40 Questions Sales Managers Should Ask Before Trusting AI Forecasts and Coaching
Your AI tools are getting smarter at pattern matching, but they cannot read the customer's actual intent or recognise when your best rep is about to lose a deal they should win. The questions you ask before acting on an AI recommendation determine whether your team stays sharp or becomes dependent on suggestions that work on average but fail on the deals that matter.
These are suggestions. Use the ones that fit your situation.
1When Salesforce Einstein flags a deal as 85 percent likely to close, does that probability include the fact that your champion just got promoted and you have not met the new decision maker?
2Your forecast has become more accurate over the past two years. Have you asked your reps whether they feel more or less confident in their own read of deal momentum?
3Einstein recommends moving a deal to a later stage. Can you name three specific actions the customer took that justify that move, or is the recommendation based on deals with similar field values?
4When you override an AI forecast, do you document why? If not, how will you know whether your judgment or the AI was correct?
5A rep has closed 12 deals this quarter. How many of those deals did Einstein correctly predict, and how many did your rep's instinct about buying signals get right that the AI missed?
6Your pipeline looks healthy according to Einstein. Have you personally called three of your largest deals to hear directly from your reps what the customer is actually doing?
7Einstein surfaces a deal slipping to next quarter. Before you pressure the rep, have you checked whether the customer delayed because of their budget cycle or because they lost interest?
8You have started using Einstein's stage recommendations. Can your reps still explain why a deal should move forward without consulting the AI first?
9When you compare your forecast to Einstein's forecast, where do they disagree most often? Are those the deals where you know something the AI does not?
10A deal is flagged as high risk by Einstein but your rep insists the customer is still engaged. Do you have a way to test which assessment is more accurate before the month ends?
Sales Coaching and Rep Development
11Gong AI highlights a rep who is not asking discovery questions in calls. Before you coach them, have you listened to the full call to hear whether they asked the questions in email instead?
12Your AI coaching tool flags that Rep A uses fewer talk-to-listen seconds than Rep B. Do you know whether Rep A's accounts are already familiar with your solution and need less education?
13Gong suggests Rep C should use the same objection-handling language as your top performer. Does Rep C's territory and customer profile actually match the top performer's, or is the AI recommending a template that will not work?
14Your coaching programme now surfaces AI-generated insights for every call. How much time are your reps spending reading AI feedback instead of thinking about what they should do differently?
15A rep refuses to use ChatGPT to draft outreach emails, saying it sounds generic. Have you tested whether her hand-written emails actually convert better, or is she just resisting change?
16Gong shows that your best rep rarely follows the recommended discovery framework. Are you confident enough in that rep to let them coach others the way they actually sell?
17Your AI coaching tool recommends that Rep D spend more time on account research before calls. What would that rep stop doing if they add that task, and is that trade-off worth it?
18You have started using Gong to identify which reps need coaching. Are you also noticing when a rep stops trusting their own judgement and starts copying what the AI recommends?
19An AI tool suggests your reps should follow a specific talk track for a common objection. Before you roll it out, have you asked your reps whether that objection actually means different things depending on the customer's situation?
20Gong flags a rep for talking too much about product features. Before you coach them, have you checked whether their customers are actually technical buyers who want those details?
Outreach Automation and Relationship Quality
21Outreach AI is sending personalised emails to 500 prospects per week. Can you name three recipients who converted specifically because of the personalisation, or is the volume masking a low conversion rate?
22Your AI outreach tool recommends the best time to email each prospect. When you A-B tested that against rep intuition, what actually won?
23ChatGPT is writing your outreach templates. Have you sent one of those emails to a customer who has worked with your company before and asked them whether it felt like it came from someone who knew them?
24Outreach AI is automating follow-up sequences. How do you know when to have a rep break sequence and pick up the phone instead of letting the automation run?
25Your reps now use AI-generated subject lines. Are your open rates actually higher, or have you only measured volume and not quality of conversation?
26Outreach is personalising email based on LinkedIn data and website behaviour. What does it not know about this prospect that your rep could discover in a 10-minute call?
27An AI tool suggests a rep should not email a certain prospect because predictive intent data says they are not ready. What if the AI is wrong and you miss the exact moment they start buying?
28Your outreach cadence is now AI-optimised. Do your reps still have permission to break the cadence if they learn something that changes their approach?
29Outreach AI is scaling your rep's outreach. Is that rep still spending time building relationships with their existing customers, or has automation crowded out the work that closes complex deals?
30ChatGPT writes your outreach emails. If an AI-written email actually generates a meeting, do you know what the prospect responded to, or do you just count it as a win and move on?
Maintaining Your Own Forecast Judgement
31When was the last time you forecast your quarter without looking at what Einstein predicted, just from your own knowledge of the deals?
32Your AI tools have improved forecast accuracy by 8 percent. Have you measured whether your team's sales judgement has gotten better or just whether the AI's predictions have gotten better?
33Salesforce Einstein recommends a rep should focus on X instead of Y. Before you pass that advice on, have you checked what that rep actually knows about their accounts that the AI does not?
34You have not personally called a deal to check on momentum in three months because Einstein's data looks solid. What early warning signals are you missing?
35A rep's forecast disagrees with their AI tool's forecast. Do you have a system for finding out which one was right and learning from it?
36Your team is now optimised for the behaviours that AI coaching tools recommend. Are those the same behaviours that close your most complex, highest-value deals?
37When an AI tool flags something as a risk, do you automatically escalate it, or do you first ask the rep whether they see it as a risk?
38You are managing your forecast mostly through AI dashboards now. What do you see in a personal conversation with a rep that does not show up in the data?
39Your AI tools have trained your team to input data in a certain way. Have you checked whether that data entry discipline is actually helping them sell, or just helping the AI?
40If all your AI tools went down for a week, could you still forecast your quarter and coach your reps with confidence?
How to use these questions
After using an AI tool to forecast, write down your own forecast for the three largest deals. Compare the two. Track which forecast was correct. This is how you keep your own judgement sharp.
When Gong or your coaching tool recommends a behaviour change, ask the rep to test it on one deal first. Do not roll out AI recommendations across the entire team until you know they work for your customer base.
Every month, pick one deal that the AI said was safe and listen to the full call yourself. Every month, pick one deal the AI flagged as risky and check whether the rep's read was better.
Before you scale outreach volume using AI, measure whether your conversion rate is staying flat. If it is declining, more volume is just wasting both your time and the customer's.
Protect time for reps to do things the AI cannot measure: building trust with a sceptical decision maker, noticing when a customer's strategy has changed, adapting to a customer's actual buying process instead of a predicted one.