Measurement
Part of When to rebuild loyalty marketing measurement and reporting
Five loyalty marketing attribution methods compared for England teams
Compare five loyalty marketing attribution methods for England teams, from last click to incrementality, with the trade-offs, evidence and reporting each demands.
What to take away
- The most common mistake is attributing every redemption to the last touchpoint, which flatters paid channels and hides the work done by email, referrals and organic search.
- Five methods cover most England programmeslast click, first click, linear, time decay and incrementality testing.
- No single method suits every reward, so use last click for simple voucher campaigns and holdout tests for any claim that you caused a behaviour change.
- Whatever you choose, fix the lookback window and the conversion definition before you report a number upward.
- Attribution proves correlation until you run a control group, so pair a model with periodic holdout tests.
Why last click keeps winning by default
Most loyalty platforms ship last-click attribution because it is easy to compute. The reward fires, the sale records, the channel gets credit. It needs no control group and no statistical work.
The problem is that it rewards whichever touchpoint sat closest to the till. A member who saw a referral email in March, browsed in April and bought in May credits only the May click. The loyalty marketing measurement and reporting guide 2027 sets out how each model behaves across a full reporting cycle.
The five methods, compared
Credit rule
- Last click
- Final touchpoint takes all credit
- First click
- First touchpoint takes all credit
- Linear
- Credit split equally
- Time decay
- More credit near the sale
- Incrementality
- Credit from a control group
Best for
- Last click
- Simple voucher campaigns
- First click
- Acquisition and referral awareness
- Linear
- Long consideration cycles
- Time decay
- Frequent purchase cycles
- Incrementality
- Claims about behaviour change
Main weakness
- Last click
- Ignores earlier influence
- First click
- Ignores conversion support
- Linear
- Dilutes the decisive touch
- Time decay
- Arbitrary decay rate
- Incrementality
- Needs scale and patience
A worked example: a member sees a referral link, opens two emails, then buys a £60 item. Under last click the email takes all the credit; under linear the referral gets a third. Same data, two stories.
Same data, two stories
Last click
- Referral link
- £0
- Email 1
- £0
- Email 2
- £60
Linear
- Referral link
- £20
- Email 1
- £20
- Email 2
- £20
Where the evidence has to come from
Attribution models are internal accounting. They show how you chose to divide credit, not what caused the sale. Only a control group, with some members held back from a message, separates cause from correlation.
Segmentation should rest on real population data rather than assumption. The Office for National Statistics publishes cultural identity data for audience segmentation that lets you check whether your member base matches the audience you think you are rewarding.
Paid channels bring their own reporting conventions. Google's guidance on campaign optimisation and reporting is worth reading before you compare platform-reported conversions with your own loyalty ledger, because the two counts rarely agree.
When one channel looks weak, read the loyalty marketing measurement mistakes in England before you cut its budget.
Fixing definitions before you report
Pick a lookback window and write it down. Thirty days suits frequent grocery trips; ninety days suits considered purchases such as insurance or furniture. Mixing windows across channels makes the comparison meaningless.
Define the conversion event precisely. Is it a purchase, a redemption, a points balance crossing a threshold, or a referral completing its first order? Each definition changes which method looks best.
Consumer law also shapes what you can promise. The Consumer Rights Act 2015 carries a schedule of further amendments affecting consumer contract law, which matters when a reward term is disputed.
Glossary
- Attribution modelthe rule that assigns credit for a conversion to one or more touchpoints.
- Lookback windowhow far back a touchpoint can count towards a conversion.
- Holdout groupmembers deliberately excluded from a message so you can measure its effect.
- Incrementalitythe lift in behaviour caused by a campaign, measured against a control.
- Conversion eventthe action you count as success, such as a purchase or redemption.
Common questions
Which attribution method should a small England scheme start with?
Start with last click for simple voucher campaigns, because it is cheap and understandable. Move to holdout testing once you have enough volume to split an audience.
How long should a lookback window be?
Match it to your purchase cycle. Thirty days works for frequent, low-value purchases; ninety days suits considered buys. Apply the same window to every channel in the report.
Can attribution prove a referral programme caused extra sales?
No. It divides credit by a rule you chose. Only a control group comparison can support a claim that referrals caused additional sales.
Do we need to report different methods to different teams?
Yes, and label which model each figure uses. Finance may want last click for reconciliation while marketing wants incrementality for planning.
