ClariLayer Docs

Forecast Accuracy

Reference

Forecast Accuracy Definition

Learn the governed definition of Forecast Accuracy, including forecast snapshots, actual-result basis, variance methods, and ClariLayer Drift Risk.

Back to metric library

Metric 16 of 16

Sales pipeline

Forecast Accuracy

High Drift Risk

Forecast Accuracy measures how closely a governed forecast value matches the governed actual result for the same period, scope, currency, and business outcome.

Governed formula

1 - absolute(forecast value - actual value) / actual value

  • State the forecast snapshot time, forecast category, actual result definition, and whether accuracy is absolute or signed.
  • Use the same period, segment, currency, and revenue or bookings basis for forecast and actual values.

Commit Forecast Accuracy

Compares commit forecast at a governed snapshot time with actual closed results.

Useful for sales execution reviews, but commit category rules must be stable.

Pipeline Forecast Accuracy

Compares a pipeline-derived forecast or weighted pipeline value with actual results.

Useful for model review, but it depends on approved weighting and eligibility rules.

Signed Forecast Variance

Reports over-forecast and under-forecast direction instead of only absolute accuracy.

Useful for bias detection, but it must not be mixed with absolute accuracy under one label.

Decisions to lock

Which forecast snapshot time and forecast category define the numerator?

A week-one commit forecast and final-week forecast answer different discipline questions.

Which actual result does the forecast reconcile against?

Bookings, ARR, revenue, closed-won amount, and invoiced value can all differ for the same period.

Is the metric absolute accuracy, signed variance, or weighted error?

Direction and magnitude tell different stories and should not share one field.

Validation questions

  • Can the forecast snapshot be reproduced with the exact timestamp, scope, owner, and category rules?
  • Does the actual result use the same period, segment, currency, and bookings or revenue basis as the forecast?
  • Are absolute accuracy, signed variance, and weighted error labeled as different metrics?

Common drift traps

  • Forecast values are updated after the snapshot point, making past accuracy look better than the governed forecast at the time.
  • The forecast is measured against bookings while the actual uses recognized revenue or invoiced amount.
  • Over-forecast and under-forecast errors are netted together, hiding directional bias in the forecast process.

Source-system boundary

Forecast reconciliation spine

CRM, Spreadsheets, ERP, Data warehouse

The governed definition should state snapshot timing, forecast category, actual-result basis, scope, currency, and variance method.

Context-layer proof

ClariLayer's context layer should bind Forecast Accuracy to snapshot timestamp, forecast category, actual-result basis, and variance method so forecast agents can explain whether misses came from timing, scope, or definition changes.

Governed signals
snapshot timestamp, forecast category, actual-result basis, variance method
Review cadence
Review after forecast-category, close-process, territory, target, or finance-reconciliation changes.

ClariLayer Drift Risk

Forecast Accuracy is high risk because it joins mutable forecast snapshots with actual-result definitions that can differ across sales and finance.

Ambiguity

5/5

The metric can mean commit accuracy, pipeline accuracy, signed variance, absolute error, or weighted forecast error.

Source-system dependency

5/5

The calculation depends on CRM forecast snapshots, spreadsheet overlays, finance actuals, currency, and warehouse reconciliation.

Time-window sensitivity

5/5

Snapshot timing, close timing, late updates, and fiscal-period boundaries can all change measured accuracy.

Governance need

5/5

Forecast accuracy drives executive trust and sales process reviews, so snapshot and actual-result rules need approval.

AI-agent risk

An AI agent can assign forecast blame incorrectly if it cannot see snapshot timing, forecast category, actual-result basis, and variance method.

Related metric definitions