The Forecast That Ate the Budget
A packaged-tea company selling across Assam and Tripura forecast 96,000 cases last year; actual sales were 68,000. Production followed the forecast, leaving 21,000 cases of near-expiry stock and ₹1.4 crore of working capital tied up. At the same time, one district ran out of the 500g SKU for five weeks during the festival season. Quotas were set as last year's sales plus 15%, uniformly across districts, even though two districts had lost distributors and one faced new competition. The forecast itself was the general manager's judgment plus a flat 10% growth factor; the sales-force composite was not used because 'reps sandbag'. Next year the marketing budget is cut 15%, a new 1-litre SKU is launching, and the monsoon is expected to be normal.
1. Diagnose why the forecast failed and recommend a forecasting approach for next year, naming the methods and the level of detail.
Read the worked answer
The forecast failed because it was a single judgmental number - general manager opinion plus a flat 10% - built at company level, not district or SKU level. It ignored structural changes: two districts had lost distributors, one faced new competition, and festival 500g demand was never modelled separately. A uniform 15% quota on last year's sales repeated the error and guaranteed either sandbagging or stockouts. Next year use a three-layer build: a statistical baseline from three years of district-SKU data with seasonality (monsoon and festival), a sales-force composite collected before quotas are announced, and a jury of executive opinion to adjust for known events such as the new 1-litre SKU and the 15% budget cut. Forecast at district-SKU level, add a separate new-product forecast, and roll it monthly. Track bias and mean absolute percentage error by district. Keep low, base and high scenarios so production can flex rather than commit to one number.
2. How would you deal with sandbagging and with the way quotas are set?
Read the worked answer
Separate the forecast from the quota. Reps sandbag because the forecast is used to set targets and incentives; if a low forecast earns an easy quota, honesty is punished. Collect the composite at assistant manager level or anonymously, and explicitly do not use an individual's forecast to set that individual's quota. Set quotas from territory potential and market share, adjusted for distributor loss and competition, not from a flat growth factor. Add a forecast-accuracy metric to the manager scorecard, rewarding over- and under-forecast errors symmetrically, and publish district accuracy so the conversation is about learning, not blame. For the new SKU, use a test-market or analogous-product method and set a separate launch quota with its own incentive. Review quotas mid-year only for documented structural events, so the rule stays credible.
3. Design a monthly control report with variance analysis and triggers, and say what you would do about the near-expiry stock and the budget cut.
Read the worked answer
By district and SKU, the monthly report should show forecast, actual, variance in units and value, variance percentage, and a reason code (distributor, competition, stockout, season, price). Add a rolling three-month forecast, inventory cover in weeks, and near-expiry stock. Set triggers: a district 15% below forecast for two consecutive months gets a field review; inventory cover above eight weeks triggers a trade offer or reallocation; a SKU stockout above three days triggers a production escalation. For the ₹1.4 crore near-expiry stock, ring-fence it, push it through high-velocity dealers with a targeted trade offer before expiry, and keep it out of the new SKU's launch numbers. With the 15% budget cut, protect the new SKU launch and the highest-margin districts; cut blanket trade schemes first. Hold a monthly sales and operations meeting with production and finance.