History only forecasts the world as it was; product plans change the world. New features land extra load that no past trend captured, so a capacity plan built solely on system statistics is half-blind. User behavior — "engagement" in product terms — is another vein to mine: discussion posts per minute, posts/uploads per day per user. This is where capacity planning and product management connect. Corporate culture has historically walled product (plans the roadmap) off from engineering (builds and runs it); both forecast, but for different ends, and their data should tie together. Allspaw's best practice: a capacity planner should keep an **ongoing conversation with product management** to know the feature timeline. Having enough capacity is an engineering requirement on par with development time and resources — and the binding metric often *is* a user-interaction quantity (recall the photos-to-user ratio), which only product context fully explains. --- *Source: [[The Art of Capacity Planning]] (John Allspaw, O'Reilly 2008) — Ch 4 — Predicting Trends*