Forecast Accuracy Analysis

𝐌𝐨𝐬𝐭 𝐟𝐢𝐧𝐚𝐧𝐜𝐞 𝐭𝐞𝐚𝐦𝐬 𝐮𝐩𝐝𝐚𝐭𝐞 𝐭𝐡𝐞𝐢𝐫 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭𝐬 𝐞𝐯𝐞𝐫𝐲 𝐦𝐨𝐧𝐭𝐡 — 𝐛𝐮𝐭 𝐫𝐚𝐫𝐞𝐥𝐲 𝐥𝐨𝐨𝐤 𝐛𝐚𝐜𝐤 𝐭𝐨 𝐬𝐞𝐞 𝐡𝐨𝐰 𝐚𝐜𝐜𝐮𝐫𝐚𝐭𝐞 𝐭𝐡𝐞 𝐨𝐥𝐝 𝐨𝐧𝐞𝐬 𝐰𝐞𝐫𝐞.

That is normal. A forecast made 10 or 11 months before close is not expected to be perfect. But it still matters because early forecasts influence budgets, hiring, cash planning, and investment decisions.

𝐓𝐡𝐞 𝐫𝐞𝐚𝐥 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧 𝐢𝐬 𝐧𝐨𝐭:

“Was the first forecast right?”

𝐓𝐡𝐞 𝐛𝐞𝐭𝐭𝐞𝐫 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧 𝐢𝐬:

“How much confidence should we place in a forecast at each stage of the planning cycle?”

That is why I built this Power BI page that tracks every forecast version for each month and compares it against the actual result.

Each line shows one month’s forecast journey:

from the early estimate,

to the monthly updates,

to the final number.

The pattern is clear.

𝐄𝐚𝐫𝐥𝐲 𝐟𝐨𝐫𝐞𝐜𝐚𝐬𝐭𝐬 𝐚𝐫𝐞 𝐰𝐢𝐝𝐞𝐫 𝐚𝐧𝐝 𝐦𝐨𝐫𝐞 𝐮𝐧𝐜𝐞𝐫𝐭𝐚𝐢𝐧.

As the actual month gets closer, the forecast tightens and moves toward the truth.

In this dataset:

• Final estimate accuracy: ~0.6% MAPE

• 11-month forecast accuracy: ~1.44% MAPE

• Long-range forecasts were about 2.3× less accurate

• 36 of 66 forecasts came in below actual, showing a mild conservative bias

𝐁𝐮𝐭 𝐭𝐡𝐢𝐬 𝐫𝐚𝐢𝐬𝐞𝐬 𝐚 𝐛𝐢𝐠𝐠𝐞𝐫 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧:

Are we only asking, “How much did we miss by?”

Or are we also asking, “Do we keep missing in the same direction?”

In this case, most forecasts came in below actual. That suggests a mild conservative bias.

And that matters because consistently low forecasts can cause the business to plan too small, hire too late, or invest too cautiously before the growth becomes obvious.

Built in Power BI using Deneb / Vega-Lite

eyJrIjoiODM3MzQ2ZGUtMzkyZC00OWFhLWJjMTYtY2ZmYWRkZjM3ODJmIiwidCI6IjExYjJhMTMyLTI2YzYtNDJjNy05N2IxLWVlNGY2YTU2NjNlYiIsImMiOjEwfQ%3D%3D