Loyalty Tier Threshold Calculator

How it is calculated

Tier thresholds should be set from the distribution of customer spend rather than at round numbers, so that each tier contains a meaningful group and the step to the next is achievable.

The usual approach is percentile-based: a top tier at the 90th or 95th percentile of annual spend, a middle tier at the 70th.

Threshold = the spend level at the chosen percentile, adjusted to a memorable number.

Numbers on it

Annual spend across 12,000 customers: the 70th percentile is $118 and the 90th is $284.

Setting tiers at $125 and $300 puts roughly 28% in the middle tier and 9% in the top, groups large enough to matter and small enough for the top tier to feel meaningful.

Setting them at round numbers instead, $100 and $500, would put 38% in the middle and 3% at the top, making the middle tier unremarkable and the top one unreachable.

The gap between tiers also matters: a customer at $250 needs $50 more to reach the top tier, which is a nudge. One needing $250 more will not attempt it.

What it does not tell you

Percentiles from historical spend describe behaviour before the tiers existed, and introducing them changes the distribution as customers push to reach a threshold.

Thresholds set too low also give benefits to customers who would have spent anyway, which is the same selection problem that affects loyalty programmes generally.

What follows from it

Set thresholds where a reachable gap exists for a meaningful number of customers, just above a cluster rather than through the middle of one.

Then show progress toward the next tier, since the mechanism only works if customers know how close they are.

Why tiers work when flat programmes do not

A flat programme rewards spending that has already happened. A tiered one creates a target, and the pull toward a threshold is a genuine behavioural effect rather than a reward for existing behaviour.

The effect concentrates just below each threshold, which is where the incremental spend comes from and where the programme earns its cost.

That means the tier boundaries are the whole design. Placed well they generate incremental spend from a defined group; placed carelessly they hand benefits to people who were never going to change what they did.

Reviewing the thresholds annually against the current distribution keeps them meaningful, since a threshold set three years ago may now capture a third of the base rather than a tenth.

Tiers that everyone reaches stop being tiers, and the pull toward the next level disappears with them.

Offering a grace period at renewal, where a customer keeps their tier for a period after falling below it, materially reduces the resentment that demotion otherwise produces.

Communicating progress toward the next tier at the point of purchase, rather than only in a monthly email, places the prompt where the decision is being made.

Measuring how many customers cross each threshold, and when, shows whether the pull effect is real or whether the tiers are merely labelling existing behaviour.