Override Value Prediction

Demand Planners and other stakeholders (Sales, Marketing, Finance, Executives) bring market, customer, and product insights not available to the AI engine.

We believe in providing all stakeholders the best possible starting point in the forecasting process: an AI Optimal Forecast which combines the previous cycle human forecasts (for continuity) and current cycle machine forecasts (for responsiveness).

We recommend a collaborative process where all stakeholders and review and refine the AI Optimal Forecast as needed. however, any human overrides should be supported by new assumptions / insights behind the override.

vyan.ai tracks all past overrides for value-added and real-time alerts stakeholders if a certain override is unlikely to add value given past track record. In general, large positive overrides are unlikely to add value (this is usually pressure to inflate forecasts to bridge gaps with Annual Operating Plan or Sales Targets). Small sized overrides (frivolous tinkering) is also unlikely to help, if overrides are within the random noise level of the time series. Usually, large negative overrides are most likely to help (as Humans are biased to not reduce forecast significantly unless there is a good reason to do so).

This real-time alerting helps reduce value-destroying overrides and improves the Final Consensus Forecast Accuracy.

vyan.ai then self-learns from successful overrides by providing higher weight to human forecasts for a given time series at given lags. This delivers a self-learning self-tuning AI Optimal Forecast signal, which reduces the need for Planners to override unless there are new insights not available to the AI engine.