CHAPTERS
- 0:00 – 2:02
Why discovery is the highest-leverage PM work + Jira Product Discovery’s 4-stage framework
Aakash frames product discovery as the most high-leverage PM task and sets up how AI can accelerate it. Tanguy introduces the team’s shared vocabulary for discovery and delivery: Wonder, Explore, Make, and Impact, designed to prevent misalignment about how “real” an idea is.
- •Discovery is positioned as the #1 leverage point for PMs, with AI amplifying it
- •The team uses a simple staged model that looks waterfall but behaves iteratively
- •Stages create common language across Atlassian to avoid premature assumptions
- •Each stage implies different asks (dependencies, funding, resourcing, etc.)
- 2:02 – 4:34
Stage 1 — Wonder: rapid problem exploration through customer interviews
Tanguy explains Wonder as structured problem discovery: talking to customers until themes repeat and the pain is understood clearly. The goal isn’t scientific breadth; it’s high-signal depth and shared understanding of the problem’s facets.
- •Wonder focuses on problem exploration, not solutions
- •Typical approach: ~a dozen hour-long user interviews until themes converge
- •Output is clarity on what’s happening and why it hurts, in customer language
- •Used at both macro (new product) and micro (feature) levels
- 4:34 – 6:05
How JPD operationalizes customer feedback: rotation, tagging, and roadmap refresh
The team runs discovery continuously via a PM rotation that triages feedback, tags insights to ideas, and reviews changes weekly and monthly. They also classify work into big bets, iterations, and small “pebbles,” tying it into different roadmap layers.
- •Weekly PM rotation reviews all incoming feedback and links insights to ideas
- •Monthly cadence revisits what’s changed and whether roadmaps need updating
- •Work is categorized: big bets vs iterations vs small UX pebbles
- •Multiple roadmap layers: leadership-level bets and team-level solution roadmaps
- 6:05 – 12:27
A concrete Wonder deliverable: persona-focused “problem page” built from video evidence
Tanguy walks through a real Wonder outcome (“global fields”/previously “global entities”) driven by customers needing scale across many JPD projects. The key artifact is not a long doc—it’s curated short video clips that make the pain undeniable and discussable in one meeting.
- •Example exploration targets product ops, program mgmt, VPs, stakeholders, PMs
- •Requirement: <10 minutes of customer video clips covering multiple facets
- •Team watches silently together, then aligns quickly on importance/next steps
- •Wonder emphasizes raw customer words over polished narrative
- 12:27 – 15:31
How to run better user interviews: stop leading, embrace silence, get researcher training
Tanguy argues most PM interviews accidentally confirm existing beliefs due to leading questions and interruptions. He shares practical interviewing behaviors—minimal script, no options, no interrupting—and recommends learning directly from professional researchers.
- •A researcher review revealed his interviews were overly leading and uninformative
- •Use a simple opening script, then follow the user’s direction with probing
- •Never introduce the concept you’re testing (e.g., don’t say “feedback” first)
- •Avoid offering options; ask open questions and wait through silence
- 15:31 – 18:21
Making the Wonder doc fast: clip “aha” moments and tell a story; don’t overwrite
The “document” is primarily a narrative made of customer clip highlights that compress 10–15 hours of interviews into ~10 minutes. Writing should be minimal; the work is tagging moments right after calls and waiting until insights clearly converge before publishing.
- •Text is secondary; the clips do the convincing and create urgency
- •Tag highlights immediately after interviews to capture true “aha” moments
- •Publish when conversations start repeating (the ‘sand settles’)
- •The final page should take only 2–3 hours once clarity is reached
- 18:21 – 21:32
Tooling and systems to stay close to customers (and where AI helps vs. doesn’t)
Tanguy outlines the mechanics that make customer learning “no-thinking” work: recruiting, scheduling, recording, snippet extraction, and querying. AI is valuable for routing you to the right conversations and extracting snippets, but direct exposure to customers is still the advantage.
- •Use tools like Pendo segmentation, Dovetail transcripts/snippets, Loom reels
- •Automate scheduling and outreach (short, human emails + Calendly)
- •AI can help query transcripts, find themes, and surface moments—but shouldn’t replace listening
- •Set up customer touchpoints as the first priority when joining a team
- 21:32 – 28:46
Scaling feedback loops: community posts, in-app questions, Slack ‘speed dial,’ and CSAT results
The team runs multiple high-frequency channels (community, in-app surveys, Slack, service desk) to stay close to users even at 18,000 customers. Tanguy links this “close to the ground” approach to maintaining CSAT above 85 while scaling.
- •Community enables rapid bug response and feature experimentation at scale
- •In-app polling (with ‘tell us more’ comments) helps resolve debates quickly
- •Direct Slack relationships with key customers accelerate iteration
- •High-touch learning correlates with sustained CSAT >85 during growth
- 28:46 – 36:03
Stage 2 — Explore: validate solutions with prototypes before writing code
Explore is about iterating toward a solution users explicitly say will solve their stated problem. Tanguy shares early JPD prototypes—from a single slide to Figma concepts—showing how reactions (lukewarm vs. urgent) guided them to the right direction, now accelerated with tools like Lovable.
- •Explore exit criterion: customers explain how the proposed solution solves their pain
- •Start lo-fi (even a slide) and iterate rapidly based on user reactions
- •Contrast: ‘interesting’ feedback vs. ‘when can I get this?’ urgency signal
- •Modern approach: generate playable prototypes quickly (e.g., Lovable from Figma)
- 36:03 – 39:25
Stage 3 — Make: build with a ‘safety funnel’ rollout to avoid bad early experiences
In Make, the team commits to building but controls exposure using a safety funnel: 10 → 100 → 1,000 → broader release. This reduces negative first impressions that are hard to reverse, and it helped JPD convert from early trials to rapid paid growth.
- •Safety funnel prioritizes minimizing bad experiences over maximizing exposure
- •Controlled scaling prevents ‘not for me’ abandonment from unready product states
- •JPD’s conservative ramp enabled strong growth: ~1,000 trials to 18,000 paying in <2 years
- •Be methodical unless your distribution dynamics truly demand viral speed
- 39:25 – 43:01
“Release fast” vs. “validate the right risk”: waitlists, distribution tests, and messaging
Tanguy reframes the debate: don’t ship fast by default—answer your biggest unknown. For JPD, they validated demand and distribution via a website signup message (3,000 companies in two weeks) before the product was fully built, then used segmentation and progressive access.
- •Identify the highest-risk question (solution fit vs. willingness to pay vs. distribution)
- •Early demand test: website messaging + email capture yielded strong signal
- •Use waitlists and segmentation to manage readiness across customer types
- •Avoid exposing unready experiences to broad audiences; winning back is hard
- 43:01 – 47:07
Free users vs. paid signal: when ‘process change’ is the real commitment
They discuss whether free usage produces misleading feedback. Tanguy argues commitment can come from the effort to adopt and configure workflows (especially in B2B collaboration), but stresses validating willingness-to-pay through research (e.g., conjoint) and budget reality.
- •Adoption effort can be a proxy for payment intent in high-switching-cost B2B tools
- •JPD tracked engagement depth (e.g., daily active usage among active PMs)
- •They validated monetization assumptions with conjoint surveys and strategy support
- •Payment timing depends on market dynamics, criticality, and budget availability
- 47:07 – 56:02
Infusing discovery into Make and Impact: shaping V0, technical spikes, and rapid iteration loops
Tanguy explains how discovery continues during delivery: technical spikes reduce feasibility risk, designs are reset with engineers involved, and scope is repeatedly expanded/contracted to reach a minimal V0 quickly. They iterate using weekly demos, living feature docs, and fast internal testing via code branches before widening customer access.
- •Use technical spikes in Explore to understand feasibility and complexity
- •Reset designs in Make with PM/Design/Engineering together; then carve down to V0
- •Weekly demos + a living feature document keep scope and learning aligned
- •Branch-based previews let PMs test changes immediately and iterate faster with customers
- 56:02 – 57:48
Where to connect: LinkedIn, JPD Community, and continued resources
Tanguy shares the best ways for viewers—especially JPD users—to reach him and share stories or feedback. Aakash closes by pointing to the full podcast, the newsletter with linked artifacts, and subscription/review calls to action.
- •Primary contact channels: LinkedIn and Jira Product Discovery Community
- •Community is actively monitored and used for real-time feedback loops
- •Aakash offers a newsletter post with tools, frameworks, and public links
- •Episode wrap-up and ways to support the podcast
