To turn Gong sales calls into feature requests, connect Gong to your feedback tool so that an AI reads each finished call transcript, pulls out every product ask with a short quote and the speaker, and drops it into a review queue. A product manager then approves each one as a new request or as a vote on an existing request, with the deal attached. With FeatureOS, that is the Gong integration: it checks Gong every hour, skips private calls, stores the quote and deal details (never the full transcript), and emails the rep when the feature ships.

The rest of this post covers why sales calls are worth mining, what to pull from each call, how the manual and AI approaches compare, and how to set it up step by step.

Why collect product feedback from sales calls?

Sales calls are where prospects say what they need before they buy, in their own words, with a deal on the line. Most of that never reaches the product team. A rep hears "we would sign if you had SSO" on Tuesday, mentions it in a Slack thread on Thursday, and by the next planning cycle it is gone.

Conversation intelligence tools like Gong already record and transcribe these calls. The gap is the step after: getting the product asks out of hundreds of transcripts and into the place where you prioritize. Doing that by hand means someone listens to calls or reads transcripts, which rarely survives a busy quarter.

Feedback from sales calls is also different from what lands on a public board. It comes from people who are not customers yet, and it comes with money attached. That makes it useful for one question your board alone cannot answer: which requests are standing between you and revenue in the pipeline? If you already weigh requests by who asked, see how to prioritize feature requests by customer segment.

What should you extract from a sales call?

For each product ask in a call, capture enough to judge it later without re-listening. A useful checklist:

  • The ask, in one line. "Export reports to CSV on a schedule," not "talked about reporting."
  • A short verbatim quote. The customer's own words keep the request honest when it gets summarized later.
  • Who said it. Name and role. A request from the economic buyer weighs differently from one from a trial user.
  • A link back to the call. So anyone can hear the context in Gong.
  • The deal. Deal name, stage and amount, so the request carries pipeline value.
  • Whether it is a blocker. "We cannot buy without this" and "would be nice someday" are different requests.
  • The workaround they use today. It tells you how painful the gap really is.
  • The matching existing request, if any. Most asks are repeats. Count them as votes instead of creating duplicates.

Leave out small talk, pricing negotiation and complaints about competitors unless they name a missing capability. The goal is product signal, not a call summary.

Manual vs AI: how do the two approaches compare?

Manual (reps log it) AI extraction from transcripts
Who does the work Rep after each call, or a PM reviewing recordings AI reads each transcript, a PM reviews the output
Coverage Only calls someone remembers to log Every call that matches your filters
Wording Rep's paraphrase Short verbatim quote plus speaker
Deal context Often missing Attached from the call's deal
Duplicates Easy to create Suggested as a vote on a matching request
Delay Days, or never Hours after the call ends
Main risk Requests get lost AI flags things that are not real asks, so a human approves each one

The manual route works for a team with a handful of calls a week and a disciplined rep or two. Past that, logging becomes the first thing dropped when the quarter gets tight. The AI route does not remove the human. It moves the human from typing to approving, which is a much smaller job.

How do you turn Gong calls into feature requests with FeatureOS?

FeatureOS connects to Gong and runs the extraction for you. The integration is on the Growth plan and above, with no extra fee. Call analysis uses your workspace's existing AI credits.

Step 1: Connect Gong

In FeatureOS, go to Settings > Integrations > Gong and click Connect. Sign in to Gong as a technical administrator and click Allow.

FeatureOS asks for read-only access: basic and extensive call data, call transcripts, and users. It cannot change anything in Gong.

Step 2: Choose which calls to read

Four settings control what FeatureOS picks up:

  • Workspace IDs. Limit the integration to specific Gong workspaces.
  • Tracker IDs. Limit it to calls that hit specific Gong trackers.
  • External calls only. On by default, so internal meetings are skipped.
  • Create customers for unmatched participants. Off by default. Turn it on if you want call participants who are not already in FeatureOS added as customers.

Start narrow. One workspace and external calls only is a sensible first week.

Step 3: Let the sync run

The sync is one way, Gong to FeatureOS. Every hour, FeatureOS looks for calls that ended about three hours earlier. Calls marked private in Gong are skipped.

For each call, AI reads the transcript and pulls out the feature requests. For each request, FeatureOS stores:

  • a short verbatim quote
  • the speaker
  • a link to the call in Gong
  • the deal name, stage and amount

Full transcripts and call media are never stored in FeatureOS.

Step 4: Review the Suggestions inbox

Extracted requests land in the Suggestions inbox, not on your board. An admin reviews each one and either approves it as a new post or adds it as a vote on a matching post. Nothing reaches your board without that approval.

This is where you catch the AI's mistakes: a passing comment read as a request, or two different asks treated as one. If you already run a triage routine for other sources, fold this inbox into it. AI feedback triage workflows covers how to keep a human checkpoint in that loop.

Step 5: Prioritize with pipeline revenue

Once a request has deals attached, the post shows the pipeline revenue behind it. If the attached deals use different currencies, the total is left blank rather than adding up mismatched numbers.

Pipeline revenue is one input, not the answer. A request with a large open deal behind it may still be the wrong thing to build. Read it alongside votes, customer segment and fit with your roadmap. For turning that into a plan, see how to turn feedback into a roadmap.

Step 6: Close the loop with the rep

When the post's status changes to Completed, the rep who was on the call gets one email. They can go back to the prospect or customer with a real answer instead of a guess. That is the sales side of closing the feedback loop.

Full setup details are in the help article: Gong integration: turn customer calls into feature requests. The integration overview is on the Gong integration page.

What does a good sales-call feedback process look like?

The tool handles extraction. The process around it decides whether the requests get used.

  1. Pick one owner for the Suggestions inbox. Usually a PM. Shared ownership means nobody reviews it.
  2. Review on a fixed rhythm. Twice a week keeps the queue short and the context fresh.
  3. Prefer votes over new posts. If a matching request exists, add the call as a vote. Your board stays clean and the count stays honest.
  4. Tell sales what happens to their calls. Reps trust the process more when they know a completed feature means an email to them.
  5. Check the filters after the first month. If internal calls or irrelevant trackers slip through, tighten the settings.

What are the limits?

Be clear about what this does and does not do:

  • It reads, it does not decide. AI suggests requests. A person approves them and a person prioritizes.
  • Private calls are skipped. If a rep marks a call private in Gong, FeatureOS does not see it.
  • It is not real time. Calls are picked up about three hours after they end, on an hourly check.
  • No transcript archive. FeatureOS keeps the quote, speaker, link and deal fields. For the full conversation, follow the link to Gong.
  • It uses AI credits. Call analysis draws on your workspace's existing AI credits.

Pricing

The Gong integration is included on the Growth plan and above, with no extra fee. FeatureOS plans are seat bundles: Starter is $60/month for 5 seats, Growth is $120/month for 10, Business is $250/month for 15, and extra seats are $15/month each. There is a 14-day free trial. Full details are on the pricing page.