Airline Economics · Loyalty

Member lifetime value

What a loyalty member is worth over time, tier by tier.

A Gold member is worth more than a Base member this year, but what matters is the whole relationship: members move between tiers and some stop flying. This calculator adds up each year's margin, follows where each tier's members go next, and discounts it all back to today.

1A year of margin per tierWhat a member's flying earns, plus what the bank and partners pay for the points they give out, minus the cost of the awards members redeem and of their status benefits.
2Where members go next yearEach year a member can stay in their tier, move up, drop down, or lapse. A transition matrix holds those odds, and the same odds apply every year.
3Add it up and discountLifetime value is the expected margin in each future year, weighted by where the member is likely to be, discounted to today and summed over the horizon.
Tier economics
Per member and per year. Points are the member's yearly earn; the page prices them with the four values above.
Yearly tier transitions

Share of members in each row's tier who end the next year in each column's tier, in %. The shaded cell (stay in the same tier) is whatever is left, so each row adds up to 100.
Loading…

What a member is worth, by starting tier

Lifetime value and where it comes from

One year versus a lifetime

Annual margin is one year in the tier. Lifetime value follows the member as they move between tiers or lapse.

Where members go

Expected tier mix, year by year, for a member starting in each tier

What moves lifetime value most

Each lever moved on its own, the others held at the settings above
Value of
Move each lever by

Scenario: keep more members in a tier

Value created across the member base

Suppose a retention effort (a softer landing, a targeted offer, better service recovery) cuts the share of one tier's members who lapse each year. The members it keeps behave like the tier's other members who stay: most remain, some move up or down. Members who start elsewhere also gain when they pass through that tier.

Scenario: more members take the co-brand card

Value created across the member base

Suppose a card campaign converts some members into cardholders. Each extra cardholder brings the bank's payment for the points they earn on the card, and costs the awards those points are eventually redeemed for. The model assumes the card changes nothing else: the member flies and lapses exactly as before.

How it works

Annual margin per tier

Every member of a tier is treated as the tier's average member. Their yearly margin is the contribution of their paid flying, plus what the bank pays for the points they earn on the co-brand card (for the share of the tier holding it), plus what partners pay for the points they earn with them, minus the cost of honouring the points they will redeem and the cost of their status benefits.

m(k) = F(k) + u(k) × C(k) × p_bank + Q(k) × p_partner − [ E(k) + u(k) × C(k) + Q(k) ] × ρ × c − B(k) F flight margin ($/year) E points earned from flying u share of members with the card C card points per cardholder Q points earned with partners B benefit cost ($/year) p price paid per point (bank, partners) ρ share of points eventually redeemed c cost of honouring a redeemed point

Redemption cost is booked in the year the points are earned, for the share expected to be redeemed: the timing of the awards themselves is ignored, which is close enough for a lifetime total. The breakage model follows that timing in detail. The default prices come from the illustrative example in Loyalty 101, chapter 07: the bank pays 1.2¢ a point and honouring one costs the airline about 0.7¢.

Tier migration as a Markov chain

The states are the five tiers plus Lapsed, a member who no longer flies or earns and brings no margin. P(i, j) is the probability that a member in state i this year is in state j next year. A member's position after t years is a probability over the states, found by applying the matrix once per year, starting from certainty in the starting tier s:

π₁ = e(s) (1 in the starting tier, 0 elsewhere) π(t+1) = π(t) × P

Retention of a tier is 1 − P(k, Lapsed): the share of its members still active a year later, whatever tier they land in. A lapsed member can come back through the Lapsed → Base cell.

Lifetime value

Lifetime value from starting tier s is the expected margin in each year, discounted at rate d and summed over a horizon of N years. Year 1 is the current year and counts in full, as in Loyalty 101, chapter 08.

CLV(s) = Σ_{t=1..N} π(t) · m / (1 + d)^(t−1) in matrix form, for every starting tier at once: CLV = Σ_{t=0..N−1} (P / (1+d))^t · m → (I − P/(1+d))⁻¹ · m as N → ∞

The member base value is Σ members(s) × CLV(s). The scenarios compute it twice, with and without the change, and report the difference. The tornado chart moves one lever at a time by the chosen percentage: retention through the lapse rate (a 20% lower lapse rate turns 7% into 5.6%), card uptake by its share, and flight margin, redemption cost and benefit cost by their amounts.

Worked example

Starting tier

Where the default numbers come from

All the defaults are assumptions, chosen to look plausible for a large network airline and stored in assets/data/memberclv.json. The five-tier pyramid (5 million members, 90% of them Base) and the margins reflect the pattern described in Loyalty 101, chapter 06: a small share of members brings a large share of the money, and the top tiers cost the most to serve. The transition matrix has the usual shape: high tiers are sticky, most members who lose status drop one tier rather than to the bottom, and a quarter of Base members stop being active each year. The card economics follow chapter 07; for a sense of scale, Delta reported that American Express paid it $8.2 billion in 2025 under their co-brand agreement.

Sources

Limitations

This is a personal project and isn't affiliated with any airline. For the business side of segmentation and lifetime value, read Loyalty 101: Segmentation & Customer Lifetime Value.