AI-Powered Marketing and Loyalty for Restaurants

Most independents send the same generic monthly email to the entire database — the regulars who eat there weekly, the lapsed guest from 14 months ago, the one-time tourist from a conference. That single broadcast cadence is the most expensive marketing decision in the industry, because it trains your highest-value guests to ignore you. AI fixes it not by adding more sends, but by segmenting the database and tailoring the message.

This article is the tactical playbook. The math on revenue lift sits in the restaurant AI ROI breakdown; the broader sequencing in the 2026 restaurant AI playbook. Here we go segment by segment.

The RFM model, restaurant edition

RFM — recency, frequency, monetary — is the segmentation backbone for guest databases.

• Recency. Days since last visit. 0–30 active, 31–90 cooling, 91–180 lapsing, 181+ dormant.
• Frequency. Visits in the trailing 12 months. 1 = one-time, 2–4 = casual, 5–9 = regular, 10+ = VIP.
• Monetary. Average check size or annual spend. Top-decile and bottom-decile bands behave very differently.

Cross these three axes and you get 8–12 actionable segments. AI does the math from Toast, Square, OpenTable, or Resy guest data and writes the segment definitions; you do not need a data analyst.

The 6 segments every restaurant should run

1. VIP Active (frequency 10+, recency 0–30)

Your 4–7% of guests producing 22–34% of revenue. The wrong move is to discount them. The right moves: birthday and anniversary recognition, comp dessert on visit 12, named-server preference baked into reservation routing, owner-signed quarterly thank-you. AI tags them in OpenTable or Resy and warm-transfers any inbound call to a human host.

2. Regular Cooling (frequency 5–9, recency 31–90)

The compounding segment. A regular who has gone 60 days without a visit is on the slow drift toward lapse. AI sends a single, non-discount, named-server SMS at day 45: "Hey, it's [GM name] at [concept]. Haven't seen you since [last visit]. We've got a new spring menu — want me to hold a seat?" Recovers 22–38% of the segment per quarter.

3. Casual Lapsing (frequency 2–4, recency 91–180)

The win-back priority. Four-touch SMS drip over 6 weeks — day 0 reminder of their favorite dish (the AI pulls it from order history), day 14 new-menu announcement, day 28 modest comp offer (free dessert or app, never percentage-off entrees), day 42 final touch with reservation link. Win-back rates run 14–24% over two quarters.

4. One-Time Recoverable (frequency 1, recency 14–90, check above median)

Tourists, work-trip diners, special-occasion one-offs. Some are unrecoverable; the AI predicts which ones are local based on phone area code, address (if captured), and ordering pattern. For locals, send a single day-21 follow-up with the chef's tasting recommendation. 6–11% return.

5. Birthday and Anniversary (date-triggered, any segment)

The highest-converting trigger in restaurant marketing. AI matches birthday and anniversary dates from OpenTable and Resy guest data and triggers an SMS 14 days out: "Birthday coming up? We'd love to have you. Bring 3 friends, dessert is on us." Books at 32–48%.

6. Dormant (recency 181+, any frequency)

Stop emailing. Move them out of the active campaign list. AI runs one annual reactivation attempt — usually tied to a concept refresh or anniversary — and otherwise drops them. Keeping dormants in the active database depresses every other segment's deliverability.

Channel selection: when to use SMS vs email

• SMS. Time-sensitive, short, personal. Win-back, birthday triggers, regular-cooling outreach, last-minute availability. Cost: $0.04–$0.08 per send. Cap at 1–2 messages per guest per month.
• Email. Storytelling, menu announcements, event invites, chef's-table launches. Cost: under $0.01 per send. Cap at 2 messages per month for most segments, 4 for VIPs who opt in.
• Push from loyalty app (Toast Rewards, Square Loyalty). Lowest-cost, highest-control. Use for points-balance reminders and tier-progression nudges.
• Direct mail. Underused. AI batches a quarterly postcard for VIPs with a named-server note. $1.20 per piece, 4–6x ROI in our engagement data.

The 14-day rollout

• Days 1–3 — Audit the guest database. Pull guest records from Toast, Square, OpenTable, Resy, and the loyalty platform. Deduplicate. Identify how many guests have email vs phone vs both. Most databases are 30–50% incomplete on phone — this is the biggest blocker.
• Days 4–5 — Segment build. AI defines the six segments above against your data. Owner and GM review the segment counts. If VIP count is implausible (under 1% or over 12%) recalibrate the thresholds.
• Days 6–8 — Message library. Draft and approve the SMS and email templates per segment. AI proposes copy; owner approves. Never let the AI send copy the owner has not seen.
• Days 9–10 — Compliance check. SMS requires explicit opt-in under TCPA. Verify your loyalty signup, OpenTable, and Resy flows are capturing opt-in. Scrub the database to opted-in only.
• Days 11–13 — Soft launch. Run the win-back and birthday flows first. Measure response rates, unsubscribes, and any service-floor backlash.
• Day 14 — Full launch. All six segments live. Set the weekly review cadence.

The same cadence runs across every AI workflow in the AI enablement engagement model.

Pitfalls to avoid

Do not over-discount. The fastest way to train regulars to expect 20% off is to give them 20% off. Recognition, story, and small comps beat percentage discounts in every cohort we have measured.

Do scrub for TCPA opt-in. Texting a guest who did not opt in is a $500–$1,500 per-message statutory damages exposure. Verify opt-in capture before any SMS goes out.

Do honor unsubscribes within 24 hours. Reply STOP must remove the guest from every list within 24 hours. AI handles this; verify it monthly.

Do not send during service. Schedule sends between 10:30 a.m.–11 a.m. or 3 p.m.–4 p.m. local. Service-window sends create call-volume spikes when the host cannot answer.

Do measure unsubscribe rate per segment. A win-back segment running over 3% unsubscribe per send means the cadence or copy is wrong; tune before you scale.

What good looks like

Four metrics. Baseline first, then measure at 30, 60, and 90 days.

• 90-day repeat-guest rate. Target: 6–12 points higher than baseline by day 90.
• Average visits per guest per year. Target: 0.4–0.9 higher across the active database.
• SMS opt-in rate at point of capture. Target: above 60% on OpenTable, Resy, and Toast Loyalty signup flows.
• Unsubscribe rate per send. Floor: under 1.2%. Above 3% on any segment, pause and retune.

This works hand-in-hand with the broader review velocity story — a guest who got a warm win-back SMS is materially more likely to leave a positive Google review post-visit.

FAQ

Q: Do I need a marketing manager to run this? A: No. The GM or owner reviews monthly and approves new copy. AI handles the segmentation, send timing, and unsubscribe management.

Q: Will my regulars find SMS annoying? A: Not if you cap volume. One to two SMS per month per guest, only when there is something real to say, and the regulars become advocates. Three-plus per month is when complaints start.

Q: What about international guests? A: International SMS is expensive and unreliable. For markets with heavy tourist traffic, switch international guests to email-only by default.

Q: Does this integrate with Toast Loyalty or Square Loyalty? A: Yes. Both expose guest, visit, and points data via API. The AI reads the loyalty signals and respects tier thresholds.

Q: Can the AI write the copy? A: Drafts, yes. Sends without owner approval, no. Brand voice is too high-stakes to delegate fully.

Q: How does this fit with review reply? A: Tightly. A guest who received a win-back SMS and visited is the highest-likelihood positive-review target. AI flags them post-visit for a review-request SMS 24 hours later. Full review playbook in the review reply automation guide.

---

If you want a segmentation and SMS rollout scoped to your concept and POS — reach out. We will pull your last 12 months of guest data and propose the six segments. Or read the broader AI for restaurants overview.

SOURCES

Cited and consulted.

1. 01National Restaurant Association — Guest Behavior and Loyaltyrestaurant.org · accessed May 8, 2026
2. 02Toast Blog — Restaurant Marketing Benchmarkspos.toasttab.com · accessed May 8, 2026
3. 03Restaurant Business — Marketing Coveragerestaurantbusinessonline.com · accessed May 8, 2026
4. 04Resy Blog — Guest Engagement and Loyaltyblog.resy.com · accessed May 8, 2026