Halden & Rowe · Customer · Loyalty
Ran the silver-tier reactivation control cell in April.
8,000-member cell produced a 4.2× lift in 60-day repeat purchase — cleaner than expected, no gold-tier cannibalisation.
Context
What was in the room.
The source recommendation and the situation as it was understood at the moment of decision. Frozen — this is what the decision was actually made on.
Source · Opportunity
Lapsed silver-tier members show a clean re-engagement signal.
Customer · Loyalty
Run the reactivation mailer to an 8,000-member control cell in week 15 with a 20% holdout.
Rationale
Why it was decided.
The accountable owner's reasoning for the verdict. One paragraph; the record of judgement, not a defence of it.
The head of customer wanted a clean read before scaling; the 20% holdout was retained to protect measurement into a full rollout.
Evidence at time of decision
What Ana was reading.
The signals in play when the decision was taken. Preserved so the decision can be judged on the information it was made on, not on hindsight.
Eligible population
94,000
Prior mailer benchmark
2.8× repeat lift
Gold-tier overlap
6%
low cannibalisation risk
CRM capacity
available week 15
Template
signed off
Accountable owner
Aoife Callaghan
Head of Customer
Collaborators
- James Whitfield · Head of Digital
Outcome
Expected vs actual.
The number Ana modelled against the number that actually landed. Variance is shown honestly — ahead, in line, or behind expectation.
Expected
£18K
cell-level incremental gross profit · 60 days
Methodology
Expected 3.0× lift on the 8k cell at April's AOV, discounted 20% for holdout drag.
Actual
Ahead of expectation by 44%
£26K
4.2× lift landed cleanly; repeat rate held at 12 weeks with no fatigue signal.
Measured Fri 12 Jun
What Ana learned
The lesson, in one paragraph.
The pattern Ana takes from this decision into the rest of the organisation's memory.
The lesson
Silver-tier response is stronger than benchmark when the offer is a preview rather than a discount. The lift held past the 60-day window, which the model previously treated as the ceiling.
Applied from now on
Full-cohort recommendation now carries an Established confidence and models a 90-day repeat window rather than 60. Preview mechanics preferred over discount for the tier.