Holding your own in an agency attribution call
Five questions that work no matter how technical the room gets, and the one kind of answer that means the conversation has stopped being useful.
Three platforms, three answers, one budget to defend. Your GA4 export already holds the answer, and the first finding is free and never leaves your browser. Get your first finding free
By Joris van Huët, Founder & CEOUpdated 3 min read
Five questions carry a founder through any attribution conversation, and none of them require technical knowledge. They work because they ask about evidence rather than about method.
The five
| Question | What it surfaces |
|---|---|
| What is this number compared against? | Whether there is a counterfactual at all |
| How sure are we? | Whether the estimate is resolved |
| What share of our orders does this explain? | Coverage |
| Which channels can we not measure at this spend? | Whether the floor is acknowledged |
| What would change your mind? | Whether the position is falsifiable |
The fifth is the most useful and the least asked. An answer names a specific observation. A non-answer explains again why the current view is correct.
What a good conversation looks like
Your agency has information you do not: creative fatigue, auction dynamics, what they tried last quarter, what the platform reps have said. An aggregate model cannot see any of that, so a productive call combines their context with an independent read rather than pitting one against the other.
The framing that keeps it productive is in settling attribution debates with your agency, and the pre-agreement that prevents most disputes is in the lift number your agency will accept.
The one asymmetry to name once
Platform reporting comes from companies that sell media. That is a description of an incentive, not an accusation, and stating it plainly once at the start is better than implying it repeatedly. It is also the reason an independent read exists at all.
The response that ends the useful part
Escalating jargon in reply to a plain question. If you ask what a number is compared against and receive an explanation of the modelling architecture, the question was not answered. Ask it again, in the same words. A competent agency will recognise what happened and answer.
What to bring
Causality Engine reads one file, the Attribution paths export from GA4, and shows what each channel caused next to what last-click gave it. Bringing evidence rather than a complaint changes the tenor of the whole call.
If your agency wants to see the method beforehand, how it works is written in plain language.
The outcome to aim for
Not agreement on a number. Agreement on a test: which channel to hold steady, for how long, and what result would change the plan. That converts a disagreement into a scheduled answer, which is the only thing that ever settles these.
Related answers
Frequently asked questions
What should I ask my agency about attribution?
What the number is compared against, how sure they are, what share of orders it explains, which channels cannot be measured at current spend, and what would change their mind. The last one is the most useful and the least asked.How do I handle jargon in an agency call?
Ask the question again in the same words. If a plain question about what a number is compared against gets an answer about modelling architecture, the question was not answered, and a competent agency will recognise that when you repeat it.What outcome should I aim for in the call?
Not agreement on a number, which rarely happens. Agreement on a test: which channel to hold steady, for how long, and what result would change the plan. That turns a disagreement into a scheduled answer.
Go deeper: Incrementality testing, 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
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- 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.
- CounterfactualA counterfactual is what would have happened under conditions that did not occur, such as a customer's outcome had they not received the campaign they did receive. It is never observed, so every effect estimate has to stand in for it with a control group, comparable units or a forecast.
- ShopifyShopify is an ecommerce platform for creating online stores and selling products. Attribution modeling shows which marketing channels drive traffic and conversions within Shopify.