Describe the problem or experience related to your idea:
In inventory planning models, Target Stock Levels are commonly calculated using a Negative Binomial distribution to model demand variability and set safety stock/service level targets. This is standard practice because demand is typically overdispersed (variance exceeds mean), making Negative Binomial a more accurate fit than Normal or Poisson distributions.
However, Anaplan has no native Negative Binomial function, so this currently must be built using manual workaround logic (e.g., approximating the cumulative distribution function through iterative line item formulas, lookup tables, or custom approximation algorithms). This workaround is calculation-intensive and significantly slows down model performance — increasing calculation time noticeably whenever Target Stock Level recalculates, especially across large SKU/location combinations. This impacts not just the speed of this calculation, but overall model responsiveness and the practicality of running frequent what-if scenarios, space consumption and procedure complexity. The Negative Binomial is today representing 11,6GB for the today TSL and, according to Ilaria a native formula could save 10 GB. From a complexity point of view, today the procedure is split in 15 modules (cf. Print screen – excluding the modules for the projection)
How often is this impacting your users?
Continuously — Target Stock Level is a core, frequently recalculated metric (daily/weekly operational cadence plus ad hoc scenario planning), so the performance hit is felt on a near-constant basis, not just during periodic planning cycles.
Who is this impacting?
- Model builders, who must maintain and optimize complex workaround logic
- Solution architects, who must design around the performance constraint at scale (e.g., across large SKU/DC combinations)
- Business/end users, who experience slow calculation times and delayed insights when adjusting scenarios
- Executive-level business users, who may receive delayed updates to inventory targets during time-sensitive decisions
What would your ideal solution be? How would it add value to your current experience?
A native, optimized Negative Binomial function (e.g., NEGBINOM.DIST, mirroring Excel's function) built into Anaplan's calculation engine — rather than relying on manual approximation logic — would dramatically reduce calculation time, running as a single optimized native function instead of multiple iterative/array-based line items [DL] and space consumption.
This would add value by significantly improving model performance and recalculation speed for Target Stock Level and similar inventory/demand planning calculations, simplifying model design and maintenance by removing complex workaround logic, and enabling faster, more frequent scenario planning for inventory and service-level decisions.
Please include any images to help illustrate your experience:
(Suggested attachments: a screenshot of a workaround formula/module structure showing iterative approximation logic; and/or a calculation time comparison — e.g., recalculation time for Target Stock Level before/after disabling such a module — to quantify the performance impact.)
