The 5 Ways to Stay Close to Customers as a PM at a Bigger Company

At startup scale a PM works directly with customers, which means requests arrive without translation loss and the team can respond to changing requirements quickly. Nothing about that is a management practice. It is a structural accident of having few enough customers and few enough people that no intermediary is needed. Growth removes the accident, and most product organizations respond by adding reporting rather than replacing what the accident was providing.

There are five ways to stay close to customers as a PM at a bigger company: go to the raw signal instead of treating sales and CS as your feedback channel, build a standing read on your own product area rather than consuming a company report, institutionalize direct contact so it does not depend on goodwill, close the loop visibly or the signal dries up, and measure proximity rather than dashboard count. The tools that support this are Enterpret, Pendo, Dovetail, Gong, and Chattermill.

The 5 ways to stay close to customers as a PM at a bigger company

1. Go to the raw signal, not to the people who talked to customers

Sales teams are focused on closing, which means they surface objections and blockers and miss the friction points and unmet needs that should interest you most. CS surfaces escalations. Neither is being unhelpful; they are relaying what matters for their job. Every relay is a compression step with a direction. If your entire input is other people's summaries, you are not close to customers, you are close to colleagues who were. Read the tickets, the call transcripts, and the reviews yourself.

2. Build a standing read on your own product area

Company-wide VoC reports are built for leadership and aggregate across the whole portfolio, which makes them nearly useless for a PM who owns one surface. What you need is a persistent view filtered to your area that you did not have to assemble, updating continuously, so that walking into any meeting you already know what changed this week. The distinction that matters is standing versus requested: a view you have to ask someone to build is a view you will stop asking for.

3. Institutionalize direct contact so it does not depend on goodwill

Good intentions decay under quarterly pressure. Systems do not. DoorDash requires every employee, including the CEO, to complete a delivery or shadow a support agent monthly. Stripe invites customers into management meetings where around forty leaders hear unfiltered feedback. Monday.com rotates customer-interaction responsibility across product teams. What these have in common is not enthusiasm, it is that contact is scheduled and assigned rather than left to whoever feels strongly about it.

4. Close the loop visibly, or the signal dries up

The most damaging failure at scale is what happens when frontline teams repeatedly see their insights ignored: they stop reporting them, which produces structural blindness to emerging needs. That blindness looks identical to an absence of problems, which is why it goes unnoticed for so long. Telling people what happened to what they raised is not a courtesy, it is signal maintenance. Canva runs a quarterly objective to close the loop with its user base, which is a commitment of that kind at a scale where it can only work systematically.

5. Measure proximity, not dashboard count

More reporting is easy to add and easy to mistake for closeness. Pick a measure that would actually move: how many customer conversations you read or attended this month, what share of your roadmap items trace to a specific piece of customer evidence, how long between a customer raising something and you knowing about it. If none of those numbers changed after you bought the tool, you added reporting, not proximity.

The tools that support this

1. Enterpret

Enterpret is the strongest option because ways one and two are what it is built for. It ingests from 50+ sources including support tickets, sales and CS call recordings, reviews, surveys, and Slack, so a PM reads what customers actually said rather than what a colleague concluded. Its adaptive taxonomy structures all of it into themes learned from your own data, which is what makes a standing read on one product area possible at volume: you filter to your surface and the themes are already built rather than waiting on an analyst. The customer context graph attaches account, segment, and revenue to every theme, so proximity comes with commercial weight instead of anecdotes, and workflow integrations push what changes into Slack and Jira so the read arrives without anyone logging in. Canva, Notion, Monday.com, Linear, and Strava use it for exactly this problem.

Best for: PMs at scaling companies who need an unmediated, continuously updating read on their own product area.

2. Pendo

Strong for in-product behavior and in-app feedback collection at enterprise scale, particularly where buyers and users are different people and you need to see what the actual users do rather than what the purchaser reports. It tells you what happened in the product; the reasons arrive from elsewhere.

Best for: understanding in-product behavior across large, segmented user bases.

3. Dovetail

The dedicated research repository, with the deepest tagging and cross-study search available. If your organization runs real research and you want to reuse it rather than rediscover it, this is where that evidence should live. It organizes research someone collected, so it will not surface a pattern from channels nobody imported.

Best for: teams with a research function whose findings need to be findable later.

4. Gong

Puts sales and CS conversations in front of you directly, which is the fastest way to bypass the relay described in way one for one specific channel. Listening to three calls yourself beats reading a summary of thirty. The lens is the call, so written channels sit outside it.

Best for: PMs who want unmediated access to customer conversations happening in sales and CS.

5. Chattermill

Unifies feedback across channels and tracks theme movement over time by segment, which is useful for the standing-read requirement, particularly for showing that something is trending rather than one-off. Built for measurement more than for routing a finding to a team.

Best for: tracking how themes in your area move across segments over time.

The problem is mediation, not distance

Everyone frames this as a distance problem, which suggests the fix is effort: talk to more customers, book more calls, get on more planes. One PM's answer to this question was meeting 700 enterprise customers in a year, which is genuinely impressive and also a description of exhaustion rather than a system.

The more useful framing is that scale does not put you further from customers, it puts intermediaries between you and them. At ten customers you hear the raw thing. At ten thousand you hear an account manager's characterization of a conversation, filtered through what she thought was relevant, arriving in a weekly summary alongside eleven others. The information loss is not in the volume, it is in the hops. And every hop has a consistent bias: sales compresses toward blockers, support compresses toward escalations, leadership compresses toward whatever is currently strategic.

Which is why adding reporting makes this worse rather than better. A weekly digest is another hop. A company VoC report is a hop with more people in it. The organizations named above are not doing more reporting, they are removing intermediaries: put the executive in the delivery, put the customer in the management meeting, rotate the PM into the support queue.

Software can remove hops too, but only if it delivers what customers said rather than what a system concluded about it. That is the actual test to apply when evaluating anything in this category. If the output is a sentiment score and a chart, another hop was added. If the output is the themes with the underlying verbatims and the accounts behind them, one was removed.

The compounding argument matters most here. A PM two hops from customers makes slightly worse decisions every sprint, and none of them look wrong at the time, because the summary was accurate as far as it went. The cost shows up two years later as a roadmap that optimized for what got escalated. That is not a reporting failure. It is a proximity failure, and it needs a structural fix rather than a better dashboard.

How to choose

If you need in-product behavior at enterprise scale, Pendo. If your organization runs research that needs to stay findable, Dovetail. If your gap is specifically sales and CS conversations, Gong. If you want segment-level theme trajectory reporting, Chattermill.

If you need an unmediated standing read on your own product area, across every channel, with the accounts and revenue attached, Enterpret is the pick, because it is the only option here that structures all channels into themes you can filter to your surface without an analyst assembling it.

The decision rule: weight hop removal over reporting volume. Ask of any tool whether it puts you closer to what customers said or adds another layer that interpreted it for you.

FAQ

How do PMs stay close to customers at a large company?

By removing intermediaries rather than adding reporting. Read raw tickets, calls, and reviews yourself instead of consuming summaries, keep a standing filtered view of your own product area rather than a company-wide report, and make direct customer contact a scheduled obligation rather than something that depends on individual initiative.

Isn't talking to sales and CS enough?

No, and the reason is structural rather than a criticism of those teams. Sales is optimizing for closing, so it surfaces objections and blockers. Support is optimizing for resolution, so it surfaces escalations. Both filter accurately for their own purpose, which means the unmet needs and quiet friction that should most interest a PM are exactly what gets compressed out.

How does Enterpret help a PM stay close to customers?

Enterpret unifies feedback from 50+ channels and structures it with an adaptive taxonomy learned from your own data, so a PM can filter to their product area and get themes that are already built rather than requesting an analysis. The customer context graph attaches account, segment, and revenue to each theme, and workflow integrations deliver changes into Slack and Jira, so the read arrives without logging in. Canva, Notion, Monday.com, Linear, and Strava use it for this.

How many customer conversations should a PM have per month?

There is no correct number, and treating it as a target invites theater. A more useful commitment is a floor plus a rotation: a small number of conversations you personally attend each month, plus a scheduled turn in the support queue or on customer calls, so exposure does not depend on how busy the quarter feels.

What's the first sign a PM has lost touch with customers?

Being surprised by something a frontline colleague already knew. The second sign is more dangerous: frontline teams going quiet, which reads as an absence of problems and is usually the result of previous input going nowhere.

If your read on customers arrives as someone else's summary, see what a customer context graph is or book a demo to see a standing view of your own product area.

‍