K
KnowMBAAdvisory
ProductIntermediate7 min read

Product Discovery Process

Product discovery is the continuous practice of deciding WHAT to build before delivery teams build it. Marty Cagan's framing at SVPG: discovery answers four risks — value (will customers buy/use it?), usability (can they figure it out?), feasibility (can we build it?), and business viability (does it work for our business?). Teresa Torres' Continuous Discovery Habits formalized the cadence: weekly customer interviews, opportunity mapping, and small-but-frequent assumption tests. Companies that run discovery in parallel with delivery (dual-track agile) ship 2-3x fewer features but those features get adopted at 4-5x the rate of feature-factory peers. Discovery is the difference between shipping software and shipping outcomes.

Also known asContinuous DiscoveryDual-Track AgileDiscovery TrackProblem DiscoveryDiscovery Habits

The Trap

The trap is treating discovery as a one-time phase that happens before sprint planning, then never again. Teams 'do discovery' for two weeks, then disappear into a 6-month build cycle, only to launch features nobody uses. The opposite trap is 'discovery theater' — running interviews and workshops to validate decisions you've already made. Teresa Torres calls this 'confirmation discovery': you ask leading questions, ignore disconfirming evidence, and walk away feeling validated. Real discovery means being willing to kill your favorite roadmap item when the evidence says it solves no problem.

What to Do

Adopt Torres' minimum-viable cadence: (1) Interview at least one customer per week per product trio. (2) Maintain a shared Opportunity Solution Tree mapping the desired outcome → opportunities → solutions → assumption tests. (3) For every solution, list the riskiest assumption and design the smallest possible test (a Figma prototype, a fake door, a manual concierge service) that could falsify it. (4) Review the tree weekly with the trio and prune ruthlessly. (5) Run discovery and delivery in parallel — never block delivery while discovering, never skip discovery to feed delivery.

Formula

Discovery Velocity = Assumptions Tested per Week × % Tests that Changed a Decision

In Practice

When Teresa Torres started coaching Hope (a product team at a B2B SaaS company), they were shipping one major feature per quarter and seeing <10% adoption. Torres made them commit to one customer interview per week and one opportunity-tree update per week. Within 6 months, they were shipping smaller features more frequently, but adoption per feature jumped to 35%+. The team killed two roadmap items mid-sprint because interviews revealed the underlying opportunity didn't exist. (Source: Continuous Discovery Habits, Teresa Torres)

Pro Tips

  • 01

    Marty Cagan's rule: 'The product manager's job is to discover a product that is valuable, usable, feasible, and viable.' If you're only validating ONE of those four risks, you're not doing discovery — you're doing a partial sanity check.

  • 02

    Run discovery on next quarter's bets while delivery ships this quarter's bets. The most common failure mode is sequential ('first we discover, then we build'), which produces feast-and-famine cycles.

  • 03

    If your interviews keep producing the same insight, you're either talking to the same persona or asking the same questions. Rotate both. Torres recommends touching 3 different segments in any 4-week window.

Myth vs Reality

Myth

Discovery is a UX research function

Reality

Discovery is a triad function: PM owns value/viability, designer owns usability, engineer owns feasibility. Outsourcing discovery to a research team produces shelfware reports nobody acts on. Cagan's data: teams where engineers attend customer interviews ship features adopted 2x more often than teams where engineers only see the spec.

Myth

We don't have time for discovery — we need to ship

Reality

Teams that skip discovery don't ship faster; they ship more code that gets thrown away. Pendo's 2019 study found 80% of features in B2B SaaS are rarely or never used. The 'shipping' was illusory — most of the work was waste. Discovery is how you stop building the wrong things at full speed.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Scenario Challenge

You're a PM at a 30-person SaaS company. Your CEO has approved a 12-week build of a new analytics dashboard based on three sales calls where prospects asked for 'better reporting.' The engineering team starts on Monday. You have one week.

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Customer Interviews per PM per Week

B2B SaaS product trios

Continuous Discovery (Torres standard)

1+ per week

Healthy

2-4 per month

Sporadic

1-2 per quarter

Feature Factory

0 — only at kickoff

Source: Product Talk / Continuous Discovery Habits research

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

🧭

SVPG (Marty Cagan)

2008-present

success

Marty Cagan codified product discovery in Inspired (2008, revised 2017) based on observing how Google, Apple, Amazon, and Netflix actually built products. His core finding: the best product teams don't get a roadmap of features from the business; they get a set of business problems and the autonomy to discover the right solution. Cagan estimates that for every 10 ideas tested in discovery, only 1-2 actually move the metric they were designed to move — meaning teams that skip discovery are wrong 80% of the time.

Ideas Tested → Ideas That Work

~10:1 to 5:1

Top Teams' Discovery:Delivery Split

~50:50 of PM time

Inspired Copies Sold

200K+

Common SVPG Coaching Lift

2-3x feature adoption

If 8 out of 10 untested ideas don't work, then a team shipping every idea on its roadmap is producing 80% waste. Discovery is the cheap filter that prevents the expensive build.

Source ↗
🌳

Teresa Torres / Product Talk

2017-present

success

Teresa Torres' Continuous Discovery Habits (2021) formalized the weekly cadence: every product trio runs at least one customer interview per week, mapped to a specific desired outcome on an Opportunity Solution Tree. The framework shifted discovery from a quarterly research project to an always-on team habit. Teams Torres coaches typically take 6-12 weeks to install the cadence; once installed, they report 30-50% reductions in features that fail post-launch.

Minimum Cadence

1 interview/week per trio

Common Adoption Lift

30-50%

Time to Habit

6-12 weeks

Trio Definition

PM + Designer + Engineer

Discovery isn't a project — it's a weekly habit. The teams that win install a cadence so reliable that 'we haven't talked to customers this week' feels as wrong as 'we haven't deployed this week.'

Source ↗

Decision scenario

The Roadmap You Inherited

You're a new Head of Product at a Series B SaaS. The team has a 9-feature roadmap committed to the board for the next 2 quarters. Engineering capacity is fully booked. The roadmap was built from sales asks 4 months ago, with no customer interviews. The CEO expects you to 'execute, not re-plan.'

Roadmap Items Committed

9 features / 2 quarters

Customer Interviews in Past Quarter

0

Avg Feature Adoption (last year)

12%

Engineering Capacity

100% allocated

01

Decision 1

You have 4 weeks before the next board update. The team wants direction. You can either start delivering on the inherited roadmap or carve out time for discovery on the items still 8+ weeks from build.

Execute the roadmap as committed — re-planning so early would damage credibility with the board and the teamReveal
You ship 7 of 9 features over 6 months. Adoption averages 14% — slightly above last year. Two features are quietly retired within a quarter. The CEO is satisfied with delivery but quarterly revenue from new features comes in below model. The board's enthusiasm fades and you're now seen as 'delivering what was promised but not moving the needle.'
Features Shipped: 7 of 9Avg Adoption: 12% → 14%Roadmap Credibility: Maintained but unimpressive
Keep delivery moving on the next 2-3 items, but install a weekly discovery cadence on items 4-9 — re-scope or kill any that fail discoveryReveal
By week 6, discovery has killed 2 of the 9 items (no real demand) and re-scoped 2 more (the underlying job was different). You free up ~10 weeks of engineering capacity for two new bets surfaced from interviews. Six months in, you've shipped 7 features (5 from original roadmap, 2 new) but average adoption is 38%. Revenue from new features beats plan and the board upgrades your mandate.
Features Killed via Discovery: 0 → 2Features Re-scoped: 0 → 2Avg Adoption: 12% → 38%Eng Capacity Reclaimed: ~10 weeks

Related concepts

Keep connecting.

The concepts that orbit this one — each one sharpens the others.

Beyond the concept

Turn Product Discovery Process into a live operating decision.

Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.

Typical response time: 24h · No retainer required

Turn Product Discovery Process into a live operating decision.

Use Product Discovery Process as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.