Forecasting the SAP true-up: a method that survives the audit

The SAP true-up shock is the most predictable surprise in enterprise procurement. The finance team budgets the SAP line at last year’s number plus inflation. The SAM team measures consumption in the last quarter of the year. The number comes in materially over budget. The escalation goes to the CIO, to the CFO, to the audit committee. The vendor benefits from the panic; the buyer pays the price. The pattern repeats across most large SAP estates, every year, despite the data and the people needed to forecast the number being already in the building. The fix is a forecasting method that is run quarterly, produces a single artefact, and is calibrated to the audit context rather than to the budget cycle.

Why the standard finance forecast fails on SAP

The finance forecasting cadence works for stable, contract-defined spend lines. SAP is not that. The SAP spend line moves with three variables that finance forecasting does not see directly. Named-user growth on existing engines moves with headcount, with M&A activity, and with role-mapping decisions inside HR and IT. Engine metric consumption moves with business volume — orders, materials, deliveries, financial documents — on patterns that the procurement category manager does not see month to month. Indirect-access exposure moves with the integration estate and with third-party document creation, which sits on the architecture team’s register but not on procurement’s.

The result is that the SAP line, when forecast from finance, is wrong systematically and predictably. The over-runs are concentrated in the third and fourth quarters, when measurement data lands and finance sees for the first time what the actual position is. By that point the negotiation window with the vendor has narrowed and the price for the gap is higher.

The four inputs the forecast must capture

A forecasting model that produces a defensible number captures four inputs. The first is the named-user run-rate — the count by licence type, broken out by entity, with the headcount and role-mapping assumptions explicit. The second is the engine consumption run-rate — HANA memory, BW units, SuccessFactors employee count, Ariba spend under management, Concur transactions, every contracted engine on its own metric. The third is the integration document volume — the count of documents created by third-party systems in SAP, on the methodology covered in the digital-access document counting article. The fourth is the contracted entitlement against each of the first three.

The forecast is the gap between the run-rate trajectory and the entitlement, projected to year-end, with a confidence interval. That is the artefact. Everything else is supporting data.

The quarterly cadence

The forecast is refreshed quarterly. Quarter one establishes the baseline trajectory. Quarter two reads the actuals against the baseline and adjusts the projection. Quarter three is the critical reading; this is where the gap to entitlement becomes visible and the negotiation strategy is set. Quarter four is the close-out and the input into the next year’s contract position. Running the cadence quarterly, rather than annually, gives the procurement team time to act on the gap before the year-end forces the negotiation onto vendor-favourable timing.

The named-user projection

The named-user projection takes the current count by licence type and applies three growth assumptions. The underlying headcount growth, drawn from the corporate plan. The role-mapping drift — the rate at which existing users get upgraded into more expensive licence types as their roles evolve. The new-system rollout — planned project go-lives that will create new named users. Each assumption is documented, with the source and the responsible owner. The pattern of named-user buckets and role mapping is covered in the named-user buckets article and the role mapping article.

The output is a projected named-user count by licence type, by quarter, to year-end. The gap against the entitlement is the named-user true-up exposure.

The engine consumption projection

The engine projection is the harder of the four inputs because the engine metrics are not all measurable on the same cadence. HANA memory is measurable continuously. BW units are measurable on the monthly snapshot. Ariba spend is measurable quarterly. SuccessFactors employee count is measurable at the headcount snapshot. The model normalises each engine to a quarterly run-rate and projects to year-end on the same assumptions as the underlying business volume. The methodology is covered in the SAP S/4HANA topic page for the S/4HANA engines and in the engine metric deep-dive white paper for the older ECC engines.

The integration document projection

The integration document projection takes the digital-access baseline (built using the methodology in the digital-access counting article) and projects forward on the business-volume assumptions for the integrations in scope. New integrations planned for the year are added with their expected document volume, sourced from the integration design documents. Decommissioned integrations are removed. The output is a projected document count for the year against the contracted Digital Access tier. The gap is the digital-access true-up exposure.

The confidence interval

Each of the three exposures is reported with a 90 per cent confidence interval. The intervals are not arithmetic flourishes — they are the negotiating range. A point estimate that says ‘we will be eight per cent over on named users’ gives the vendor a number to negotiate against. A range that says ‘the named-user gap will land between four and twelve per cent over, ninety per cent confidence’ gives the buyer the lower bound as the negotiating starting point and the upper bound as the planning reserve. The discipline of always reporting the range, rather than the point estimate, materially changes how the conversation goes when the year-end approaches.

What the forecast looks like when the audit arrives

The forecast, run quarterly and documented in a single repeating artefact, becomes the defensive evidence base when the audit notification arrives. The audit team will ask for the buyer’s reading of named-user counts, engine consumption, and integration document volume. The buyer who has the forecast in hand can produce, within the first two weeks of the engagement, a documented internal record that shows the position was understood, was monitored, and was being managed against the contractual entitlement. The audit settlements that arrive on this footing close materially lower than settlements where the buyer is reconstructing the position from raw extracts under audit pressure. The pattern is consistent with the broader license compliance assessment work we run.

The monthly tracking against the quarterly forecast

The quarterly forecasting cadence works because it is supported by lighter monthly tracking. The monthly pack reads the named-user count, the engine consumption snapshots, and the integration document volume against the quarterly forecast. The variance signals whether the quarterly model is still calibrated. Monthly variances within the confidence interval require no action. Variances outside the interval trigger an interim re-forecast. The discipline is borrowed from financial planning practice and applies cleanly to the SAP licence position. The monthly tracking takes one to two days of analyst time per cycle once the model is in place.

The tooling question

The forecasting model can be built in spreadsheets, in a SAM tool, or in a custom analytics environment. The choice matters less than the discipline of running it on a fixed cadence with documented assumptions. Spreadsheet models are easiest to build and easiest to defend; the assumptions are visible to anyone who opens the file. SAM tool models capture the data flows automatically but often hide the assumptions behind tool-specific logic that the audit team will not accept without explanation. Custom analytics environments are appropriate at the largest scale but carry build cost. For most enterprises, a spreadsheet model run by a senior SAM analyst on a quarterly cadence outperforms a tool-based model that no one fully owns. The pattern is covered in the SAP S/4HANA topic page for S/4HANA-specific engines and in the usage analytics article for the analytics dimension.

The scenario analysis on the forecast

The forecast carries more analytical weight when it is presented with two or three scenarios alongside the central case. A conservative scenario that holds volume flat. A central scenario aligned with the corporate plan. An aggressive scenario that captures the upper end of the planning range. The scenario set frames the negotiation reserve and the contingency budget. When the audit notification arrives, the scenario analysis is what the audit team reads to understand how the buyer is thinking about the position — and the discipline of presenting the central case alongside the alternatives consistently produces better engagement than a single point estimate that the audit team can challenge in isolation.

For SAP estates over twenty-five million in annual licence spend, the forecasting cadence pays back many times over against the year-end true-up risk. The insurer named-user reclassification case file documents an engagement where the forecasting artefact carried the matter through a contested audit.