Six talent agencies attribution methods compared for UK reporting

What to take away

• Most agencies default to platform-reported conversions because they are easy to screenshot, not because they are accurate.
• Codes and tracked links are cheap and auditable, but they miss view-through influence entirely.
• Post-purchase surveys capture what codes cannot, yet they depend on sample size and honest answers.
• No single method survives a client review on its own. Pair two, and state which leads.
• Agree the method in the contract before launch, not after the invoice.

Why single-source numbers fail

Attribution arguments start when the numbers do not match. The platform reports one figure, the client's analytics another, and the agency ends up defending a gap it cannot explain.

The gap is structural. Each method counts a different event: a redemption, a click, or whatever the platform defines as a conversion on its own timeline.

Two reports can both be right and still disagree. Each counts a different moment, and platform dashboards often update after the client's own numbers are frozen.

The measurement and reporting guide for 2027 treats method choice as a contractual question rather than a dashboard setting. If the method is not named in the agreement, the argument is already lost.

Six methods compared

Method	What it counts	Main weakness
Promo codes	Redemptions	Undercounts, favours loyal audiences
Tracked links	Clicks and sessions	Blind to view-through
Platform-reported conversions	Platform-defined events	Definitions and delays vary
Post-purchase surveys	Self-reported source	Small samples, recall error
Incrementality tests	Difference against a holdout	Costly, slow, needs volume
Blended modelling	Weighted mix of the above	Depends on input quality

Codes and tracked links are the cheapest to run and easiest to audit. Both reward the same behaviour: an audience already primed to buy. A creator with a slow-burn, high-consideration audience looks weak on both.

Platform-reported conversions are convenient, but the definitions sit with the platform. The Advertising Standards Authority publishes rulings showing how endorsement claims are adjudicated in the UK, a reminder that a number and a claim are different things: see the ASA rulings database.

Surveys fill the view-through gap. Incrementality tests answer the harder question of what would have happened anyway. Blended modelling combines them, but inherits every weakness of its inputs.

Cost divides the options as well. Codes and tracked links cost almost nothing to run, while incrementality work needs volume, a holdout and time before it answers anything.

Choosing between them

1. Write down the decision the number will inform: media plan, renewal or fee negotiation.
2. List the methods already in your stack. Do not buy a tool to answer one question.
3. Pick one lead method and one supporting method, and state the lead in the report.
4. Set the reporting window and settlement date in the contract.
5. Run a small incrementality test on your largest account before rolling it out.
6. Publish the method, the window and the known gaps in every client report.
7. Review the choice each quarter against actual spend and outcomes.

Step three is where most teams stall. Two methods that disagree beat one method presented as fact, provided you say which one you trust and why.

A written method note prevents more arguments than any dashboard. One page naming the lead metric, the reporting window and the exclusions is usually enough.

Evidence a client will accept

The strongest evidence pack combines a lead metric, a supporting metric and a written explanation of the gap between them. Clients rarely reject a gap. They reject a gap nobody mentioned.

Where the two methods disagree, show both numbers and explain the difference plainly. Finance teams accept uncertainty they can see.

Digital marketing coverage, such as the digital marketing news and analysis from The Drum, shows how creator spend sits inside wider digital budgets, which is often the frame a finance team uses.

If the client is a public body, procurement rules shape what can be bought and how it is evidenced. Read the guidance on procurement in the national interest before proposing a measurement package to a public sector buyer.

Benchmarks matter too. Comparing your results against sector norms is only fair if those norms used the same method. The benchmark research on English agencies explains its sampling before quoting any figure.

Common questions

Which attribution method should a small agency start with?

Start with tracked links plus a post-purchase survey. Both are cheap, and together they cover click-based and view-through influence without a modelling budget.

Can we report platform numbers as fact?

Only if the platform's definition and reporting delay sit alongside the figure. Otherwise label it platform-reported, not verified.

How often should the method be reviewed?

Quarterly works for most rosters. Review sooner if a client changes its analytics setup or a platform changes its conversion definition.

Do public sector clients need a different method?

Often yes. Procurement rules can require auditable evidence, so agree the method before the tender rather than after the award.