Start a discovery call with the change the buyer wants, then understand the current process, the problem and its impact, what success means, and how a decision will be made. If a prospect wants to reduce manual reconciliation, ask what causes the work, who it affects, and how they would recognize improvement before discussing budget or presenting a demo.
A discovery call is a two-way conversation about need and fit. Its useful outcome is a shared understanding of the problem and a decision about a relevant next step: a demo, another conversation, a later follow-up, or no further action. Use the questions below as a guide, not a script. Skip information that is already current and follow the buyer’s answers.
Before the call, review the relevant CRM record and recent activity. Treat existing information as preparation, not as a substitute for confirming what is still true.
What discovery call questions should you ask?
The shortest useful sequence is: understand the current state, identify the problem and its impact, define the desired outcome, explore timing and the decision process, then agree on a specific next move. Open with a brief agenda, confirm the time available, and ask permission to explore the prospect’s situation before moving into qualification details or a solution presentation.
A useful discovery call produces a shared understanding of the problem and a clear decision about what happens next, not a completed checklist.
Use a question sequence that follows the buyer’s reasoning
1. Understand the current state
- “Could you walk me through how you handle this today?”
- “What part of your role or process does this affect?”
- “What have you already tried?”
Ask for a concrete example when the process is unclear. Capture the buyer-described process, relevant role, and any known stakeholders. If the detail is unavailable, record it as an open question rather than filling the gap from assumptions or pre-call research.
2. Identify the problem and its impact
- “What are you trying to change?”
- “What tends to cause that problem?”
- “Who else is affected, and what happens when it occurs?”
- “What makes this a priority now?”
- “Why has it not been addressed before?”
Invite the buyer to describe consequences in their own terms. Do not suggest a revenue loss, cost, or risk and then record it as confirmed. If the impact matters but is not yet clear, ask for an example and leave the field unresolved until someone can verify it.
3. Define the desired outcome
- “What would a better outcome look like?”
- “How would you recognize that the change worked?”
- “Is there a measure or target you already use?”
- “If you do not choose a product, what is your plan for addressing the problem?”
Capture the stated result and any measure with its unit and context. If the buyer describes a goal but gives no number, record the goal without inventing a success metric. An existing alternative plan may also clarify urgency and the consequences of inaction.
4. Explore timing and the decision process
When the need and outcome are clear, ask what must happen to make a change:
- “What is driving the timing?”
- “Who else will weigh in?”
- “How does your team evaluate options?”
- “Are there implementation, legal, or procurement steps we should account for?”
- “What resources or budget would be involved?”
Ask about budget when it is relevant and appropriate to the conversation. BANT means budget, authority, need, and timeline. It can organize qualification, but it is not a validated universal score or an automatic reason to reject a prospect. A buyer may have a real need before a budget is approved or every decision-maker is known. Record gaps as unresolved and agree how to learn more.
Run the call as a conversation, not a questionnaire
Set expectations at the start: explain that the purpose is to understand whether there is a useful fit, confirm the time available, and ask permission to explore the prospect’s situation. Ask one question at a time. When an answer raises a material point, follow that thread instead of moving mechanically to the next prepared prompt.
Before discussing a solution, reflect the situation back in the buyer’s words: “It sounds like the reconciliation work is slowing the monthly close, and you want to reduce it before the next planning cycle. Have I understood that correctly?” Invite a correction. If the picture is coherent and a plausible fit exists, suggest a relevant next step. If a critical point is unclear, clarify it or assign follow-up. If discovery is complete, do not fill the remaining time with an unrequested product tour.
Use empathy without manufacturing certainty. “That sounds frustrating” can acknowledge the experience, but the CRM should contain the buyer’s stated facts, not the rep’s interpretation. A learning mindset is more useful than a feature-led pitch.
Record call evidence separately from deal qualification
A call note records what was discussed in one activity. A deal field represents the team’s current, reviewed understanding of one opportunity. A contact field is better suited to durable facts about the person, such as their role, rather than a qualification conclusion about a particular deal.
For example, “wants to reduce manual reconciliation” is an observation from one call. It becomes a deal-level desired outcome only after a rep verifies that the statement applies to this opportunity and updates the deal record. Keep the activity as the source and link the accepted deal value back to it where the CRM configuration allows.
A transcript statement belongs to one call; it is not automatically the current truth about a deal or contact. Preserve the source activity, then let a reviewer decide whether its evidence changes the deal’s accepted qualification fields.
HubSpot documents activities such as calls, meetings, notes, and tasks separately from ordinary CRM object properties. Records can display associated activities, but availability and behavior depend on the object, integration, subscription, configuration, and permissions. See HubSpot’s guidance on activity properties and logging activities on records.
For a deal-focused process, save the conversation as an activity and maintain reviewed, deal-specific qualification fields on the relevant deal. Store contact information on the contact when it describes the person, not the opportunity. If your team is unsure which record should own each field, review your CRM systems consulting options.
Distinguish four answers in the CRM: unknown means not established; not discussed means the call did not cover it; declined means the prospect chose not to answer; and no means the prospect explicitly said no. “Budget not discussed” is not “no budget.” Useful fields include problem, impact, desired outcome, success measure, timing status, decision roles, purchasing steps, open questions, source activity, next-step owner, and due date.
Use AI for candidate extraction, with a human review gate
For eligible configurations, HubSpot Conversation Intelligence can transcribe and analyze calls from supported recording sources. Availability depends on subscription, seat, recording source, and configuration. HubSpot also documents different record behavior for calls and synced meetings. Check the current recording and transcript guidance and transcript synchronization guidance for your setup. Confirm recording consent, privacy, retention, and access requirements before enabling recording or transcription.
A transcript can help a reviewer find candidate values, but the documented capabilities do not establish a universal native workflow that extracts every discovery answer, validates it, obtains approval, deduplicates observations, and writes qualification fields. The following is a proposed implementation pattern, not a HubSpot feature guarantee:
- Input: receive a transcript and source activity ID with an intended deal ID. Check that the activity belongs to the correct prospect and opportunity.
- Candidate extraction: identify only statements supported by the supplied transcript and return evidence text or a timestamp. Return unknown for absent or ambiguous information.
- Validation: verify required fields, permitted values, activity-to-deal association, and contradictions with accepted values. Use deterministic rules for missing owners, dates, associations, or required fields.
- Human review: have a rep or manager inspect evidence, resolve conflicts, and approve consequential values such as budget, authority, business impact, qualified status, or stage advancement.
- Write-back: update the deal only with approved values, preserve reviewer and review time, and create a follow-up task when the review identifies an unresolved question.
{
"source_activity_id": "illustrative-meeting-4821",
"target_deal_id": "illustrative-deal-903",
"extraction_version": "illustrative-v1",
"observations": [
{
"field_name": "desired_outcome",
"candidate_value": "Reduce manual reconciliation",
"evidence": "We want to reduce manual reconciliation.",
"evidence_timestamp": null,
"review_status": "needs_human_review"
},
{
"field_name": "budget_status",
"candidate_value": "unknown",
"evidence": null,
"evidence_timestamp": null,
"review_status": "not_stated"
}
]
}The example is illustrative and is not a HubSpot-published schema. At the raw observation grain, one row represents one candidate field from one source activity and extraction version. If multiple evidence passages are retained, store them as separate evidence spans rather than combining them into an ambiguous observation. A proposed uniqueness key is source_activity_id + field_name + extraction_version. If multiple evidence spans for the same field must be retained, use an evidence-span identifier in the key instead.
Concurrent processors should enforce that key with a database uniqueness constraint or transactional upsert. A lookup followed by create can still produce duplicates. Preserve the activity ID, evidence timestamp where available, reviewer, review time, previous value, and new value. Do not let a replay overwrite an accepted deal value without review. These are implementation recommendations, not HubSpot-published schema requirements.
For guidance on aligning HubSpot records, properties, and workflows, see HubSpot systems consulting.
Choose a capture method that fits the team’s process
| Method | Best fit | Validation and owner |
|---|---|---|
| Live playbook | Reps need consistent prompts during a call. | The rep checks responses and associations. HubSpot playbooks are available in Sales Hub or Service Hub Professional and Enterprise, subject to seats and permissions. Updating properties from responses requires Enterprise conditions and appropriate permissions. |
| Rep-reviewed notes | Calls vary, or structured capture is unavailable. | The rep records the buyer’s wording, open questions, and next step as an activity, then updates reviewed deal fields. |
| Transcript-assisted review | Recording and transcript review are appropriate and available. | The rep verifies candidate evidence and approves consequential field changes. The reviewer checks the recording source, association, access, and consent requirements. |
HubSpot documents an important playbook behavior: if multiple questions update the same initially empty property, only the last iteration is saved. Map separate questions to distinct properties unless that overwrite is intentional. A playbook can standardize prompts and capture responses, but it is not by itself an approval process. Do not use free-text notes as the sole trigger for an automated deal-stage change. See HubSpot’s playbook documentation for current availability and permission conditions.
Close with an owned, dated next step
Summarize the problem, impact, desired outcome, and unresolved questions. Ask the prospect to correct the summary, then agree on a specific action. “Follow up next week” is incomplete unless the action, owner, and date are clear. Save the activity and follow-up task to the intended CRM record.
If fit is uncertain, record what must be learned and who will learn it. If it is not a fit, note the reason supported by the conversation rather than converting unanswered questions into negative answers. A qualification completeness gate can require a documented problem or desired outcome, at least one stated consequence, a known or explicitly unresolved decision context, a timeline status, an owned next step, and no unresolved record or privacy exception before progression.
- The prospect has confirmed or corrected your summary.
- The next action is specific and has a named owner and due date.
- The activity and any approved qualification update are on the intended record.
- Unknowns or conflicts have a named person responsible for resolving them.
- The follow-up task is saved, not left as an informal promise.
Teams can review process quality using internal measures such as calls with an owned next step, unresolved record associations, duplicate observations, field corrections after review, opportunities advanced without required evidence, and time from call completion to reviewed CRM update. These are operational measures, not universal sales benchmarks or proof that a question framework improves conversion.
Frequently asked questions
How long should a discovery call last?
There is no verified universal duration benchmark in the sources reviewed here. Schedule enough time for the purpose of the conversation, confirm the available time at the outset, and stop when useful discovery and a clear next step are complete.
Is a discovery call a sales interview?
No. It is a two-way conversation about the prospect’s needs, current process, decision context, and fit. The rep should listen and adapt rather than run through every prepared question or turn the call into a disguised product pitch.
Does budget need to be known on the first call?
No. Ask about budget when it is relevant and appropriate, but record it as unknown or not discussed when it has not been established. Agree on how to resolve it if budget is material to the next decision.
What should the CRM contain after the call?
Save the call or meeting as an activity, preserve the buyer’s stated problem and outcome, record reviewed deal-level qualification separately, and create a follow-up task with an action, owner, and due date. Keep unanswered questions visibly unresolved.
References: HubSpot’s discovery call question guidance and sales qualification guide provide editorial context. The latter describes BANT as a framework, not a validated prediction model. Product details above are based on the linked HubSpot Knowledge Base documentation and remain subject to the applicable account configuration.