Learning guide and study notes
What does the evidence show? Use it to decompose a measurable decision.
Analytical thinking breaks a complex, measurable question into verifiable components—denominators, baselines, segments, and drivers—so decisions rest on sound quantitative logic rather than headline averages.
Make the question, data, assumptions and uncertainty visible before trusting a number. The three moves below adapt the UK AQuA guidance; they are not an official named framework.
Work through these steps in order, applying the tools above. Write down the answer to each step before moving on.
Specify the outcome, comparison, scope and intended use of the analysis. [1] Source: UK Government — The AQuA Book (opens in a new tab)
Check data quality, units, baselines and assumptions; validate calculations against the question. [1] Source: UK Government — The AQuA Book (opens in a new tab)
Show limitations and sensitivity before making a recommendation; revisit it as evidence changes. [1] Source: UK Government — The AQuA Book (opens in a new tab)
Pick a rising metric. Define its baseline, denominator, segments and one alternative explanation; say what additional data would change your conclusion.
Situation: Log storage is getting expensive.
How to apply: Separate data growth, retention period and unit price; compare options using the same workload assumptions.
Outcome: Present the cost and retrieval trade-off with an explicit uncertainty range.
Situation: Weekly average response time stays flat, yet users report frequent timeouts during peak hours.
How to apply: Split the aggregate average by percentile (p50, p95, p99), time window, and endpoint mix to expose tail spikes.
Outcome: Target the specific slow endpoint during peak concurrency instead of scaling the entire service tier.
Situation: Submitted support tickets fall by 25% the week after a new intake portal launches.
How to apply: Compare portal completion rates, abandoned sessions, and direct chat escalations using the same user denominator.
Outcome: Fix a broken validation field that was blocking legitimate requests rather than celebrating a false improvement.
Illustrative examples showing how the lens changes decisions; not measured outcomes.
Watch out: A precise-looking figure can conceal weak inputs or a misleading denominator.
Combine this lens with others rather than relying on it alone. Browse all thinkings
A number is useful only when its definition, comparison, assumptions, and uncertainty fit the decision.
Use a problem of your own. Write a short answer to each prompt, then return after trying the method.
Answer from memory before opening the suggested answer.
Separate unit price from volume, retention, workload mix, and the comparison period; use consistent units and scope.
A range can show how uncertain inputs affect the decision. Explain the assumptions behind the bounds rather than implying unjustified precision.
These sources support the method; the examples and exercises are original teaching illustrations.