A practical framework / Six steps
Turn a customer growth question into a repeatable analytical skill.
Start with the decision. Make the assumptions visible. Build something a customer can use—and a team can improve.
- Use it when
- A customer question keeps recurring
- Bring
- Context, available data and a decision owner
- Leave with
- A skill brief and a practical next step
Define the decision
Name the customer question, the person making the decision and the action the answer could change. Start with an observable business need.
Output: One decision, a decision owner and a proposed next action.
Capture the customer context
Record the audience, business model, priority products or pages, geography and meaningful business events. Preserve the terms the customer uses.
Output: A short context record and explicit scope.
Establish the measurement
Specify the population, unit of analysis, data sources, time window and refresh cadence. Identify unavailable data before designing the conclusion.
Output: Metric definitions, a baseline and known coverage gaps.
Specify the method
Explain how the inputs become a recommendation. Identify comparison groups, possible confounders and the conditions that should stop or qualify the analysis.
Output: A reproducible method with failure conditions.
Produce something actionable
Return a recommendation, prioritized work queue, experiment or client-ready explanation. Keep the supporting signals visible.
Output: A deliverable with reasons, scope and an accountable next step.
Improve through delivery
Review the recommendation with the customer. Capture what was useful, what was misunderstood and what happened after action was taken. Update the reusable method.
Output: A feedback record and the next version of the skill.
Illustrative worked example / AI search
Which buyer questions deserve our next content investment?
Imagine a B2B team choosing a small set of questions to investigate. Start with customer conversations and available search demand. Define a repeatable prompt set, observe the answers and citations, and connect each question with an existing or proposed destination.
Visibility
Presence and citation within a defined prompt set, engine and observation window. A sampled answer does not establish total market exposure.
Referrals
Observed visits attributed to identifiable AI sources. Missing attribution stays unknown.
Leads
Defined business events and CRM records, with known join and attribution limits.
Revenue
Opportunity or revenue outcomes where the CRM linkage supports them. Keep observed association separate from causal contribution.
Useful output: a short question-to-page backlog with evidence, missing information, a proposed intervention and a review date.
This is a framework example, not a historical client result or a claim that the four measures have been joined successfully.
Your working brief
Make the next question concrete.
Complete the fields for one real decision. Entries stay in this page for this visit; download or print a copy to keep them.
Before you run the workflow
Check the foundations.
0 of 6 items checked
A planning aid, not a release approval or evidence that the analysis has passed validation.