Discovery gap analyzer

After the call, list the questions that were not asked, compared with your published discovery standard. Each gap is work for the next meeting. It is not a verdict on the AE, and it is not a coverage percentage.

A summary records what the buyer said. A post-call summary and CRM auto-update writes those facts into fields. Gap analysis asks the other question: against the topics you require, what never received a question? Summaries miss absence on purpose. They are built to compress what happened.

If that list lands in a performance packet, the job has changed. Reps will stop recording, or they will spray questions that sound like discovery and never wait. Grade the file: what is still unknown on this opportunity. Coach the person in a separate conversation, after a manager has actually listened.

Skip the coverage score. A precise-looking percentage still punishes a short intro and rewards a long call that never landed a number. The file has an answer, a cite that the topic was never asked, or a named skip because this meeting was not supposed to cover it.

Keep four required topics, written down by enablement or the sales manager, and do not rewrite them every quarter:

• Pain: what is broken, who feels it, what they already tried.
• Metric: how they would know it got better, in a number they already use.
• Buyer: who else must agree, who can stop it, who will live with the change.
• Process: how they buy, what else is in flight, what would keep them on the current way.

Run the transcript against that list, not against a generic methodology poster. If the team runs MEDDIC, empty metric and economic-buyer fields after MEDDIC auto-extraction are a hint that a topic may be missing. Extraction writes what was said. Gap analysis flags what was never asked. A methodology scorecard then scores whether the AE ran your play. Keep those three jobs apart. One call-quality number hides which job failed.

For each required topic, write one of three states:

• Answered, with a quote from the buyer.
• Unasked, with a cite from silence: the moment a thread was available and no question followed.
• Out of scope for this meeting type, with the type named (intro, working discovery, demo, technical deep-dive, executive checkpoint).

Cap the list. Three unasked questions, ranked by whether the next meeting can actually ask them, is enough. Twenty flags is a report nobody opens.

Do not write gaps into the CRM as if they were facts. "Metric unknown" is a next-step, not a filled field. The AE still confirms any field update.

Real-time call coaching can surface a missed question while the buyer is still on the line. After the call you are not prompting. You are building the next agenda from what the file still lacks.

Every unasked question needs a cite that the topic was available and not probed. Useful cites look like this:

• Buyer: "Exception handling is eating the ops team." AE: starts a product walkthrough. Cite: pain named, no follow-up on impact, owner, or current workaround.
• Buyer names two teams. AE never asks who else has to sign off. Cite: people mentioned, buying group not mapped.
• The rest of the call never opens how they measure the problem today. Cite: metric topic never started.

A cite from silence is a timestamp plus the adjacent turn. It is not a guess about what the buyer probably feels. If they never mentioned cost, overtime, or attrition, do not emit "uncover the cost of attrition." That is invented pain. The model is writing a playbook, not reading this call.

The same rule catches a quieter miss: treating a passing complaint as finished discovery. "We looked at a few tools" is not process. "It has been messy" is not a metric. "We will circle back" is not a buying step. Leave those as unasked, with the actual line attached, so the AE can ask next time instead of typing fiction into the CRM.

Phrase the output as questions the AE can put on an agenda, not as diagnoses. "How many exceptions does the morning pile usually hold, and who works it?" is usable. "They are drowning in cost from exceptions" is a claim the buyer did not make.

Label the meeting type before you flag gaps. A handshake, a fifteen-minute teaser, or a first intro should not be scored as a full discovery. Two empty topics on a short intro often means the meeting was scoped correctly. The list should say "ask these on the working session," not "this AE failed discovery."

Picture a first meeting with an operations director at a mid-market shipper. The AE opened a pre-call account briefing that noted a recent warehouse system rollout. Twelve minutes in, the director says the rollout "created a pile of exceptions the team works by hand every morning." The AE empathizes, shares a screen, and spends the rest of the slot on a dashboard.

The gap list should not say the call went poorly. It should say, with cites:

• Pain follow-up unasked. Cite: "pile of exceptions... every morning." No question on volume, who works the pile, or what they already tried.
• Metric unasked. Cite: the remainder is a product walkthrough. No question on a number they already review, such as exceptions per morning, overtime, or on-time percentage.
• Buyer unasked. Cite: the director speaks as "the team" and "we." No question on who else feels the pile, or who would have to approve a change.
• Process: defer. Meeting type is first intro, about twenty-five minutes, no procurement language. Missing RFP timing is not a failure on this file.

The AE puts those three questions at the top of the next invite, in the agenda the buyer will see. The manager does not paste the list into a quarterly review. If the next recorded call still never asks them, then you have a coaching problem. The first list was only the file.

The AE on the opportunity owns the output. They accept, rewrite, or drop each unasked question. Kept questions go on the next calendar invite. A gap that never reaches the agenda the buyer sees is decoration.

Send the list to the AE first. A manager can read it in deal review after the AE has already used it, the same way they read CRM fields. If the first viewer is always the manager, you are grading the AE. Recording will thin out, and the transcripts you still get will be the ones people were willing to be judged on.

Gong, Chorus, Salesforce, and Hyperbound sit in the same class when they read a transcript and surface missed topics. Some of that lives in conversation intelligence. Some of it sits next to a CRM activity. Some of it lives in a coaching product. Do not rank them. Do not assume a native CRM note is a gap list, or that a specialist coaching product is automatically better at absence. Inspect three things: whether the required topics are yours; whether each gap cites a moment of silence rather than a generated pain; whether the AE can turn the list into the next agenda without a dashboard tour.

After a stretch of calls, check whether flagged questions actually got asked on the next recording. That is the operational test. If they did not, the list is too long, the meeting types are wrong, or the AE does not trust the output. Fix those. Do not add a coverage percentage to the scorecard.