Test an attribution tool inside its refund window
Treat the refund window as the project deadline it is. Five checks in order, starting with the one that ends most evaluations before the second day.
€99 once, excl. VAT. Full refund within 30 days, no questions asked. You keep the read.
By Joris van Huët, Founder & CEOUpdated 3 min read
A refund window is a project deadline, and evaluations fail because nobody treats it as one. Five checks, in this order, fit inside almost any window and the first one ends a surprising number of evaluations on day one.
The five checks
| Order | Check | Fails if |
|---|---|---|
| 1 | Does it produce a number from your actual data? | The input requirements do not match what you can export |
| 3 | Does it agree with a period you already understand? | It disagrees wildly and cannot explain why |
| 4 | Does it name what it cannot measure? | Every channel gets a confident figure |
| 5 | Would it have changed a decision you already took? | You cannot name one |
Check one, on day one
Run your real export, not a sample. The most common evaluation failure is discovering in week three that the tool needs a data shape you cannot produce, or a collection period you have not run. Find that out in the first hour.
Check three is the one people skip
Pick a period where you already know what happened, ideally one containing a deliberate change: a channel you paused, a campaign that ran only in one market, a month with an obvious spend step. Run the tool on it.
A tool that reproduces something you already know is a tool you can extend trust to. A tool that disagrees may still be right, but it now owes you an explanation, and how it handles that request is itself the evaluation. This is the same logic as the placebo test.
Check four separates honesty from confidence
Feed it a channel with very little spend. A well-built method will decline to score it, because its effect cannot be separated from noise at that level. A method that returns a confident figure is filling the gap. The arithmetic is in the measurability floor.
Check five is the decision
Take a budget decision you made in the last six months and ask whether this tool's output would have changed it, or made you more confident in it. If neither, the tool is interesting rather than useful, and interesting does not survive a renewal review.
Diary the deadline
Put the last day of the window in your calendar on the day you start, with a reminder three days before. Refund windows are missed by drift, not by decision, and the vendor has no incentive to remind you.
What this looks like here
The €99 one-time read is designed for exactly this sequence: one export, what each channel caused next to what last-click gave it with a data-health score rather than a confidence interval, and a full refund within 30 days, no questions asked. No subscription is attached, so there is no window to miss. Pro at €299 a month, cancellable at any time, is a separate decision.
How to use GA4 exports covers check one.
The order matters
Run them in sequence and stop at the first failure. Evaluations that run all five in parallel produce a mixed picture and a decision nobody wants to defend.
Related answers
Frequently asked questions
What should I test first when evaluating an attribution tool?
Whether it produces a number from your actual export rather than a sample. The most common evaluation failure is discovering in week three that the tool needs a data shape you cannot produce or a collection period you have not run.How do I check whether an attribution tool is trustworthy?
Run it on a period where you already know what happened, ideally one containing a deliberate change such as a paused channel or a single-market campaign. Reproducing something you know is evidence; disagreeing is acceptable only with an explanation.Why do teams miss refund deadlines?
Drift, not decision. Put the last day in the calendar when you start, with a reminder three days before. The vendor has no incentive to remind you and evaluations expand to fill whatever time is available.
Go deeper: Causal attribution, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Have an audience of ecommerce operators? Earn 20% recurring commission by recommending the €99 read.
Keep reading
Terms in this article
- AnalyticsAnalytics is the systematic computational analysis of data. It reveals customer behavior and measures campaign performance.
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- Attribution DebtAttribution debt is the gap between what your ad platforms claim drove revenue and what actually caused it, carried quarter after quarter into the budget. It is how marketing debt accrues: allocate on claimed conversions long enough and the plan itself becomes the liability.
- CausalityCausality is the relationship where one event directly causes another, essential for identifying specific actions that drive desired outcomes in marketing.
- ClickClick is the action a user takes to interact with a digital advertisement, redirecting them to a website or landing page. Clicks are a fundamental metric for measuring ad engagement and a primary input for click-based attribution models.
- Confidence IntervalConfidence Interval is a statistical range of values that likely contains the true value of a metric. In marketing analytics, it quantifies uncertainty around estimates, indicating the precision of an outcome or causal effect.
- IncrementalityIncrementality is the part of an outcome, such as orders or revenue, that a marketing activity caused: the extra results compared with what would have happened without it. It is a causal effect, not the number of conversions credited to a channel.