At the start of each quarter, you scramble to put together your sales forecast. Your sales reps make their commits, your sales ops team is neck-deep in Excel spreadsheets, and your CFO expected it before the end of last quarter.
As sales leaders, forecasting plays an important role in our lives, yet most of us don’t take the time to measure our accuracy. If we did, we would quickly realize there is room to improve. To do so, however, requires some work. This blog post provides a guide to improving your forecast accuracy. The trick is to combine multiple techniques into one ensemble forecast.
As its name implies, an ensemble forecast is a group of forecasts that are viewed as one with the goal of improving accuracy. It is like diversifying your stock portfolio, by purchasing multiple stocks you reduce unsystematic risk. Sales forecasting is no different. If you make multiple forecasts, you can put them together to improve your accuracy and reduce your risk.
This forecasting technique is based on the number of sales reps you have, their assigned quota, their tenure, and seasonality in your business. You calculate this forecast based on how much quota you’ve assigned and how much of you expect to be attained this quarter. This is based primarily on the experience of your team and your expected sales ramp rates.
This forecasting method is based on the commitment you get from each of your sales reps. Each sales manager rolls up these commitments and passes them up to be combined into an overall forecast. This is a common approach found in many sales organization.
This approach is not based on the number of opportunities in your funnel and their perceived value. This is not based on your sales reps’ committed opportunities. This approach might involve calculations leveraging average deal size, expected win rate, average days to close, etc. or it might just be a simple multiple, like a 3x goal.
This forecast is only based on data that you are able to observe through time. It considers business trends, cyclical business changes, seasonality, and random noise. This type of forecasting does not observe the open opportunities in your sale pipeline, but can still be a very good estimate of sales. This is especially true if your business does not rely on only one or two deals each quarter.
This approach is similar to roll-up forecasting, but instead of asking your sales team to make the commitments you use an algorithm to determine which opportunities will win and which will lose. The appeal to this type of forecasting is its ability to remove human bias. You no longer have to worry about sandbagging or over commitments.
It is time to construct your ensemble forecast based on each of your individual forecasts. You can do this by simply averaging them. This gives each forecast an equal weight and should get you closer to the predicted outcome. If you don’t think each deserves equal weight, you can determine how much confidence you have in each forecast and assign weights accordingly.
In order to improve the accuracy of your ensemble forecast, you will eventually need to tune the weights according to performance. You should compare each forecast to the actual outcome to determine the weights. You can use a simple regression or can leverage machine-learning models to determine the appropriate weights. Either way, your forecasts will improve based on measured accuracy.
Now that you know how to put together an ensemble forecast, you are ready to improve your sales forecasts. The key is to get a process in place and measure the result of each forecast. At ORM we specialize is in sales forecasting. If you have any questions, or would like help improving your sales forecast, let us know at email@example.com.
For more information on sales forecasting, our white paper on The Importance of Sales Forecasting gives a detailed method of obtaining an ensemble forecast to ensure you never miss another quarter.