Retraining the Predictive Opportunity Scoring Model

AB-210 By islever Editorial Last updated

Configure AI-powered intelligence and insights features Optimize opportunity management

Short answer

DE Retrain the predictive opportunity scoring model on a smaller, recent data set and set it to ignore empty values so current conversion patterns are learned.

Question

You manage Dynamics 365 Sales for a company.
The predictive opportunity scoring model was trained ten days ago.
Sales managers report that the scores no longer reflect current conversion patterns.
You need to manually retrain the model to improve predictive scoring accuracy.
Which two actions could you configure? Each correct answer presents a complete solution.
NOTE: Each correct selection is worth one point.
  1. Increase the number of users in the environment.
  2. Create additional models.
  3. Enable Dataverse auditing.
  4. Use a smaller data set. Correct Answer
  5. Ignore empty values. Correct Answer

Key Takeaways

It tests whether you know model accuracy is controlled by training data scope and missing-value handling, not by environment size, auditing, or extra models.

This AB-210 question covers retraining the Dynamics 365 Sales predictive opportunity scoring model when scores drift from current conversion patterns. The correct actions are using a smaller data set (D) and ignoring empty values (E).

Common mistake: Choosing B (create additional models) — extra models create a competing prediction instead of retraining the existing scoring model that managers are already using.

Expert Analysis

What the question is really testing The scenario is about a trained predictive opportunity scoring model in Dynamics 365 Sales whose results have drifted from current conversion behaviour. The exam is checking whether you understand what actually influences a model when you retrain it: the data you feed it and how you handle that data. Retraining is a data-shaping exercise, not an environment or infrastructure change.

Why D and E are the answers Option D is correct because narrowing the training set lets the model learn from recent, relevant records rather than years of stale conversion history — the ten-day-old model is stale precisely because older periods dominate the pattern. Option E is correct because sparse or missing attribute values degrade training quality; instructing the model to ignore empty values removes noise so the signal on current conversion patterns survives. Together these are the two levers you can configure on the existing model.

The adjacent decision The distractor that tempts most candidates is B: creating additional models. Additional models would give you separate scoring for a different segment or a fresh baseline, but they do not retrain the model already reporting scores — you would end up managing two competing predictions. The comments make exactly this point: creating more models does not retrain the existing opportunity scoring model. A is irrelevant because licence/user count has no bearing on model accuracy, and C (Dataverse auditing) merely records changes; it supplies audit history, not training signal.

What would flip the answer If the question asked how to capture more historical data going forward, or how to isolate a brand-new scoring prediction, the reasoning would shift toward auditing or creating a new model. Because the wording is "manually retrain the model to improve predictive scoring accuracy," only data-set scope and missing-value handling qualify.

Operational consequence Leaving the model stale means managers keep prioritising pipeline from outdated patterns, reps chase deals the score overrates, and trust in Sales Insights collapses — which is why scheduled retraining with a current, clean data set is the operational fix.

Official Reference

https://learn.microsoft.com/en-us/dynamics365/sales/

This link opens the official vendor documentation in a new tab.

Exam Strategy

When a question says a model must be retrained, eliminate any option that changes the environment rather than the data — user counts, auditing, and infrastructure never affect training accuracy. Then ask whether the option acts on the existing model or creates something new; 'additional models' is a trap because it produces a second prediction rather than a refreshed one. Read the constraint wording closely, since a differently framed question about capturing more history would change which options qualify.

Frequently Asked Questions

Why does a predictive opportunity scoring model become inaccurate after ten days?

The model learned from historical conversion patterns that no longer match current selling behaviour. Retraining on a recent, well-populated data set refreshes the patterns the score is based on.

Does creating additional models retrain my existing scoring model?

No. A new model produces a separate prediction you would have to manage alongside the original; the model already serving scores stays unchanged until you retrain it.

How does ignoring empty values improve scoring accuracy?

Sparse or missing attributes add noise during training, weakening the relationship between fields and conversion outcomes. Excluding empty values lets the model train on complete, meaningful records.

More AB-210 FAQ →

Community Votes

D
50%
E
50%

50% of anonymous learners picked answer D. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 1 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 1

  1. DGee 👍 1
    Creating more models does not retrain the existing opportunity scoring model. You can modify the training data used by the model. Using a more recent and relevant subset of data can help the model better reflect current sales conversion patterns. Missing or sparse data can negatively affect model quality. Configuring the model to ignore empty values can improve the effectiveness of the training process and resulting scores.

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Practice All AB-210 Questions

Access 100 questions with complete answers and detailed explanations.

View Full AB-210 Practice Test →

← Back to AB-210 Study Guide