How much of a deal is won during the discovery call? I have my own theory: a lot more than most people think.
Recently, I reviewed 50 recorded discovery calls from our sales team to understand what separated the deals that kept moving from the ones that slowed down or disappeared altogether (because sometimes it’s the case too, whether we like it or not).
In this article, I’ll share the 10 AI use cases that emerged from that analysis, along with examples from our own sales process and commentary from our account executive, Jackie Kyrylenko.
What Is a Discovery Call?
A discovery call is the first structured conversation between a sales professional and a potential customer. The representative’s job is to determine whether that interest aligns with a problem the product can solve. It’s important to note that both sides are qualifying each other at the same time.
I expect four things to come out of this conversation:
- The problem: What is the prospect trying to fix, and how urgent is it?
- The people: Who makes the decision, who influences it, and who might block it?
- The budget: Is money allocated, or is this still exploratory?
- The timeline: Are they buying this quarter, or “maybe next year”?
If any of these four stay unclear after the call, the deal moves forward on assumptions, and assumptions are, as I call them, “deal killers”.
To make sure all four get covered, sales teams typically use qualification frameworks:
- BANT (Budget, Authority, Need, Timeline) is the simplest. It checks whether the money, the decision-maker, the problem, and the timing are all present.
- MEDDIC (Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, Champion) maps out the full buying committee and the internal evaluation process.
- SPIN (Situation, Problem, Implication, Need-Payoff) is a questioning sequence. It moves the prospect from describing their current setup to quantifying the cost of not changing.
All three answer the same core question: should both sides keep talking?
Why Does the Discovery Call Set Your Win Rate?
The discovery call sets your win rate because it determines whether the rest of the sales process is built on accurate information or on guesses. If the rep confirms the use case, identifies the decision-maker, and understands the budget on the first call, the demo, the proposal, and the negotiation all move faster because every step is built on facts, not assumptions.
The deals that moved further usually had three things in common:
- The representative confirmed the prospect’s use case during the call (the one that surfaced after five minutes of questioning, which is usually different from what the prospect wrote on the form).
- The representative addressed security and compliance concerns early.
- The call ended with a specific, scheduled follow-up rather than a vague “let’s circle back.”
Naturally, less successful deals were missing at least one of those three: the specialist assumed the use case instead of confirming it, compliance questions were left for later, or the call ended without a concrete next action.
I wanted to see whether this held up in our own pipeline, so I analyzed 50 of our team’s recent discovery calls. Here is what I observed across those calls:
|
Discovery Call Metric
|
Closed Deals
|
Stalled Deals
|
|---|---|---|
|
Open-ended questions asked per call
|
5-15
|
2-4
|
|
Use case confirmed on first call
|
85%+
|
Under 30%
|
|
Security/compliance addressed on first call
|
80%+
|
Postponed to call 2 or 3
|
|
Follow-up sent after the call
|
Within 1 hour
|
Next day or later
|
|
Specific next step scheduled before hanging up
|
90%+
|
Under 40%
|
|
Average deal cycle length
|
3-4 weeks
|
8+ weeks or stalled indefinitely
|
The conclusion is simple: the more complete your discovery, the faster the deal moves. Every question you skip on the first call becomes a blocker later. This is why I decided to document how my team uses AI across the full discovery process, because the tool’s biggest impact is making sure nothing gets missed, before, during, or after the call.
10 Ways to Use AI in Your Discovery Call Process
In short, the 10 AI use cases I found in my team’s discovery calls are: pre-call briefing from past meetings, personalized competitor comparisons, battle cards built from buyer questions, post-call Q&A documents, follow-up email drafting from the transcript, account handoffs without sync meetings, proposals that update as the deal moves, newcomer onboarding from recorded calls, training recordings turned into step-by-step guides, and deal re-qualification from your meeting library.
Every use case below comes from a workflow I found in my team’s recorded calls. I also talked to our account executive, Jackie Kyrylenko, to get her opinion on the best AI use cases for sales discovery calls.
1. Brief Yourself on Past Discovery Meetings
Open your meeting library, pull up every conversation your team has had with the account, and ask the AI a direct question: “Did we discuss data hosting regions?” or “What business tools are they currently replacing?” This way, you do not need to skim through an hour of audio or guess which call covered which topic.
This matters most when more than one person on your team has talked to the same prospect. The prospect expects everyone at your company to know the conversation details, and when they have to repeat their use case to a third person, they start to question whether your team communicates internally.
Jackie described using Sembly in a similar way to settle a question mid-deal:
She also described the broader shift in how she works:
At 40-plus calls a month, no representative can hold all the details in their head, and that’s okay. All you need is a searchable knowledge base that keeps context alive and prevents deals from going cold between the second and third conversation.
2. Build Personalized Competitor Comparisons
After a discovery call where the prospect mentions your competitors (in our case, it used to be Fireflies or Gong), ask your AI to build a comparison document. The process takes three steps:
- Point the AI at the meeting you just had so it knows what the prospect cares about.
- Ask it to pull competitor data from the web.
- Specify that you want a value comparison for the prospect’s industry.
Jackie does this on almost every deal. Here is how she described one of the AI use cases for HR consultancy:
The result is a competitor comparison built for one prospect’s situation. Furthermore, it is delivered while the conversation is still fresh. What else can you ask for?
3. Create Battle Cards That Reflect Buyer Questions
Feed your AI a batch of recorded discovery calls plus web data, and ask it to produce a competitive battle card. The output combines two layers: objections and questions that prospects have raised in recorded conversations, and publicly available competitor data with source links for verification.
Most battle cards are built from internal brainstorms, which means they address the differentiators the marketing team came up with rather than the ones prospects bring up on live calls. A battle card built from recorded customer conversations reflects buyer concerns because it was extracted from buyer conversations.
Jackie had a similar experience with an engineering company:
The source references (meeting timestamps and web URLs) stay visible in the card, so when a prospect challenges a comparison point, the representative can verify where the claim came from.
4. Send a Q&A Document Instead of a Generic Summary
After the discovery call, generate a table of every question the prospect asked and every answer you gave. Send it as a follow-up instead of a bland “great talking to you” email.
A Q&A table ensures a decision-maker can scan it for the one question they care about, read the answer, and forward it to procurement or legal. A meeting summary paragraph in an email thread (even if it’s great) rarely gets forwarded. A brief table, on the other hand, does, which means your deal now has a document circulating inside the prospect’s organization within an hour of the call.
Jackie sends these after most of her discovery calls:
5. Draft Personalized Follow-Up Emails
Ask your AI to draft the follow-up email from the meeting transcript. Why? The transcript likely has all the specifics you promised during the call: the Trust Center link, the trial workspace, and the link to the onboarding documents.
A follow-up that says “great connecting, here are some resources” gets archived. However, a personalized follow-up that says, “Here is the Trust Center link you asked about, I will open your workspace on Sembly, and here is a link to the onboarding materials,” gives the prospect something to act on.
6. Handoff Client Accounts Without a Sync Meeting
When a deal moves from sales to customer success, share the meeting folder instead of scheduling a briefing call. The new account owner can read through the AI meeting minutes, ask internal AI chat for a timeline, and start the relationship already familiar with the use case, objections, and open commitments. This pretty much removes the 30-minute handoff meeting from the process entirely and solves the most common complaint prospects have about vendor transitions: having to explain everything again to a new face on the team.
I once again was curious whether our own sales team actually does it, so I asked Jackie:
This way, the prospect does not re-explain their setup, because the new representative already has the full context from every past call.
7. Generate Sales Proposals That Update as the Deal Moves
Point your AI at a folder of meetings with one prospect and ask it to draft a client deliverable. Structure it in a way the prospect needs. As new meetings happen and terms evolve, add the latest call to the chat and regenerate the document. For deals with five or ten conversations over a quarter, this keeps the proposal current without having to rewrite from scratch after each call.
Jackie described how one customer uses this approach:
As a result, each version reflects the full history of conversations with the prospect, so nothing said on call three gets lost by call seven.
8. Onboard New Sales Representatives on Real Conversations
Create a collection of your strongest discovery calls and share it with every new hire. Then ask the AI to generate a FAQ document from those recordings: common prospect questions, best answers, and a glossary of terms the team uses. The new representative can learn the best ways of working with clients by studying how experienced colleagues handled objections, data loss concerns, and pricing conversations on recorded calls, which is more useful than a training deck written two years ago.
As it appears, this is something Jackie is no stranger to:
9. Use Recorded Trainings as a Source for Step-by-Step Guides
The idea here is to upload training sessions, compliance recordings, or strategy meetings and ask the AI to produce a checklist with stages, action items, and source references. Every claim in the output traces to a specific recording and timestamp, which matters in compliance and regulated industries where an unsourced summary can create liability.
Jackie used this when our newcomer went through certification:
10. Re-Qualify Deals From Your Meeting Library
For longer sales cycles, keep qualification up to date by searching past meetings instead of scheduling check-in calls to reconfirm budget or a timeline with leadership. The prospect shared their budget range three calls ago and named the decision-maker on the second call. Asking again tells them you did not track what they said, which erodes confidence in your team.
Jackie uses the library for this daily:
The longer you think about it, the more sense it makes. I mean, for a deal with multiple contacts and a six-month sales cycle, searching your meeting history keeps the qualification fresh without making the prospect repeat themselves.
How to Use AI With BANT, MEDDIC, and SPIN Frameworks
Most sales teams follow a framework on their discovery calls, but the framework only works if the information it produces gets captured and used. AI handles the capture, and it makes the framework more useful after the call.
Here is how each framework works with AI meeting intelligence:
|
Framework
|
What It Tracks
|
How AI Helps
|
Best For
|
|---|---|---|---|
|
BANT
|
Budget, Authority, Need, Timeline
|
Tags each answer to the transcript. After the call, the rep checks whether all four are confirmed. If budget did not come up, the representative sees the gap and addresses it next time.
|
Shorter sales cycles with a single decision-maker
|
|
MEDDIC
|
Metrics, Economic Buyer, Decision Criteria, Decision Process, Identify Pain, Champion
|
Organizes meeting notes by all six MEDDIC headings. Blank sections show what is missing. Over multiple calls, the AI gathers the full picture across conversations.
|
Complex enterprise deals with buying committees
|
|
SPIN
|
Situation, Problem, Implication, Need-Payoff
|
Captures which stage the questioning sequence reached. If the representative covered Situation and Problem but never got to Implication, the summary shows that gap.
|
Consultative sales where the prospect needs to quantify the cost of inaction
|
AI Across the Stages of a Discovery Call
My team’s AI workflows are split into three groups that match the call lifecycle:
- Before the call: The rep searches past meetings with the account, reviews what colleagues discussed, and identifies open questions. For a brand-new prospect, they research the company first. I remind newer reps that the first thing they need to understand is the prospect’s business: their geography, their customers, what they do day to day. Everything on the call depends on that foundation.
- During the call: The AI agent records, transcribes, and separates speakers across more than 40 languages, including mixed-language calls. The rep stays in the conversation instead of typing notes. Jackie described the principle behind this: “The concept of Sembly is to have all of the processing after the meeting. We want our clients to concentrate on the conversation and not think about taking notes during the meeting.” Results come back quickly: usually 8 to 12% of the call duration.
- After the call: Meeting recaps, Q&A documents, competitor comparisons, proposals, and CRM pushes. Jackie described how this changed her daily work: “I don’t even remember what it’s like to write everything down. Sembly is there always, and I’m not even stressed about remembering things from the call.”
The three stages you see above work as a system. Skipping the pre-call research means you go into the conversation blind. Skipping the post-call processing means the conversation produces nothing. Most teams only use AI in the middle, which is also the stage where it has the least to offer.
Common Mistakes I Saw Sales Teams Make With AI During Discovery Calls
Across 50 recorded calls, three patterns showed up often enough that I think every sales team evaluating AI should know about them before they buy or deploy.
- The wrong tool for the job: Several prospects came to us directly after trialing Sembly competitors. In every case, the complaint was the same: the tool produced a transcript that they still had to process manually. They had not bought a note-taker; they thought they were buying a deliverable generator. Knowing the difference before you buy saves three months of a wasted subscription.
- Skipping pre-call research: On calls where a representative had not searched the meeting library beforehand, prospects occasionally flagged that they had already covered a topic on a previous call. It came up gently, but it came up. At 40-plus calls a month, no agent remembers every detail, and the ones who checked the library before dialing in never had that moment.
- Misunderstanding where AI gets its information: Multiple prospects assumed that AI competitor comparisons were pulled from some general knowledge base and were skeptical of the accuracy. Once they understood the tool pulls from their own recorded meetings and cites web sources separately, the skepticism flipped. They trusted it more, not less, because the sourcing was transparent.
All three patterns point to the same thing: AI on discovery calls works when the rep understands what the tool is for and uses it deliberately at each stage.
If you are evaluating AI for your sales team’s discovery calls, start with two use cases: pre-call research from your meeting library and post-call follow-up generation. Those two cover the prep and the admin, which is where representatives lose the most time.
Wrapping Up
Discovery calls have always been about information: getting enough of it, capturing it accurately, and using it before the deal goes cold. With AI, a sales representative no longer has to choose between staying in the conversation and remembering everything from it.
The 10 use cases in this article come from analyzing 50 recorded meetings of the Sembly sales department, but the underlying principle applies to any team. AI is most useful on discovery calls when it handles everything around the conversation, including preparation, documentation, and deliverables. This way, the representative can focus entirely on the client.
FAQ
What is a discovery call in sales?
A discovery call is the first structured conversation between a sales representative and a prospect who has shown buying intent. Its purpose is to confirm four things: the problem the prospect is trying to solve, who makes the buying decision, whether the budget is allocated, and what the timeline looks like.
How long should a discovery call be?
The ideal discovery call runs 25 to 30 minutes. This gives enough time for rapport, 12 to 15 qualification questions, a brief solution overview, and confirming next steps.
What is the ideal talk-to-listen ratio on a discovery call?
Top-performing sales representatives talk 40% of the time and let the prospect fill the remaining 60%.
What is the difference between a discovery call and a demo?
The short answer is that a discovery call happens before a demo. Its goal is to qualify the prospect and understand their use case.
A demo presents the product based on what the discovery call revealed. Running a demo before a discovery call often means presenting features the prospect may not care about.
What should you do after a discovery call?
Send a follow-up within the hour that includes every commitment made on the call, a Q&A document capturing the prospect’s questions and your answers, and a scheduled next step.
You can use agentic solutions like Sembly to auto-generate all 3 from the meeting content.
What are the best AI use cases for discovery calls
The best AI use cases for discovery calls are: searching past meeting recordings before the call, generating competitor comparisons after it, sending a comprehensive Q&A document post-meeting, drafting follow-up emails, and passing accounts to new team members.
Each one targets a specific stage where sales representatives lose the most time.
Can AI replace sales representatives on discovery calls?
No. AI handles documentation, recall across past meetings, and deliverable generation. The conversation, the judgment calls, and the relationship stay with the representative.
How does AI help with lead qualification after a discovery call?
AI maps conversation data to qualification frameworks, such as BANT or MEDDIC, and shows which criteria were confirmed and which are still open.
This way, representatives can search their meeting library to re-qualify deals without scheduling additional calls, and managers can review qualification gaps across the full pipeline.
How do you train sales representatives on discovery calls using AI?
The fastest approach is building a collection of your strongest recorded discovery calls and generating a FAQ document from them: common prospect questions, the answers that worked, and a glossary of product and industry terms.
