A consultant is preparing a client's Dynamics 365 Sales environment to get the most reliable results from AI opportunity scoring and Copilot-generated summaries before go-live. (Select TWO.) Which two actions should the consultant prioritize?
Select all correct answers, then click Submit.
Short Explanation
If you want these AI features to actually earn their keep on day one, there are two things worth focusing on before go-live. First, the model and the summary generator can only work with what has been captured, so getting sellers into the habit of logging their emails, meetings, and notes against the opportunity record matters more than any setting in the system. Second, the scoring model learns from history, so if the client has deal history sitting in spreadsheets or an old CRM, bringing that into the system gives the model real examples of what winning and losing look like instead of starting from nothing. Turning off a field sellers use for their own notes does not make the AI smarter, and trimming the sales process down to fewer stages is not a lever that improves prediction quality either. Data volume and data capture are the two things that actually move the needle.
Full Explanation
The correct answers are C and D. Both predictive scoring and Copilot summaries depend on activity data being tracked to the record, so confirming sellers log emails, meetings, and notes against opportunities ensures the AI features have material to analyze, as described in option C. Predictive scoring also depends on learning from a meaningful volume of historical won and lost deals, so migrating past deal history where it is missing, as described in option D, gives the model enough examples to identify patterns before go-live. Option A is incorrect because the probability field is a standard seller-editable field independent of AI scoring, and disabling manual entry does not improve model accuracy; it only removes a data point sellers use for their own tracking. Option B is incorrect because there is no requirement or accuracy benefit tied to minimizing the number of sales stages; stage count is a process design decision, and collapsing stages to satisfy a scoring preference is not a documented or necessary step, and could actually reduce the granularity of data the model can learn from.