Tässäpä jatkoa Pekan aiempaan artikkeliin.
From Abstraction to Practice
Public discussion often approaches machine learning from a somewhat abstract and futuristic standpoint. In contrast, potential ML use cases are actually very practical. One of such use cases is forecasting of sales opportunity hit rates which can benefit salespeople and management in B2B organizations. This blog post introduces two potential use cases for data-driven opportunity hit rate forecasting. Because listing ML use cases is not enough, I have posted another blog that describes the building process of a “hit rate engine”.
How Much Money do I Currently Have in the Sales Pipeline?
Firstly, hit rate forecasting can be useful for sales management and coordination purposes. Once your hit rate forecasting model has been trained and validated, it can be employed to generate a hit rate forecast for each individual opportunity in the sales pipeline. Aggregation of these results yields a financial estimate for the entire sales funnel.
1. Machine Learning Allows Intelligent Sales Funnel Forecasting
Total expected revenue can be estimated by multiplying each hit rate forecast by the corresponding opportunity price and summing up these results. The method can be further extended to consider expected aggregate profit if margin for each opportunity is known or derivable from the dataset. These ML-based forecasts can be integrated to BI reporting.
Implementation of sales funnel forecasting requires that your data source systems have sufficiently detailed feature information regarding individual opportunities. Forecasting accuracy improves as the level of attribute completeness increases. The most obvious data sources are CRM and CPQ (configure, price & quote), but potentially relevant data from miscellaneous sources may be readily available in a data warehouse. Keep in mind that the results will be only as good as your features and model.
What Price Should I Offer in this Situation?
Having a model that predicts sales opportunity hit rate is also beneficial in the sense that you can tell how hit rate evolves as price is increased or decreased. When the curve of forecasted hit rate is flat at the opportunity price, the salesperson should increase the price to make more profit. On the contrary, the opportunity price may be too high so that it would be optimal to ask less in order to boost hit rate and achieve maximum expected profit. It is critical to know where the steep curve exists with respect to price!
The forecasting model can be refined to indicate the theoretical optimum price point. Such “what-if analysis” requires that price and margin are inputs of your hit rate prediction model. The forecasting model produces a predicted hit rate for each corresponding price level within a given range. It is also relatively straightforward to carry out profitability modelling as hit rate has been predicted for a certain interval of prices and margins.
2. Hit Rate Engine as a Sales Assistant
Estimates for sales opportunity hit rate and expected profit can assist sales personnel in pricing. A nifty graphical presentation conveys how margin and theoretical profitability behave when price is increased. The engine can be extended with other attributes to can tinker with as well. In practice, the most natural place for this type of a tool to exist is embedded in your CRM or CPQ system.