A consultant is preparing to enable predictive lead scoring for a client whose Dynamics 365 Sales environment holds clean historical data spanning eighteen months, with qualified or disqualified outcomes recorded on roughly 60,000 leads. Which two actions should the consultant take before activating the model? (Select TWO.)
Select all correct answers, then click Submit.
Short Explanation
Before flipping this feature on, the consultant needs to check two things that both come down to trusting the fuel going into the engine. First, were the past leads actually labeled honestly and consistently as won or lost, because a model taught on messy or missing labels will confidently learn the wrong lessons. Second, does the organization genuinely have enough of that clean history to train something reliable, since a rough count of records is not the same as confirming the data actually clears the bar. What does not belong on this checklist is hand-typing scores onto leads that have not even closed yet, since the model does not learn from guesses, and touching owner-assignment rules, which is an entirely separate mechanism from figuring out how likely a lead is to convert.
Full Explanation
The correct answers are C and D. Predictive lead scoring learns by comparing patterns in past leads to their recorded outcomes, so the model is only as good as the data feeding it; the consultant must confirm the qualified or disqualified field was populated consistently, since gaps or inconsistent labeling teach the model the wrong lessons. The consultant must also confirm the organization's data volume and history actually clear the threshold for a reliable org-specific model rather than assuming eighteen months and 60,000 records automatically qualifies, since data quality and completeness matter as much as raw count. Option A is incorrect because manually scoring open leads has no bearing on model training, which learns from closed historical outcomes, not from scores seeded onto records that have not yet reached a conclusion. Option B is incorrect because lead scoring evaluates conversion likelihood and has nothing to do with owner assignment; disabling assignment rules would disrupt lead routing for no benefit to the scoring model and conflates two unrelated capabilities within the platform.