Conflicting performance reports
Agree metric definitions, data sources, and comparable periods before interpreting the numbers.
Data analytics and product insights
Understand what your product data and customer feedback can tell you before deciding what to change. ACW structures measurement, examines patterns, and connects findings to product questions, with the limits of the evidence visible.
Discuss your analytics questionStart with the decision you cannot make
Organise the evidence around the question
Data structure
Define the metrics, source fields, and population being measured.
Behaviour analysis
Examine journeys, cohorts, and changes in usage where the available data supports them.
Feedback themes
Group repeated issues while preserving their context and source.
Decision priorities
Identify findings that support action and questions needing more research or testing.
Match the analysis to the product context
SaaS
Define activation, returning usage, and retention for the product’s actual use cycle.
AI products
Examine task completion, output review, failures, and feedback alongside usage.
Early products
Identify which signals are observable now and which assumptions remain untested.
Partner delivery
Agree client-specific definitions and reporting responsibilities rather than applying one template to every product.
What the work produces
Defined metrics and documented source limitations.
Reports, dashboards, or tracking work within scope.
Findings with supporting data and relevant segments.
Feedback themes linked to source observations.
Recommended actions, hypotheses, and follow-up questions.
Choose the evidence you need
From data collection to a decision
Agree the question and permitted sources. Review completeness and definitions, then analyse the relevant period and population. Discuss findings with the team and record what is observed, what is inferred, and what should be checked next.
Keep the basis of every finding visible
1/4
1/4
A decision first
Start with what the team needs to decide.
Questions before starting
Do we need new analytics tools?
Review the existing data and collection gaps first. Tool setup should follow the measurement requirements.
Can you analyse a small amount of data?
Yes, with conclusions limited to the available evidence. Small samples often identify questions rather than establish a stable trend.
Will this tell us what to build?
It supports prioritisation. Delivery cost, strategy, and further user research may still affect the decision.
Bring the question behind the report
Tell us what you need to decide and which data or feedback you have today.