Search
  • Bill Kantor

Entering the fourth dimension

Updated: Jul 16

In our recent blog posts, we discussed how data quality and breadth impact the accuracy of sales forecasts. While it may seem counterintuitive, having highly accurate and complete sales data isn’t as important as it might seem, nor does sales forecasting require dozens of internal and externally sourced data fields.

More data depth trumps data accuracy—especially when making predictions in aggregate.

In fact, we noted that Funnelcast only requires six fields that are already being captured by your CRM.


Data quality, breadth, and depth. How much do you need?

Photo by Volodymyr Hryshchenko on Unsplash


That brings us to the third data dimension, namely depth, or history. Turns out that more data depth trumps data accuracy—especially when making predictions in aggregate, or across your entire sales pipeline. (Admittedly, better data quality will improve the fidelity of individual opportunity forecasts.)


How much data is required to get started? Three to four sales-cycles, are typically sufficient, with two to three cycles used for training the forecasting model and one or two cycles for model validation. You will also need enough successful closes to train the model. This translates into 50+ closed opportunities per business segment that you want to analyze. That’s remarkably little data compared to typical forecasting problems and speaks to the power of Funnelcast’s unique modeling methodology.


In fact, statistical forecasting can produce extremely useful and accurate predictions—so long as the underlying data properties don’t change. The technical term for this is “stationarity”. It doesn’t take a data scientist to figure out when your sales data has changed. Here are some common events we’ve encountered in working with clients that can disrupt stationarity and potentially degrade the accuracy of your forecasts:

  • You have introduced a new product for which you have no history.

  • Your product or service is launched in a new business segment (e.g., a geography or industry) for which you have no history.

  • An economic shock (like Covid-19) causes your B2B customers to slow or accelerate buying decisions.

  • Competitors offer a similar product or service for free to gain market share

  • You experience particularly favorable or unfavorable press or analyst coverage.

Being aware of these events is key to knowing when to trust your data history and the accuracy of your forecasts. Still in many of these cases, forecasts based on what you know about the past can be used to provide a sense of how things would be if the changes had not transpired or if new products/business segments perform like others.


If you've gotten this far in this blog post, we have to congratulate and thank you. If you are counting, you may be wondering what the fourth dimension is? Would you have clicked if we said "Entering the third dimension?"


We’ll cover data issues related to forecasting individual opportunities in a future blog post.

1 view0 comments

Recent Posts

See All