A consultant is configuring Dynamics 365 Sales for a mid-market client whose marketing team wants to identify which inbound leads are worth converting to opportunities, while the sales team separately wants to know which active opportunities are most likely to close. The client asks whether one AI model can serve both needs. What should the consultant recommend?

The recommendation is to enable predictive lead scoring for lead records and predictive opportunity scoring for opportunity records because each model is trained on outcomes specific to its own record type. Predictive lead scoring predicts conversion likelihood from lead attributes and marketing engagement, while predictive opportunity scoring predicts win likelihood from opportunity-stage activity and deal characteristics. Running both gives the marketing team a lead-qualification signal and the sales team a close-likelihood signal, matching what each group asked for. The approach that enables only predictive opportunity scoring and configures it to run against lead records is incorrect because opportunity scoring is not designed to evaluate lead records; the two entities carry different fields and historical outcome data, so pointing the opportunity model at leads would not produce a valid score. The approach that enables only predictive lead scoring is incorrect because a lead's score does not transfer to the opportunity created from it; once a lead converts, the opportunity needs its own predictive opportunity score built from opportunity-stage data. The approach that builds a single custom model outside Dynamics 365 Sales to score both leads and opportunities uniformly is incorrect because it assumes native Dynamics 365 Sales AI scoring is limited to a single entity per environment, which is false; both models can run in the same environment against their respective entities without a custom external solution.