Skip to main content

For growth operators

Follow the signal past the funnel.

An acquisition answer matters only if the next experience can use it appropriately. Make the journey, changed response and observed behavior inspectable together.

Scope your API pilot Follow the workflow ↓

Acquisition → onboarding

Keep the source attached to the intent.

A campaign click is not the same as a stated goal. Separate inferred acquisition signals from explicit onboarding answers and agree which can inform the product.

Onboarding → product

Change a response you can explain.

Choose one consuming surface: a recommendation, a next step or a return interaction. Keep the default response alongside the proposed treatment and let the customer correct the relevant context.

Product → observed response

Instrument the action, dismissal and correction.

Define journey events and their relationship to the response. Review whether the product used the intended signal before attributing any downstream conversion change.

Measurement choice

Choose the attribution design before reading the result.

Where feasible, use a randomized treatment and control. Otherwise scope a cohort or before/after comparison and record selection effects, concurrent campaigns and other limits. An observed purchase does not establish causal lift.

Fictional example · illustrative, not a customer result

Journey and attribution note

Compare the current handoff with a proposed, source-connected record.

Before · context lost

  • Acquisition: campaign for returning practitioners.
  • Onboarding: “I want a brief routine.”
  • Product: generic recommendation; purchase recorded separately.

After · proposed pilot artifact

  • Permitted signal: explicit routine preference, linked to its onboarding source.
  • Treatment: brief-routine recommendation; default retained for comparison.
  • Observed events: recommendation shown → accepted/dismissed/corrected → purchase.
  • Attribution choice: randomized comparison if feasible; record limits otherwise.

Review question
Can an analyst reconstruct what was shown and why, without treating correlation as lift?

Prepare the conversation

Bring the work. Agree on what to evaluate.

Customer inputs

  • Campaign and onboarding sources with permitted-use boundaries
  • The product surface and default experience
  • Event definitions, identity linkage and a measurement owner

Evaluation questions

  • Which signal is stated and which is inferred?
  • Can response exposure be connected to the observed event?
  • What concurrent changes limit attribution?

Scope the next step

Clarity API enterprise pilots

Custom quote · Scoped pilot

Explore company-brain and hyper-personalization use cases with your own data, boundaries and evaluation criteria.

Inputs, integrations, boundaries, deliverables and evaluation criteria are agreed during scoping.

Scope your API pilot Compare the offers →