Method note / Causal measurement
The sales you were going to get anyway
Some of the sales you credit to marketing were coming anyway. A practical way to separate reported conversions from the demand your spend actually caused.
Read the methodTangentix Research / Methods and field notes
The methods behind the numbers you'll have to defend.
Research mandate / 01
Anyone can hand you a conclusion. We publish the working, and the standard we hold it to. Read it, argue with it, take it to your own team.
No trend summaries. No AI spectacle. No precision the evidence can't carry.- Editorial owner
- Tangentix Research
- Publication unit
- 4 method papers
- Record floor
- Revision, review date, and primary-source check
- Attribution boundary
- Organization-led publication until individual approval
Research domains
Four places your reporting runs out of answers.
- 01
Causal measurement
Decision it informsHow much of that reported return did you actually earn?Something moved. The hard part is knowing how much of it you actually caused.
- 02
Media economics
Decision it informsWhich channel has earned your next dollar?Your channel report tells you what happened on average. Your next dollar needs to know what happens next.
- 03
Enterprise experimentation
Decision it informsWhich journey constraint should enter the experiment portfolio?Everyone has a theory about the drop-off. Only a real test settles it.
- 04
AI discovery
Decision it informsWhich evidence gap is suppressing AI recommendation?Being visible and being recommended are not the same thing. This is the gap between them.
Latest paperR / 001Jul 10, 2026
Inspect the working evidence behind the method.
A real, anonymized deliverable. These are the checks we run before we let anyone call demand incremental.
Approval record Real methodology / anonymized evidence
Decision docket Approval record / source deliverable A recommendation remains distinct from authorization and realized outcome.
Source record / TX-BRIEF-INC-01 Executive decision record Curated source deliverable / decision artifact, not a live product view
Open evidence record Executive brief / TX-BRIEF-INC-01
- Artifact
- Executive brief
- Evidence status
- Real methodology / anonymized evidence
- Record ID
- TX-BRIEF-INC-01
- Record
- Sanitized client methodology / values withheld
- Decision question
- How much of that reported demand did you earn, and how much was coming anyway?
A real, anonymized deliverable. Five independent checks have to agree before we'll call anything incremental.
Publication register
You can check our work on every paper here.
The complete register, including the paper above. Every entry keeps the method behind it, the revision it's on, and the date we last checked its sources, so you always know how fresh the thinking is.
Method note / Causal measurement
Read The sales you were going to get anyway
Research brief / Media economics
Read ROAS is not a capital-allocation system
Field note / Enterprise experimentation
Read A purchase path is not a funnel
Research brief / AI discovery
Read AI visibility is a citation system
The Tangentix standard
If a finding can't change a decision, we don't publish it.
- 01 / Counterfactual
- What would most likely have happened anyway?
- 02 / Uncertainty
- Which assumptions, if wrong, would flip the conclusion?
- 03 / Commercial hurdle
- Is the effect big enough to be worth your money?
- 04 / Action
- What should you fund, stop, or test on Monday?
Apply the research
Bring us a question your reporting can't settle.
Tell us what you're trying to decide. We'll work out the strongest evidence we could realistically build for it, and tell you honestly whether it's worth going deeper.
Work with us