The Marginal Retention-Acquisition Ratio: A Single-Score Allocation Tool for Foodservice Marketing

Section 1 — Why a cross-industry average is the wrong instrument for a single-operator decision

The Reichheld-Bain finding from 1990 — that a five-percent increase in customer retention produces a twenty-five-to-ninety-five-percent increase in profit — is one of the most-cited and most-misapplied results in hospitality marketing. The original study, published in the Harvard Business Review by Frederick Reichheld and W. Earl Sasser, is a cross-sectional analysis across multiple industries and maturity stages. The seventy-percentage-point spread is the arithmetic consequence of that heterogeneity, not the measurement of a universal effect. A restaurant operating at twenty-two-percent return-rate with EUR 45 customer acquisition cost faces a different allocation question than an operator at fifty-eight-percent return-rate with EUR 12 CAC. The Bain finding gives them the same recommendation. For at least one of the two, that recommendation is wrong.

The decision problem chained-foodservice operators face is not whether retention matters — every cross-industry analysis confirms that it does. The decision problem is allocation: at the margin, does the next euro of marketing budget produce more return through retention investment or through acquisition investment? An industry-average finding cannot answer that question. The Marginal Retention-Acquisition Ratio (MRAR) operationalises the allocation decision in a single calculable score.

Section 2 — The MRAR formula and the four zones

The composite is intentionally narrow. MRAR = (CLV_avg × RR_gap × 2.5) / CAC. Four variables, each tied to data that any chained-foodservice operator with a competent POS and a marketing controller can produce within a working day. The 2.5 multiplier is not an estimate; it is the conservative midpoint of the empirical 2.0-to-3.0 band documented in Reinartz and Kumar's 2003 Harvard Business Review study and confirmed in Gupta and Lehmann's 2005 monograph Managing Customers as Investments — a one-percent increase in return-rate translates into a two-to-three-percent increase in customer lifetime value across the studied categories. The 60-percent benchmark anchoring RR_gap is the Starbucks-Rewards reference: aspirational but documented as achievable for engaged operators. Setting the benchmark at the foodservice industry median of thirty-five to forty percent would zero out MRAR for half the sector and break the decision logic.

The composite returns a value between zero and roughly six in plausible operating ranges. The four zones partition the decision space.

Two thresholds carry analytical weight. MRAR equal to 1.0 is the break-even line — retention and acquisition deliver identical marginal return, and any allocation shift is speculation rather than analysis. MRAR above 5.0 is the upper boundary at which a calibration check is warranted: at that level the gap to the sixty-percent benchmark is so wide and the CAC so high that loyalty investment promises exceptional returns — which is also the structural signature of a measurement error. Either CLV_avg is being overstated, CAC is being understated, or RR_current is being measured against the wrong cohort definition.

Section 3 — Saturation zone: where the formula breaks

Above sixty-five-percent return-rate the formula loses its decision power. RR_gap turns negative against the sixty-percent benchmark, and the composite either flips sign or returns a meaningless figure. The methodologically correct read of MRAR in this band is not "retention priority confirmed" — it is "the formula does not apply; a loyalty-yield check is required." Reinartz and Kumar named the underlying mechanism deadweight: loyalty programmes credit points or discounts on transactions that would have happened identically without the programme. For operators with high organic return-rates the deadweight share is structurally high — programme spend funds behaviour that needs no programme stimulation. The arithmetic compliment becomes an analytical trap.

An MRAR above 5.0 paired with a return-rate above sixty-five percent is therefore not an excellence signal. It reads as an indicator of structural misallocation. The corrective is not blanket budget reduction. It is segmentation: which guest cohorts sit at return-rate below forty percent despite a CLV_avg profile that justifies activation? That cohort is the genuine loyalty target — not the regular guest who already visits three times a week and accrues programme points on autopilot.

Operators with high organic return-rates do not exit the loyalty programme. They redirect the loyalty budget toward the marginal-customer cohort — guests within reach of the regular-visitor frequency threshold but not yet across it. The deadweight diagnostic does not invalidate the loyalty programme. It re-targets it.

Section 4 — Three retrospective cases

Starbucks Rewards US 2008–2012 (MRAR > 3.0, Retention Priority). The starting position in 2008: return-rate approximately thirty-five to forty percent, app launch in progress, high CLV_avg through mass-market frequency, moderate CAC. The gap to the sixty-percent benchmark sat at twenty to twenty-five percentage points; combined with high CLV_avg and moderate CAC, the retrospective MRAR sits comfortably above 3.0 — Retention-Priority territory. The decision to invest aggressively in loyalty was correct on the framework. The validating outcome — Rewards members visiting outlets 5.6 times more frequently than non-members, the programme contributing forty percent of total revenue today — is not retrospective marketing narrative; it is a structurally verified effect documented in Starbucks investor relations since programme inception. The retention case is unusually clean because the brand also operated as a resilience compounder across the period — a related thread covered in Resilience Asymmetry in Chain Foodservice.

McDonald's MyMcDonald's DE 2021–2023 (MRAR 1.0–2.0, Balanced). The methodologically more demanding case. In a QSR segment with structurally high organic visit-frequency, CAC is low, CLV_avg is low, and RR_gap is small — MyMcDonald's operates near the Balanced threshold. The deadweight problem is particularly sharp here: a customer who visits the brand multiple times per week is rarely behaviourally altered by points-based incentives — points accrue on transactions that would have happened regardless. An incremental CLV uplift through retention investment is structurally harder to achieve in this segment than in casual-dining formats with infrequent visitation and high RR_gap. The Balanced signal reads as a warning, not as a green light. McDonald's Deutschland has not published incremental-sales figures from MyMcDonald's — a quiet tell about programme economics that the framework reads as consistent with the Balanced zone.

Vapiano DE 2018–2020 (MRAR < 1.0 + concept-fit failure). The framework's edge case. By 2018, Vapiano showed return-rate below thirty percent (driven by quality issues and chip-card friction at the order interface), CAC above EUR 60, and like-for-like comparable revenue minus 4.2 percent in the first nine months of 2019 against industry growth of plus 5.3 percent — a structural concept-fit failure underway. MRAR points to acquisition priority. But a CAC problem rooted in declining net promoter score is not solvable through acquisition campaigns: new guests arrive, encounter the quality problem, and do not return. The chain entered insolvency in April 2020 with 235 stores across 33 countries at peak. The case demonstrates the framework's edge: MRAR allocates between retention and acquisition within a viable concept; it cannot diagnose or correct a concept-market-fit defect underneath. The layer below MRAR is addressed in The Concept-Market-Fit Score for Foreign-Chain Entry.

Section 5 — The proxy route for operators without POS tracking

MRAR presupposes POS tracking — loyalty cards, reservation systems, or linked payment data that identifies guests across visits. For operators without that infrastructure, the proxy route allows qualitative MRAR estimation using two readily available variables: current return-rate band (low, medium, high) and current CAC band (below EUR 25, EUR 25 to 50, above EUR 50). The proxy table maps the four-zone diagnostic onto observable inputs, providing directional guidance without precise calculation. CAC orientation for German restaurant categories sits at EUR 17 to 30 on a pure ad-spend basis (WordStream Google Ads benchmarks 2025, Enrich Labs Meta Ads benchmarks 2025); the realistic operating-context band is EUR 25 to 60 depending on segment density and metropolitan-area cost premiums.

An operator who cannot calculate MRAR from POS data but can answer two questions — what is the approximate return-rate band, and what is the approximate CAC band — has a directional read. An operator who can answer neither question has no allocation-decision basis at all and is operating loyalty budget on heuristic alone.

Section 6 — Three operating consequences for marketing allocation

Measure return-rate before raising loyalty budget. Without an RR baseline, every loyalty investment is speculation. The share of foodservice operators who can reliably calculate current return-rate from POS data is materially below the share that runs loyalty programmes. The gap is not technical incapacity — modern POS systems export the data — it is the absence of return-rate as a steering metric. The remediation is not a new tool. It is the routine inclusion of return-rate in monthly operating reviews.

Knowing CAC is mandatory. An operator who has not calculated customer acquisition cost — marketing spend on new guests divided by new guests in the period — cannot apply MRAR and is making allocation decisions without price comparison between the available levers. This applies equally to single-unit operators and to chains administering centralised marketing funds. Setting budgets by habit or by competitor benchmark, without knowing one's own marginal ROI, is not an analytical decision in any meaningful sense. The German CAC band of EUR 25 to 60 across Meta and Google Ads provides initial orientation but does not substitute for operator-specific calculation. The cost-side context that frames CAC at country level is documented in The DACH Operating Environment.

MRAR is a monitoring instrument, not a one-time check. Markets shift, loyalty programmes alter return-rate, CAC moves with competitive intensity and platform pricing. An MRAR of 4.2 today (Retention Inclination) can compound into 1.5 (Balanced) within six months through rising CAC or through a successful loyalty programme that closes the RR_gap. An allocation decision based on a stale MRAR is worse than no decision at all. Quarterly recalculation is not analytical luxury — it is the minimum cadence that operates the instrument as steering metric rather than as one-time report.

Section 7 — What the score does not measure

MRAR sits inside a defined boundary. It does not measure concept-market-fit — the layer addressed by the Concept-Market-Fit Score The Concept-Market-Fit Score for Foreign-Chain Entry. It does not measure channel allocation between dine-in and delivery — the orthogonal allocation question covered by the Channel Cannibalization Index The Channel-Cannibalization Index: Delivery vs. Dine-In. It does not measure product-margin architecture, brand-equity development, or talent-side retention — the last of which is covered separately by the Pre-Quit Signal Score The Pre-Quit Signal Score: Frontline Foodservice, the companion methodology on the personnel side.

MRAR is replicable across operators because the inputs are standard POS and marketing-controlling outputs. The instrument is not proprietary; the discipline of computing it quarterly is. The methodological assumption — that CLV_avg increases linearly with RR — holds for typical foodservice configurations. For niche concepts with saturation-resistant visit frequency or for operations with unusually low regular-guest spend, the linearity may bias the estimate; operators in those categories should treat MRAR as directional rather than calibrated.

Section 8 — Prescriptive close

We compute MRAR before reallocating loyalty budget, not in retrospect. Operators who shift retention spend on the strength of the Bain finding alone shift on a cross-industry average — and the average is the wrong instrument for a single-operator decision.

Related research

• The Concept-Market-Fit Score The Concept-Market-Fit Score for Foreign-Chain Entry — concept-fit underneath MRAR allocation
• The Channel Cannibalization Index The Channel-Cannibalization Index: Delivery vs. Dine-In — companion allocation methodology on the channel side
• The DACH Operating Environment The DACH Operating Environment — CAC level and margin architecture as MRAR inputs
• The Pre-Quit Signal Score The Pre-Quit Signal Score: Frontline Foodservice — talent-side retention companion methodology

Sources

• Frederick Reichheld and W. Earl Sasser — Harvard Business Review 1990 (the original five-to-ninety-five-percent retention-profitability finding, cross-industry cross-sectional)
• Werner Reinartz and V. Kumar — Harvard Business Review 2003, "The Mismanagement of Customer Loyalty" (sixteen-thousand-customer challenge to the Bain finding across four categories)
• Sunil Gupta and Donald Lehmann — Managing Customers as Investments, 2005 (the 2.0-to-3.0 multiplier range for retention-to-CLV translation)
• Starbucks Corporation — Investor Relations 2008–2025 (Rewards programme effects, 5.6 times visit-frequency among members versus non-members, forty-percent revenue share)
• Vapiano SE — annual report 2019, insolvency filing April 2020 (like-for-like minus 4.2 percent at nine months 2019 against industry growth plus 5.3 percent; 235 stores across 33 countries at peak)
• McDonald's Deutschland — press releases 2021–2023 (MyMcDonald's app rollout)
• WordStream Google Ads industry benchmarks 2025; Enrich Labs Meta Ads benchmarks 2025 (CAC orientation for restaurant categories: EUR 17–30 ad-spend basis; EUR 25–60 realistic operating context)