The Three Modes of Customer Understanding
Separate targeted Discovery, always-on Continuous Conversations, and passive Customer Listening — each needs its own machinery.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 88%
A taxonomy that distinguishes three distinct customer-understanding practices that teams routinely conflate: Discovery (targeted research to optimize a specific workflow), Continuous Conversations (weekly customer calls auto-scheduled onto the whole product team's calendars), and Customer Listening (systematically scarfing up the signals customers already emit across support, reviews, NPS verbatims, churn surveys, and closed-lost). Each requires different infrastructure, and the enemy of all three is friction.
Origin
Oji Udezue's framework, built across Twitter, Atlassian, Calendly, and Typeform; he credits Teresa Torres for the specific continuous-discovery tactic of a site pop-up linking to a Calendly booking, which he ran in Calendly's community.
Core principles
- 01Discovery, continuous conversations, and listening are three different jobs, not one
- 02The death of customer discovery is friction — if you ask individuals to do it, they won't
- 03Talking to one customer is useless; talking to 100 is super useful — targeting the right customer is everything
- 04Customer signals arrive constantly whether you listen or not; processing them efficiently is a core PM job
- 05The job is to meld customer delight with business ambition
How to run it
- 1
Run targeted Discovery for specific optimizations
When you want to optimize a particular workflow, identify the people most likely to be affected and use research strategies to talk to them about that specific thing.
- 2
Institutionalize Continuous Conversations
Arrange for PMs, designers, engineering managers, and PMMs to have customer conversations show up automatically on their calendars every week, with no effort on their part to book.
Pro tip Use a pop-up on your site or in-product inviting visitors to talk, linking to a scheduling tool so the call auto-appears on the calendar (the Teresa Torres tactic).
Watch out If you require individuals to arrange the calls themselves, the friction kills the practice — it's the thing that dies in innovative companies.
- 3
Train the team on how to have the conversation
Talking to the right customer well is a skill; equip the team so conversations are productive, not just calls with any customer.
Watch out Talking to one customer is useless and talking to the wrong customer is worse — put intelligence up front about who to invite and why.
- 4
Build a Customer Listening pipeline
Systematically capture the signals customers already emit — social, App Store and G2 reviews, churn survey verbatims, NPS verbatims, bug reports, Zendesk support tickets, and Salesforce closed-lost — and look at their frequency distribution.
Pro tip At Calendly, Udezue jury-rigged this through workflows and Slack with individuals triaging the signals.
- 5
Process the signals efficiently as a core discipline
Treat efficient processing of these constant signals as a big hairy audacious goal of product management, feeding what actually matters back into the roadmap.
Pro tip Target listening prompts at users in their activation or conversion journey — churned or inactive users won't talk, but users mid-journey often will.
In the wild
Following Teresa Torres's tactic, Calendly put a pop-up in its community (calendly.com community) inviting visitors to book a feedback call via Calendly, so conversations landed on the calendar automatically. Udezue notes the group was self-selected but still representative enough to be worth talking to.
→ A low-friction, always-on stream of customer conversations without individuals having to hunt for participants.
Udezue enumerates the signal sources firing constantly at a company: social, App Stores, G2 Crowd, churn surveys, NPS verbatims, Firefox-style bug reports at Atlassian, Zendesk support frequency distributions, and Salesforce closed-lost. He jury-rigged Calendly's triage through Slack workflows.
→ A structured way to hear what customers are already saying instead of only what you go asking about.
Common mistakes
Delegating discovery to individual initiative
Asking PMs and designers to arrange their own customer calls guarantees they won't happen — friction is the death of discovery. Conversations must be auto-scheduled onto calendars by default.
Talking to any customer instead of the right one
Volume without targeting is worthless — talking to one (or the wrong) customer is useless. Without intelligence about who to invite and why, the conversations don't represent your actual ICP.
Is it for you?
Best for
Product leaders building a repeatable customer-understanding operation across a whole R&D team.
Not ideal for
Very early solo founders who can and should talk to customers directly without formal infrastructure.
From the transcript
“the the death of customer Discovery is friction if you ask individuals to do customer Discovery they will not do customer Discovery”
“customer conversations show up on their calendar every week automatically without them getting involved”
“customer listening is different it's not really Discovery it is the scarfing up of customer signals that are happening constantly anyway”
“talking to one customer is useless talking to 100 is super useful”
From the episode
Picking sharp problems, increasing virality, and unique product frameworks
Oji Udezue (Typeform, Twitter, Calendly, Atlassian)