Anaplan projects for FP&A fail when the model infrastructure has not been designed to be a natural fit with the model import data.
We have learnt this over the course of many health checks and model refurbs having seen the same mistakes made repeatedly across all sectors.
❌ The level of detail contained in actuals data and other imports is incompatible with forecasting and/or budgeting!
➡ Lists and hierarchies almost always need to be updated and mapping tables maintained manually when import data and modelling requirements are not aligned. This is mainly because these lists often do not contain codes or a natural logic to design self-governing mechanisms so must be managed externally.
❌ Fewer, longer and deeper hierarchies are used!
➡ The model design gets itself tied up in knots as these lists tend to feel forced and unnatural as they have been designed to maximise modelling efficiency rather than to enhance user experience or model effectiveness. Never sacrifice effectiveness for efficiency especially if the user experience is compromised.
❌ Excessive use and reliance on list sub-sets
➡ The lack of a natural, self organising structure often mean a model needs to use many more list sub-sets to differentiate and target parts of hierarchies for specific processes and calculations. These often need to be managed manually as the logic required to differentiate members is often lacking in the data model design.
Thankfully, we have also discovered a few key design approachs which help us overcome these issues.
✅ Align imported data with forecast/budget requirements - match detail
◼ Build lists based on your imported data
◼ Incorporate forecast/budget lists into hierarchies using these new lists
◼ Only use imported data that is relevant to the modelling requirements
✅ Use shallower and shorter hierarchies - use parallel hierarchies
◼ Create core lists from imported data and keep these flat
◼ Create processes to build multiple parallel hierarchies from these flat lists
◼ Utilise flat, balanced and ragged hierarchies
◼ Ensure all leaf level or terminal list items use common codes
◼ Design parallel hierarchies which fit naturally with the organisation
◼ Use these natural hierarchies to reduce the need for list sub-sets