Business Health
Wikipedia as a Knowledge-as-a-Service (KaaS) platform. This page applies business metrics to Wikipedia's community data, treating editors as the workforce, edits as output, and bytes as revenue.
1. Community Composition
How is the editing workforce distributed across activity levels? The five edit-count bands scale with the selected period: the power threshold is
How is this calculated?
Tier(editor, period) = bucket(period_edit_count, period_months × [1, 5, 25, 100])
Each editor matching the selected user types is classified once per calendar period. The monthly power threshold is 100 edits, the quarterly threshold is 300, and the annual threshold is 1200. All lower boundaries scale by the same 1×, 5×, 25×, and 100× monthly rates. The stacked area is a workforce snapshot, not a progression funnel; an editor may appear in different tiers across periods.
2. Acquisition Funnel
SaaS equivalent: Conversion funnel. For each cohort year of new registered editors, what fraction ever reached 5+, 25+, or 100+ cumulative edits? Unlike the composition chart above, this tracks individual editors across their full lifetime. Declining conversion rates signal that newer cohorts are less likely to become committed contributors, though recent cohorts also have less time, so expect a natural drop-off for the last few years.
How is this calculated?
Conversion(milestone) = Editors ever reaching milestone / Cohort Size × 100%
Each registered editor is assigned to a cohort based on their first-ever edit year. Their total cumulative edits across all time are counted, and we check whether they ever crossed 5, 25, or 100 edits. The conversion rate is the fraction of the cohort that reached each milestone. Recent cohorts (last 2-3 years) naturally show lower rates because editors haven't had enough time to accumulate edits. Note: this metric always uses registered editors regardless of the user type filter.
3. Lifetime Contribution Value (LTV)
SaaS equivalent: Lifetime Value. Instead of revenue, LTV is the estimated total net bytes a typical editor from each cohort contributes over their entire tenure, accounting for attrition. Older cohorts have higher LTV because survivors compound their contributions year over year.
How is this calculated?
LTV(cohort) = Σ over years (Survival Rate × Avg Net Bytes per Editor)
For each cohort year, LTV = sum over all years of (cohort survival rate x average net bytes per editor in that year). Average bytes per editor is computed from article namespace (ns 0) data for the selected user types. Cohorts with fewer than 10 initial editors are excluded.
4. Edit Survival Rate (Longevity Index)
SaaS equivalent: Product quality / NPS. Instead of measuring how long an edit "lives" before being overwritten, we track the fraction of edits that survive (are not reverted). High survival means consensus-driven, high-quality writing. A decline signals rising edit wars or declining content standards.
How is this calculated?
Edit Survival = (Total Edits − Reverted Edits) / Total Edits × 100%
Edit survival rate = (total edits - reverted edits) / total edits. The dashed green line at 90% marks a healthy threshold. Respects selected user types and namespaces.
5. Net Knowledge Retention (NRR Proxy)
SaaS equivalent: Net Revenue Retention. Are existing power editors producing more over time? This shows the output (net bytes) by activity tier. If the top tier's output grows even as casual editor output shrinks, the platform has strong net retention: its most engaged users are getting more productive.
How is this calculated?
Net Bytes(tier) = Σ byte_diff WHERE activity_tier = tier
Net bytes produced by each activity tier per period. Positive growth in the top tier over time indicates strong net retention. Respects selected user types.
6. Controversial Equilibrium Score
A metric that tracks the ratio of talk page edits to content page reverts. Healthy community governance is marked by more discussion and fewer reverts. A high talk-to-revert ratio signals that editors resolve disputes through consensus; a low ratio signals a "toxic" environment where edit wars replace deliberation.
How is this calculated?
Equilibrium = Talk Page Edits / Content Page Reverts
Talk page edits (namespaces 1, 3, 5, 7, etc.) divided by content page reverts (namespaces 0, 2, 4, etc.). A ratio above 1 (dashed line) means more words are spent discussing than reverting. This metric always uses all namespaces regardless of the namespace filter, but respects the user type filter.
7. Productivity by Activity Tier
Bytes per editor broken down by activity tier reveals whether power editors are becoming more or less efficient. If the top tier produces declining bytes per editor, it may signal burnout, increasingly administrative work, or a shift from content creation to maintenance.
How is this calculated?
Bytes per Editor = Net Bytes / Unique Editors (per tier, per period)
Net bytes divided by unique editors within each activity tier per period. Respects selected user types.