen? One column each. Anything in the 'cheapen' column either gets redesigned to serve the concept — or retired before it quietly erodes what guests pay you for.
-
The cockpit
The five-number dashboard that runs the house
All restaurant data reduces to five weekly numbers: occupancy per service, no-show rate, RevPASH, repeat-visit share and prime cost. One dashboard, fifteen Monday minutes, each number owned by a system from these guides — that's data-driven without drowning in dashboards.
The promise of "restaurant analytics" usually arrives as forty charts nobody opens after week two. The houses that actually run on data do the opposite: brutally few numbers, looked at without fail, each wired to an action. Restaurant analytics done right is a cockpit, not a museum.
The Monday cockpit — five numbers, five owners Number Healthy It's the report card of… Occupancy per service 85%+ weekend / 60%+ midweek Reservations & demand shaping No-show + late-cancel rate < 3% Confirmation chain & deposits RevPASH Trending up Pacing, pricing, turns (finance guide) Repeat-visit share 30%+, growing to 50% Experience & retention marketing Prime cost ≤ 60–65% Menu & labour systems Notice what happened: the five numbers are the five other guides. The dashboard isn't another project — it's the nervous system connecting everything you've built, automated by your analytics so Monday's fifteen minutes are reading, not collecting.
Data with a conscience
Two closing disciplines. First, act on one number a week — the dashboard exists to start projects, not to be admired. Second, let the same data serve more than margin: portion forecasts that cut waste and energy insights from quieter services feed the sustainability ledger too — the rare project where the planet and the P&L agree.
Do this tonightDraw the five-number table on paper and fill in what you know today. Every blank cell is a system from one of these guides waiting to be switched on — and now you know exactly which guide to open next.
The metric that predicts next month before it happens
Add a sixth number when you're ready: booking pace — covers already on the books for each of the next four weeks, compared to the same point last cycle. It's the only restaurant metric that looks forward: pace down 20% three weeks out means the quiet fortnight is preventable (a campaign, a waitlist nudge, an event) instead of survivable. Hotels have run on pace for decades; restaurants with reservation data have it sitting unused.
How hard does your tech actually work?
Eight yes/no checks across the stack. Tick what runs today — not what's planned. Your score saves locally for the return visit.