Loyalty benchmarks: how to check one before you reuse it

What to take away

Decide what the benchmark is for

Loyalty benchmarks answer a question your own data cannot, usually because the scheme is new or small. If you hold three years of clean history, your own trend line beats any published average. Write the question down before you search. "Is our 22 per cent redemption rate low?" is answerable; "how are loyalty schemes doing?" is not.

Before you open a report, fix the internal definitions you will compare against: what counts as an active member, how a redemption is counted and how breakage is treated. Beside every figure you keep, record the publisher, the fielding date, the sample size and the exact definition used; a number without those companions is not evidence. An external figure can only land on a stable internal baseline.

What makes two benchmarks comparable

Comparability rests on who was measured, over what period and how the metric was defined. A grocery scheme with weekly shoppers shows different redemption and breakage from a subscription service. A figure published three years ago also reflects a different cost base.

Check the sample before the headline

Measure the audience first. A scheme open to all shoppers behaves differently from one limited to a paid tier, and members recruited during a promotion inflate early engagement. Ask how the publisher treated lapsed members, because a benchmark that drops them will look flattering. Ask for the response rate as well as the sample size, and treat a sample as adequate only when it can be broken out to the cohort you run — tier, tenure and channel.

Compare the definitions, not just the numbers

Redemption rate, active member and breakage are not standard terms. One publisher may count a member as active after a single transaction in 12 months; another may require two in six. Put the definitions side by side before the numbers.

Check whether the headline is a mean or a median. A few very large programmes pull an average upwards and make a healthy scheme look weak. Ask for quartiles if the publisher holds them.

Referral benchmarks need the same test

Referral metrics vary even more than redemption metrics. A referral rate per member, a rate per invitation sent and the share of referred customers who complete a first purchase are three different numbers, and only the last speaks to acquisition cost. Ask which one the publisher used before you set it against your own.

Benchmark reward value against your terms

A redemption figure says little if your terms make points hard to use. Expiry windows, minimum spend thresholds and exclusions all change what a point is worth, and those conditions sit inside a consumer contract. The Consumer Rights Act 2015 sets the fairness test for terms in consumer contracts, so a comparison that ignores your terms compares two different products.

Where published benchmarks come from

Most free figures come from platform vendors, trade bodies or academic teams. Vendor reports often carry the largest samples and the thinnest disclosure, so ask for the method note before you quote one. Trade bodies tend to describe practice rather than performance, which suits process comparisons but not results. The DMA measurement pillar is a sensible starting point for documenting marketing effectiveness.

Teams that skip the comparability work repeat the same faults: a headline read against the wrong definition becomes a budget line, and the mismatch surfaces only after the money is spent.

Who should read the benchmark

The person interpreting a benchmark needs to know the difference between a cohort and a snapshot, and between a rate per member and a rate per transaction. Without that grounding, errors travel straight into the board pack. CIM data and insights training gives a team the vocabulary to question a supplier's method.

Reading a published figure against your own data

Suppose, for example, a retailer runs a points scheme with 40,000 active members. Last year it issued 12 million points and members redeemed 3.4 million, a redemption rate of roughly 28 per cent. A published benchmark puts the sector at 34 per cent, so the team drafts a £50,000 budget request to close the gap.

Before spending, they rebuild the benchmark. It counted redemptions against points issued over 12 months, while their own points expire after 24 months. On a 12-month basis their rate is nearer 31 per cent. The remaining gap is small, so the money goes to a retention test.

Illustrative figures only.

Common questions

Can I reuse a competitor's published figures?

Only when the sample and definitions line up with yours. Otherwise quote it as context and note in the board pack that definitions differ.

How many benchmarks should I use?

Two or three from different publishers. A single figure anchors the discussion, while a spread shows how much variation is normal. If the spread is wider than the gap you are closing, the research cannot settle the decision.

When should benchmark figures be refreshed?

Annually suits most scheme metrics, and sooner when reward costs or interest rates move sharply. Diarise the review so no board pack quotes a stale figure.