How to Achieve 80%+ Accuracy on Sales Forecasts

January 2nd, 2018


Content Team


Sales Forecasting



Estimated Time

16 Min

EVP Sales Perspective

In 2012 Cisco acquired a series of companies, that when integrated grew to become Cisco’s fastest growing business units. Revenues were looking great and the pipeline was strong. However, high demand created a pull effect on the product organizations and to ensure an efficient and effective supply chain, accurate sales forecasting was required.

Finance Perspective

The forecasts were prepared by the controllers and updated monthly. was the source of the pipeline data and an Oracle Datawarehouse was used to retain history from other areas. Finance built the business forecasts in MS Excel. The forecasts always began widely off and as it came closer to the end of quarter the accuracy improved substantially. But the business needed accurate forecasts from the beginning of the quarter, not the end.

Hypothesis for Analysis

Financial controllers’ forecasts seem to be fairly accurate in representing closed deals at the end of the quarter – yet early in the quarter the forecasts show large variances. The analysis was to focus on why the poor start of quarter forecasts.

Data Preparation

This was a situation of many moving targets: the sales territory structure had a number of shifts, causing regional roll-ups to change over the quarter. The pipeline data from Oracle and the pipeline data from Salesforce did not contain the same number of records as indexed by DealID. Normalization of the data was a troublesome task requiring a programmer that had business context.


Was carried out over two 4-week cycles, each cycle building and refining the algorithms and outputting data to business dashboards where they could be accessed by business users to validate the findings. The data discovery automation features providing the most useful clues for the analyst.


The sales reps were remarkably accurate when they entered the deal value, getting a 90+ % accuracy almost every quarter. But looking more closely, deals were being entered late into the quarter and closed fast. Sales reps were entering deals right before or soon after the deal actually closed. Also the best performing sales reps and partners were not offering the largest discounts. When looking at reps who sold the least, contrary to initial assumptions these reps were not poor performers but they were global account managers who were cross-selling this division’s products into their existing accounts.

Changed Behavior

Reps that entered deals into the system as early as possible allowed the organization to increase predictability. Reps not in the division were provided more product sales training.

Free eBook!
How to identify the real problems in your account based sales process.
Download eBook