How to answer the objection you cannot answer
Sometimes the sceptic is right and your data cannot settle the question. Three responses that keep your credibility, and the one that ends it.
By Joris van Huët, Founder & CEOUpdated 3 min read
Occasionally a sceptic raises something your data genuinely cannot settle, and how you respond in that moment decides whether anything you said before is believed. Three responses work. One does not.
The three that work
| Response | When to use it |
|---|---|
| "That is outside what this design can resolve" | The objection is about a mechanism the method cannot see |
| "Here is the test that would settle it" | The objection is empirical and testable |
| "You may be right, and here is what I would expect if you were" | The objection is plausible and you disagree |
The third is the strongest and the least used. Stating what you would expect to observe if the sceptic were correct turns a disagreement into a shared prediction, and shared predictions get checked.
The one that does not work
Answering a question you were not asked. Responding to "how do we know this is not just seasonality" with an explanation of the modelling approach reads as evasion, whether or not it is meant that way. If you do not have the answer, say the words.
The three objections that are usually correct
Seasonality, when the window spans a period with a known pattern and you have not adjusted for it. Say so and re-run on a comparable period.
A structural change the model cannot see: a competitor exiting, a pricing change, a supply problem. An aggregate method has no visibility of any of these, and the person raising it usually has direct knowledge. Concede it and note it as a limitation on that channel.
Creative and auction dynamics. A causal read on channel-level data cannot distinguish a channel that stopped working from creative that stopped working. That is a real boundary and it is worth stating before it is raised.
What to do with a testable objection
Convert it. "We can settle that by holding this channel dark in half our regions for four weeks" is a much better meeting outcome than winning the argument. The design is in the geo testing guide and the sizing questions in how to measure incremental lift.
Agreeing the test in the room also fixes the threshold in advance, which is what stops the result being relitigated later.
Why conceding does not weaken you
Because the alternative is defending a claim past what the evidence supports, and everyone in the room can see when that is happening. A report that names its own limits and then survives the parts it did claim is more persuasive at the end of the meeting than one that claimed everything and was chipped at throughout.
Channels it cannot measure are named rather than scored.
The phrase worth having ready
"This design cannot resolve that. Here is what would." It is short, it is honest, and it ends the exchange in a better place than any defence.
Related answers
Frequently asked questions
What should I say when I cannot answer an objection?
Say so plainly, then name what would settle it. Answering a question you were not asked reads as evasion whether or not it is meant that way, and it costs more credibility than the admission would have.Which objections to attribution are usually valid?
Seasonality when the window spans a known pattern, structural changes the model cannot see such as a competitor exiting or a pricing change, and the inability of channel-level data to separate a failing channel from failing creative.Does conceding a limitation weaken my case?
No. Defending past what the evidence supports is visible to everyone in the room. A report that names its own limits and holds on the parts it did claim ends the meeting stronger than one that claimed everything.
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.
- 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.
- Holdout TestA holdout test is an experiment that keeps a random part of an audience from seeing a campaign and compares its results with those of the part that could see it. The held-out part is the control group, so the gap in outcomes estimates what the campaign added.