RFM, recency, frequency, monetary, is the most useful customer segmentation framework ever built. It was invented in 1995 for direct mail catalogues, it has survived every marketing fad since, and it still outperforms 90% of the AI-branded segmentation in the market today. And almost no restaurant operators use it.
This is the definitive playbook. You'll learn what RFM is, how to score your guests on all three dimensions, how to build six actionable segments from the scores, and exactly which campaign each segment should receive.
The three dimensions
Recency, how long since a guest's last visit. Strongest predictor of future behaviour. A guest who visited yesterday is far more likely to visit next week than one who visited 60 days ago.
Frequency, how many times a guest has visited in a defined window (usually 12 months). A measure of habit.
Monetary, total spend in the same window. A measure of value.
Each guest gets a score of 1-5 on each dimension, where 5 is the top quintile of your base. A guest who is R5 F5 M5 is in your top 20% across all three, a "champion." A guest who is R1 F1 M1 is at the bottom on all three, "dormant" or "lost."
How to score
Pull your last 12 months of POS-linked loyalty data. For each guest:
Recency, days since last visit. Rank guests from newest to oldest. Top 20% get R=5, next 20% R=4, and so on.
Frequency, count of visits in window. Top 20% by visit count get F=5.
Monetary, total spend in window. Top 20% by spend get M=5.
The output is a single three-digit code per guest (545, 411, 323, etc.). There are 125 possible cells, which is far too many to action. You collapse them into six segments.
The six segments (and what to do with each)
Champions (R5 F4-5 M4-5), recent, frequent, big spend. Your top 5-10% of guests generate 30-50% of revenue. Action: recognise, not discount. Early access to new menu items, invitations to tastings, personal thank-yous from the manager. Never interrupt their flow with mass promotions.
Loyal (R3-5 F4-5 M2-4), frequent visitors at moderate spend. Your volume base. Action: progression programs, bonus-point multipliers, upsell campaigns, referral mechanics to bring a friend. These guests will respond to structured rewards far more than Champions will.
Promising (R4-5 F1-2 M1-3), recent but new, low frequency. Guests who've just entered the program. Action: second-visit conversion. The gap between one visit and two visits is the most important transition in restaurant LTV. Bonus points on visit #2, within 14 days of visit #1.
At-risk (R2 F3-5 M3-5), previously loyal, now slipping. Used to come weekly, haven't been in 30-60 days. Action: win-back. Personal-feeling message, meaningful but bounded incentive (not a 50% firesale), short redemption window.
Needs attention (R2-3 F2-3 M2-3), middle of the distribution, drifting. Action: nudge campaign. A reason to come in the next two weeks, a content hook (new menu, seasonal event), moderate incentive.
Dormant / Lost (R1 F1-2 M1-2), haven't been in 90+ days, historically low engagement. Action: final-try reactivation with aggressive but time-bound offer. If they don't reactivate, suppress them from further sends to protect deliverability and your unsubscribe rate.
The math that matters
Two numbers explain why RFM is so valuable:
Moving a guest from At-Risk to Loyal typically adds 4-8 visits over the next 12 months. At AED 70 average check, that's AED 280-560 in additional revenue per recovered guest. A campaign that hits a 500-person At-Risk segment and recovers 12% is worth AED 17,000-34,000 , from one send.
Moving a Promising guest to Loyal often doubles their annualised value. The second-visit conversion is the single highest-ROI intervention in restaurant loyalty, and it only works if you're segmenting on recency + frequency to know who needs the nudge.
Common mistakes
Using only monetary. A high-spend guest who hasn't been in four months isn't your top customer, they're at-risk. R and F predict behaviour; M measures past value.
Blasting champions with discounts. Champions don't need price incentives to come back. Discounts train them to wait for the next one and erode your margin on the segment that least needs discounting.
Not re-scoring. RFM is a live score, not a one-off label. A guest who was a champion in January can be at-risk by April. Re-score at least monthly, ideally nightly.
Ignoring the bottom. Dormant guests cost you deliverability and inflate your unsubscribe rate. Suppress them aggressively after one failed reactivation.
Where RFM fits with everything else
RFM is the foundation layer for every other kind of restaurant marketing: campaigns (which segment gets which message), menu engineering (which items your champions vs at-risk guests are ordering), win-back automation, referral programs, and direct-channel conversion. Without it, every campaign is a blast. With it, every campaign is targeted.
SmartSegments runs RFM on your loyalty data automatically, re-scores nightly, and exposes each segment as a targetable audience inside the campaign builder. The work described in this guide, score the base, collapse into six segments, act differently on each, happens without a spreadsheet. But the framework is the same regardless of the tool: score on recency, frequency, monetary; segment into six; match message to segment; measure.