Before rolling predictive lead scoring out to the full sales team, a consultant reviews the model performance page and finds the lift chart shows the top-scored decile of leads converting at roughly the same rate as a randomly selected group of leads. What should the consultant conclude?
Select an answer to reveal the explanation.
Short Explanation
A lift chart is basically asking one question: if you only worked the leads at the very top of the ranked list, would you close more deals than if you'd just grabbed a random handful? When the answer comes back 'not really,' that's a red flag, not a passing grade, because it means the ranking isn't actually separating the promising leads from the long shots yet. Reading a flat curve as a win gets the logic backwards. It's also not a chart that's only looking at the current month, so waiting a few weeks for the calendar to turn over won't magically fix a weak signal. And piling more fields onto the intake form isn't the answer either, since the problem lives in how well the data already being captured relates to actual outcomes, not in how much data there is. The right call here is to hold off on a full rollout and dig into why the signal is weak first.
Full Explanation
The correct answer is C. A useful predictive model should separate strong leads from weak ones, and a lift chart demonstrates that by showing whether the highest-scored group converts meaningfully better than a random sample; when the top decile converts at roughly the same rate as chance, the model is not yet finding a reliable signal, so pausing the full rollout until performance improves protects the sales team from chasing a prioritized list that is no better than working leads in any order. Option A is incorrect because a flat lift curve is the opposite of good performance; it indicates the model is not differentiating leads at all, not that every lead has been fairly and accurately evaluated. Option B is incorrect because the lift chart is not artificially restricted to the current calendar month, and waiting for month-end would not change a fundamentally flat result caused by weak training signal. Option D is incorrect because adding more picklist fields to the lead form does not address the diagnosis; the issue the lift chart reveals is about the strength of the relationship between existing data and outcomes, not the quantity of fields captured on the form. The consultant should investigate training data quality and volume before recommending a broader rollout.