A consultant is enabling predictive lead scoring for a software vendor that sells both a low-cost self-serve subscription and a high-touch enterprise platform through the same Dynamics 365 Sales lead queue. Self-serve leads convert in days on small deal sizes, while enterprise leads take months and involve many more touchpoints. The client wants one scoring setup that produces meaningfully accurate grades for both groups. What should the consultant recommend?
Select an answer to reveal the explanation.
Short Explanation
Picture two completely different sales motions running through the same funnel: one where someone clicks buy on a cheap plan within a week, and one where a deal creeps through months of meetings before it closes. If you train one scoring model on both groups mixed together, the model ends up learning a blurry average that does not really describe either kind of deal, so the grades it hands out feel off no matter which type of lead you look at. The fix is to split the training data along the line that separates the two businesses, so each model only ever sees leads that behave like each other. That gives you two scores that are each honestly calibrated to their own world. Trying to patch this by just switching scoring off for the slower-moving deals throws away a workable tool for no good reason, and telling sellers to eyeball-adjust a number in their head just reintroduces the guesswork that scoring was meant to replace.
Full Explanation
The correct answer is A. Predictive lead scoring builds its model from historical patterns in the population it is trained on, so mixing two groups with very different conversion timelines and deal sizes produces a blended model that fits neither group well. Scoping separate models to each segment, using a field that distinguishes self-serve from enterprise leads, lets each model learn from a coherent population and produce grades that reflect real conversion behavior for that segment. Option B is incorrect because the platform does not automatically adapt a single model to serve dissimilar populations; averaging distinct patterns together degrades accuracy rather than improving it. Option C is incorrect because predictive scoring is not restricted to short sales cycles; it can be configured for longer-cycle segments as long as there is enough consistent historical data, and disabling it for enterprise leads discards a workable option. Option D is incorrect because asking sellers to manually discount an automated score defeats the purpose of the feature and relies on inconsistent human judgment instead of a model actually suited to the segment; the platform supports scoping models rather than requiring workaround interpretation.