A consultant supports a client that just completed an acquisition, roughly doubling the volume of leads flowing into Dynamics 365 Sales and introducing a new industry the client previously had little exposure to. The consultant is concerned the existing predictive lead scoring model, trained before the acquisition, will misjudge these new leads. (Select TWO.) Which two actions should the consultant recommend?
Select all correct answers, then click Submit.
Short Explanation
An acquisition that suddenly changes who your leads are is the kind of shock a model trained on the old normal was never built to handle gracefully. The smart move is not to wait around for the calendar to catch up, it is to force a fresh look now so the model starts learning the new industry's patterns as soon as possible. And because no retrain is guaranteed to nail it on the first try, someone should keep an eye on how those new leads score and actually convert over the next several weeks, adjusting if the numbers still feel off. What would be overkill is assuming the whole feature has to be switched off because two industries are now mixed together, that is well within what the model can handle once it has retrained. And throwing away the older history to make room for the new industry only weakens the model further, since that older data is still useful evidence of what predicts a win.
Full Explanation
The correct answers are B and C. A sudden, large shift in the mix of incoming leads is exactly the kind of change that can make a model trained on the old mix produce misleading scores, so triggering a manual refresh rather than waiting for the next scheduled retrain gets the model learning from the new industry's patterns sooner. Because a retrain does not guarantee an immediately perfect fit, the consultant should also recommend watching how the new industry's leads score and convert over the following weeks to confirm the model is now tracking their real outcomes, adjusting further if it is not. Option D is incorrect because it is a drastic overreaction; Dynamics 365 Sales predictive scoring can absolutely learn from a lead population spanning multiple industries, it just needs enough retrained history to do so well. Option A is incorrect because discarding the pre-acquisition closed history throws away legitimate signal the model could still learn from, and shrinking the training data right when more history is needed works against the goal rather than helping it.