Synthetic Data — Analytical Simulation, Not Real Company Results

Inspect the evidence.

Executed Python analytical modules · fixed seed 42 · 10,000 modeled accounts · 90 calendar days. Explore chart hover, zoom and exports. All measurements describe the supplied model and its assumptions.

01 / Mature cohort retention

Weekly signup cohorts; blank cells have no eligible users. Partially mature cells include only eligible accounts. Daily active retention, pooled arms.

02 / Meaningful retention curves

Two or more successful AI requests on an exact user-age day; denominators shrink with administrative observation time.

03 / Calendar activity

DAU reflects cumulative acquisition and modeled weekday variation; it is not an experiment treatment estimate.

04 / Return frequency

All mature registered users, including never-active accounts. First 30 user-age days. Histograms avoid active-user conditioning.

05 / Engagement distribution

Successful funded requests in the first 30 days; tail values are preserved.

06 / Revenue distribution

Cash receipts, not recognized accounting revenue. Most free accounts generate no receipts; renewals after age29 are excluded.

07 / Segment profitability

Exploratory first-30-day contribution per mature account. Small subgroup results are not powered rollout recommendations.

08 / Conversion milestones

Registered and billing counts use mature ITT accounts. These are marginal milestones; paid is not necessarily a subset of repeat-use accounts. Inspect ordered SQL funnel separately.

09 / Economic sensitivity

One factor at a time, three seeds per value; plotted mean and range reflect model response, not real causal evidence. Changing daily amount changes treatment entitlement.