What continuous discovery really is
Many teams confuse initial, project-based discovery with a continuous practice. They run a discovery phase, define the problem, scope the work, and then shift entirely into delivery mode. Teresa’s framework flips this on its head.
“My continuous discovery framework was designed for outcome-focused teams. They're not starting with a solution. They're not even starting with a customer problem. They're actually starting with an outcome.” – Teresa Torres
This is a critical distinction. Instead of beginning with a customer pain point, you start with a desired business outcome, such as improving a key metric or hitting a strategic objective. This outcome acts as a filter. Customer needs are infinite, but the outcome tells your team which needs to focus on. The goal is to discover the customer needs that, if addressed, will help you reach your target outcome.
The prerequisites for success
For continuous discovery to take root, a few things need to be in place.
-
Empowered teams, not autonomous teams: Leadership must delegate authority, but within clear guardrails. Teams aren't given a blank check to do whatever they want. Leaders must define the strategic context—mission, vision, revenue model—so teams can make decisions that align with the broader business. “A lot of organizations have an implicit strategic context,” Teresa notes, “which leads to frustration when teams suggest things that leaders would never consider.”
-
Direct customer access: The people building the product need to talk to the people using it. In many companies, Sales or Customer Success act as gatekeepers. This barrier has to come down.
-
Good analytics: While not a dealbreaker (nearly half of product teams have no analytics), good product instrumentation becomes crucial as your discovery efforts mature.
Why discovery is so hard?
Even with the right intentions, both startups and large enterprises struggle to implement continuous discovery.
The startup challenge
Surprisingly, Teresa believes startups are often the toughest environment for discovery. Why?
-
Founder vision: Most founders start with a strong solution in mind, not a problem to solve. It takes experience—and often getting "knocked down a few times"—to become receptive to feedback that challenges their "baby."
-
The race for revenue: The pressure to survive makes it incredibly difficult to say "no" to a potential customer, even if their request pulls the product in a direction that doesn't serve the broader market.
The enterprise challenge
In larger companies, the culture is overwhelmingly solution-focused. Roadmaps are built around features, meetings are about prioritizing ideas, and success is measured by output. Teresa offers a brilliant analogy:
“In an organization, big or small, ideas are like ice cream and discovery is like broccoli. It's an unfair fight. The vast majority of people are not gonna pick broccoli over ice cream. But just like with health, we get better outcomes if we eat broccoli.”
So, if you’re an individual contributor on a team stuck in an "ice cream" culture, what can you do? Don't try to change the organization.
“The biggest mistake people make is they try to change their organization,” Teresa advises. “The level where you should try to enact change is with you individually.” Start by adopting small habits. Take one tiny step. As you begin to work differently, others will get curious. That’s when you gain the opportunity to influence.
The art of the user interview
To uncover real customer needs, we have to fight against our own psychology. Humans are notoriously bad at summarizing their own behavior. This is where story-based interviewing becomes a superpower.
Instead of asking a speculative question like, “What do you like to eat for lunch?”, ask a specific, memory-based question: “Tell me about the last time you had lunch.”
The first question invites a fast, error-prone answer influenced by cognitive biases—what you ate yesterday (recency bias) or what your aspirational, healthy self wishes you ate. The second question forces the brain to recall a specific instance, providing a much more reliable picture of actual behavior.
Teresa shared a classic example of asking a woman her criteria for buying jeans.
-
Direct question: "What's your criteria?"
-
Her answer: "Fit, then price and style."
-
Story-based question: "Tell me about the last time you bought a pair of jeans."
-
Her story: "I bought them on Amazon."
The room chuckled. How do you evaluate fit on Amazon? It turned out they were on sale. Her actual criteria were 1) getting a deal, 2) convenience, and 3) fit. The story revealed the truth that the direct question obscured.
Discovery in the age of B2B and AI
Navigating B2B complexity
In B2B, you have to serve two masters: the end user who uses the product daily and the economic buyer who signs the check. How do you connect a user’s needs to the buyer’s definition of value?
The key is to ensure your product outcome is directly tied to a business outcome. You’re not trying to satisfy end users just for fun; you’re doing it because it drives renewals or attracts new customers. Your strategic context must connect the dots: we believe optimizing this user workflow will improve retention by X%.
The AI trap: prototypes and synthetic personas
The rise of AI has made it incredibly easy to build prototypes. A product leader demos an AI-powered prototype to one customer, gets positive feedback, and declares it validated. Teresa warns this is a classic case of confirmation bias.
“We don't build our products for one customer. We build our products for the market.”
Furthermore, there’s a massive difference between seeing a demo and actually using a tool. A shiny object looks great until you try to integrate it into your messy, real-world workflow.
What about using AI for discovery itself, like with synthetic personas? Teresa’s rule is simple: AI should be additive, not a replacement.
“If you are building a product used by humans, you probably need to be talking to those humans. Full stop,” she insists. Use AI-powered transcripts from sales calls (like Gong) to supplement your own interviews, not replace them. Use AI prototyping tools to test a single, specific assumption, not to validate an entire solution.
Final thoughts
As AI makes it easier and faster to build software, what you choose to build becomes the ultimate differentiator. Without a deep understanding of customer needs, we risk creating what Teresa calls a "Homer Simpson car"—a Frankenstein vehicle of tacked-on features with no coherence or real value.
The goal isn't to perfectly follow a rigid process. It's to build habits and create fast feedback loops with your customers. As Teresa says, "Most of the work in discovery is not following a process, it's managing the cycles." Start small, stay curious, and choose the broccoli. Your product will be healthier for it.
