Approach
Measure first,
then spend.
Most acquisition problems are measurement problems wearing a disguise. Our method front-loads the unglamorous work so that later decisions are made on evidence rather than argument.
Principle one
Unit economics before channel selection.
If you do not know what a customer is worth, no channel can be judged.
We start with the value of an acquired customer, the acceptable cost to acquire one, and the time horizon over which that is measured. Only then does the conversation move to which channels can plausibly deliver at that cost.
Principle two
Attribution is agreed, not assumed.
When every platform reports its own attribution, the sum of claimed conversions exceeds the number of actual customers.
We agree a single attribution model up front, reconcile platform reporting against it, and report the difference rather than quietly picking whichever number looks best. Budget decisions then rest on one version of the truth.
| Common failure | What we do instead |
|---|---|
| Each platform's self-reported conversions taken at face value | Reconcile to a single agreed model and report the variance |
| Last-click by default, unexamined | Model chosen deliberately for the buying cycle, and stated |
| Cost per lead as the headline metric | Cost per qualified outcome, tracked downstream |
Principle three
Partner quality beats partner volume.
A large partner list is easy to build and hard to defend. We recruit against the brief, require disclosure of promotional methods and traffic sources before activation, monitor for breaches, and remove partners who do not comply.
What we check before activation
- Who the partner is, and what they actually operate.
- How they generate traffic, including whether they use sub-affiliates and who those are.
- What claims their promotional material makes, and whether the proposition supports them.
- How they collect any personal data, and what permission is captured.
- Whether their audience is plausibly the audience in the brief.
Principle four
Permission is evidenced, not inherited.
Where a campaign generates leads, "consent exists somewhere upstream" is not an answer.
We require that permission is captured at the point of collection, with the wording that was shown, its version, the source, and the time. That record travels with the lead. Opt-outs are honoured across the programme and treated as standing until there is a fresh, evidenced agreement.
Principle five
Report the misses.
Reporting covers performance against the measure agreed at stage one, including what underperformed, what was removed and what changed as a result. A report containing only good news tells you nothing you can act on.