Model Based Recommendations & Lead Generation

Prediction rules are very straight forward to implement when it comes to recommendations or generating leads.  Model analysis for is based on completely different concepts.Once Oracle Sales Cloud has enough historical opportunity revenue data that provide strong statistical correlations between customers and buying patterns for meaningful results, using the predictive models instead of the predictive rules become a more sophisticated option for analysis.
Specific customer attributes like industry, location, size of company, past products sold are used to determine future product recommendations. Data mining algorithms like k-means clustering and association model along with Oracle Business Rules engine are used to generate the recommendations.The data mining of historical data provides the insight and the business analyst will author rules based on data mining patterns that provide targeted recommendations to the customers of a sales rep.

But although you could see it as a self learning engine, we can influence how the model is learning:

• Eligibility rules that define if a customer should ever be considered for a product recommendation still apply here
• We can analyze the quality of the attributes that the model is taking into account.  The quality of these attributes reflects whether or not we have enough data to properly take them into consideration.  If the quality of a few attributes is insufficient, we could consider uploading more historical data to reinforce their quality to become more accurate predictions.
• We can have the model focus on certain attributes as the importance of them might vary from industry to industry.

Model based lead generation

Model based lead generation is no different from Rule based lead generation.  Once recommendations have been identified through rules or models, then can be converted into leads as explained in the rule base lead generation post.