Autodesk: 80% Faster Revenue Forecast Roll-Up with Anaplan

Autodesk began its subscription transition in 2016, committing to eliminate perpetual license sales entirely by 2021. This wasn't a simple pricing change — it fundamentally altered every aspect of the financial model. Revenue recognition shifted from point-of-sale to ratable recognition over subscription periods. The key metrics changed from units sold and average selling price to ARR, churn rate, net dollar retention, and customer lifetime value. The legacy spreadsheet models, built over decades to track perpetual license performance, were structurally incapable of representing the new SaaS model accurately.

The pain was acute: during peak business model transition, Autodesk's finance team was simultaneously producing two completely different views of the business — the old perpetual metrics for historical comparisons and the new SaaS metrics for forward-looking management. Each required separate models, separate data pulls, separate reconciliations. The manual workload was consuming the team's capacity entirely.

• 500+ people touched the annual forecast process across global business units
• Revenue forecast roll-up took 3+ weeks per cycle at peak
• Expense forecast variance regularly exceeded ±$1 million per quarter
• Parallel modelling of perpetual and SaaS metrics doubled the data processing burden
• Multiple versions of 'final' forecasts circulating simultaneously created decision confusion

Autodesk evaluated several enterprise planning platforms before selecting Anaplan. The decisive factor was Anaplan's 'connected planning' architecture — the ability to build separate models for different planning domains (ARR, renewals, expenses, workforce) that automatically propagate changes through the hierarchy. When the ARR model updates, the P&L model refreshes; when the P&L updates, the cash flow model refreshes.

This connectivity eliminated the manual 'stitching together' that had consumed so much finance team time. Previously, updating a revenue assumption in one spreadsheet required manually updating 8-12 other linked files. In the Anaplan architecture, one update cascades automatically through all 18 connected models.

Autodesk's implementation wasn't a single monolithic model — it was an ecosystem of 18 connected models, each responsible for a specific planning domain. This modular approach allowed different teams to own their domain while automatic connections ensured consistency across the consolidated view.

• ARR waterfall model: new logo, expansion, and churn projections by product and geography
• Renewal forecasting model: renewal probability by contract cohort and customer health score
• Capacity planning model: headcount and quota capacity for the sales organisation
• Expense planning models (8): one per major cost centre, all connected to the consolidated P&L
• Geographic rollup models: regional P&Ls aggregating from 50+ country-level inputs
• Consolidated P&L model: the single source of truth that all other models feed into

Autodesk resisted the temptation to build all 18 models simultaneously. Instead, they followed a deliberate phased implementation: start with the highest business-value use cases, prove success rigorously, then expand. Phase 1 (months 1-4) focused exclusively on ARR waterfall and renewal forecasting — the two metrics that mattered most to investors and management. Phase 2 (months 5-9) added expense planning for the top 3 cost centres. Phase 3 (months 10-18) completed the remaining models.

Each phase had clearly defined success criteria established before go-live: a specific accuracy improvement target, a cycle time reduction goal, and a user adoption rate. If any metric missed target, the phase was held open until it was achieved. This discipline prevented the 'good enough' creep that derails many enterprise software implementations.

• Phase 1: ARR waterfall + renewal forecasting — highest strategic value, validates architecture
• Phase 2: Top 3 expense cost centres — builds finance team confidence with lower-complexity use cases
• Phase 3: Remaining models + geographic rollups — completes the connected planning ecosystem
• Each phase gate: accuracy target met? Cycle time target met? User adoption rate above 85%?

The headline results — 80% reduction in forecast roll-up time and 97% improvement in expense forecast variance — were achieved within two quarters of completing Phase 3. But these numbers don't fully capture the scope of the transformation. The more profound change was qualitative: what the finance team was doing with its time changed completely.

• Revenue forecast roll-up: 3+ weeks → 4 hours (80% time reduction)
• Expense forecast variance: ±$1M → ±$30K per quarter (97% accuracy improvement)
• Analyst time split: 80% data gathering / 20% analysis → reversed to 20% / 80%
• Finance team headcount: reduced through natural attrition while output volume increased 3×
• Board reporting cycle: condensed from 10 days to 3 days post-close

Autodesk's Finance leadership has shared the implementation lessons that they'd prioritise if starting over. These are relevant regardless of which planning platform you use or what industry you're in.

• Define your metrics framework completely before building models — every undefined term becomes a reconciliation problem
• Get business unit buy-in before go-live, not after — the tool only works if business partners use it to make decisions
• Start with the highest-visibility use cases to build executive sponsorship — visible wins fund the next phase
• Build in a 'parallel run' period where both old and new processes run simultaneously — it's painful but essential for trust
• Invest in training before go-live, not during — users who feel confident are far more likely to adopt
• Document model logic exhaustively — models become institutional knowledge that outlasts the people who built them