The regional forecasts do not add up to the company plan

Finance has a $2 million revenue forecast. The regional teams forecast $1.8 million between them. Both versions appear in next quarter's planning papers. Before choosing which number to circulate, establish whether they cover the same business. If they do, the disagreement needs a forecasting decision, not an unexplained $200,000 adjustment.

Forecast reconciliation produces a set of estimates whose parts and totals agree. It does not prove that the forecasts are accurate or that a sales target is achievable. The distinction matters when different teams use the numbers to commit staff, stock or spending.

Define the measure, currency, forecast date and target period. Confirm that every team is forecasting the same kind of revenue, rather than mixing bookings, billings and recognized revenue. Use the company's agreed accounting basis and distinguish the management target from the current expectation.

Next, identify the smallest reporting groups needed for the decisions. Each sale should belong to the appropriate group once. A regional view and a product view can describe the same sales; adding their totals together would count the business twice.

For an invented example, the starting forecast is:

The four region-product cells form one consistent $1.8 million population. They support both regional and product reports without treating those reports as separate revenue sources. A simple geographic hierarchy nests locations within regions; crossing regions with product categories creates a grouped structure. Forecasting: Principles and Practice explains this distinction and the requirement for coherent totals.

Check omitted or overlapping groups before debating prediction methods. A corporate forecast that includes a planned launch while local forecasts exclude it has a scope difference. Name the missing group and its assumptions. Do not spread its revenue across existing products merely to remove the reconciliation gap.

Save the original regional, product and corporate estimates with their dates and assumptions. These are the starting forecasts, not mistakes to erase. Each may contain useful information that another view misses.

Ask the corporate forecaster to explain the additional $200,000 in the example. Does it come from different retention, pricing, delivery or demand assumptions? Ask regional owners whether those assumptions are already included locally. The same planned contract must not be counted once in a regional model and again as a corporate adjustment.

Classify the difference before resolving it:

This classification keeps a data repair from being mistaken for a more optimistic commercial judgment.

A bottom-up approach accepts the forecasts for the smallest groups and adds them. In the example, the corporate forecast would become $1.8 million. A top-down approach starts with the corporate estimate and distributes it using defined proportions. Both can produce coherent numbers, but they rely on different information. Detailed groups can be noisy; fixed historical shares can miss a change in the business mix. Those tradeoffs are described in the forecasting text's single-level approaches.

If the company spreads the entire $2 million proportionally over the existing $1.8 million detail, every cell rises by about 11.1%. North services becomes roughly $444,444 instead of $400,000. That is a documented allocation calculation, not evidence that the team can deliver the extra work.

Do not choose the method merely because it preserves the preferred headline. Record which level has credible evidence for the decision and test the result against delivery and customer assumptions. If the information cannot support a firm choice, show the uncertainty rather than disguising it as an agreed forecast.

Suppose a review of the hypothetical case supports another $50,000 in North subscriptions, $50,000 in North services and $20,000 in South services. Assume these are distinct changes, use the same revenue basis and are not already in the starting forecast. The agreed detail becomes:

The example assumes evidence supports those changes; the amounts are not actual company results. The corporate forecast now falls from $2 million to $1.92 million, while the detailed forecast rises by $120,000. There is no need to invent another $80,000 just to preserve the original corporate number.

Keep an adjustment log naming each assumption, its owner and the evidence that changed it. If management's target remains $2.2 million, show the $280,000 target gap separately. Closing that gap requires an operating plan, not another reconciliation entry.

A business with many related series may need more than a meeting and a small table. Statistical methods can adjust forecasts across levels using information about their errors. Wickramasuriya, Athanasopoulos and Hyndman's original reconciliation research develops a method using the covariance of forecast errors under stated assumptions. Its mathematical objective is not a guarantee of better results for every company dataset.

Ask an analyst to compare an appropriate method with simpler alternatives at the horizons the business uses. Require disclosure of sparse histories, unstable categories and constraints such as nonnegative revenue. A forecast that adds up but produces implausible local values still needs investigation.

Publish the approved detailed estimates and generate regional, product and company totals from them or from the accepted reconciliation system. Check every relevant sum after adding a product, changing a territory or importing a new forecast version. Do not allow a presentation copy to become a second independently edited forecast.

Retain the starting estimates, adjustments and released result. When actuals arrive, compare both the original and reconciled forecasts with outcomes at the levels used for decisions. That record can show whether reconciliation improved the useful predictions, merely repaired inconsistent totals or concealed a recurring weakness in one team's assumptions.