Three case studies, each from question to recommendation. Data type is stated on every one.
Real public data · 110,521 appointments
A SQL analysis of real outpatient appointments: the no-show rate rises from 4.6% (same day) to 33.0% (booked 31+ days ahead), and bookings made 8+ days ahead drive 57% of all no-shows. Includes the SMS confounder, charts and tested recommendations.
Synthetic data
Funnel and cohort analysis in SQL: 62% of carts never reach checkout. Lifting cart-to-checkout to 48% would add about 27% more orders on the same traffic (scenario). All numbers are synthetic and labelled as such.
Anonymised · imcure
Problem, as-is vs to-be process map and recommendation for a manual KYC onboarding flow. Manual verification time fell 30% at imcure (measured); all other figures are labelled assumptions.