Continuous discovery: talking to users every week

Teresa Torres's continuous discovery is the habit of talking to customers every week, done by the product team. How to start and the opportunity solution tree.

info Quick Definition
Continuous discovery is a practice from Teresa Torres: at a minimum, the team that builds the product has weekly contact with customers. That contact is not meant to validate an idea already decided, but to uncover opportunities: needs, pain points and desires that guide what is worth building.

What is it and what is it not?

Torres defines continuous discovery, in her book Continuous Discovery Habits (2021) and in her talks, with a simple sentence: “At a minimum, weekly touch points with customers”, by the team building the product. The rhythm is what matters. Discovery is not a phase at the start of a project that ends before building begins. It is a habit that does not stop while the product is being developed.

According to the transcript of her 2021 talk for Business of Software, the activities are small and aimed at a desired outcome. The first of them is usually interviewing to understand the opportunity space. It is not a mass survey or a months-long study: it is a conversation with one person, every week.

Who does it?

Torres insists that the responsibility belongs to the team, not to a separate research department or only to whoever leads the product. In her Y Oslo 2024 talk she puts it this way: the team that builds the product has the contact. In her approach, that means the product trio: whoever manages the product, whoever designs and whoever builds. When contact with customers is delegated entirely to one person, the rest of the team receives conclusions and loses the conversation that produced them.

This idea connects with dual-track agile: discovery is a workflow of its own, feeding the building team continuously.

Why are interviews the central habit?

According to Zeda’s summary, Torres calls weekly interviews a keystone habit, because they build the motivation for the habits that follow. They serve to uncover opportunities, not only to validate solutions. A team that talks to customers every week starts to see patterns that do not appear in a large annual study.

Torres also advises not over-optimizing the first interview. In her Y Oslo 2024 talk she recommends not fixating at first on the process or on participant selection: “Just find a customer to talk to”, and repeat it the following week. What matters is starting: “Something is better than nothing”.

What is the opportunity solution tree?

The opportunity solution tree is the central visual tool of the framework. It starts at the root with a desired outcome. From there, opportunities branch out: needs, pain points and desires that customers have mentioned. And from each opportunity come candidate solutions, which are validated with assumption tests.

Product Talk lists a clearly defined outcome as a prerequisite for building the tree: the root is what gives everything else its meaning. That is why the tree is related to the difference between outcomes and outputs: without a clear outcome, opportunities have nothing to be measured against.

The solutions in the tree are tested with assumption tests, which, according to Product Talk, can assess desirability, viability, feasibility, usability and ethics. The tree does not decide by itself: it organizes the work so each solution has a reason and a way of being checked.

How much time does it take?

Less than people assume. At Y Oslo 2024, Torres suggests that a team start with one small habit, for example a single weekly conversation of 20 to 30 minutes, and let it compound over the weeks. In her 2021 Business of Software talk she adds a practical tip: it is easier for someone to say yes to 15 or 20 minutes than to 30.

Torres’s warning is clear: do not wait for a perfect process, because an imperfect one is better than none.

Teams often try to do it perfectly, and then they do nothing. A modest, steady rhythm is worth more than an ambitious process that is abandoned after a month.

Example

Illustrative example: a medical appointment booking app team defines the outcome as “people book their first appointment without calling the clinic”. Every week, one person from the team talks to a recent patient. The conversations reveal that many people fear choosing the wrong specialty. The team records that opportunity in the tree and proposes three solutions: an assistant that suggests a specialty, a symptom list or an orientation call. Before building any of them, it tests the cheapest with five people. In the first week of practice, the team interviewed only two people. That was enough to change the priority.

For your team: choose an outcome, schedule a conversation with a customer this week, and note three needs you hear. That list is the first level of the tree.

When should you NOT rely on it?

  • When the team has no access to customers. Without real conversation, discovery becomes assumption in a nice format.
  • When the decision is already made for a reason that does not depend on customers, such as a regulatory obligation.
  • When the team expects immediate results. Continuous discovery pays off over time, not in one week.
  • When AI is used to replace people. At Y Oslo 2024, Torres says she uses AI for tasks such as sentiment analysis or checking her own synthesis, but that, for now, we should keep talking to humans. She also removes all identifying information about customers before feeding interview data into AI tools.

Common mistakes

  • Treating discovery as a phase that closes before building.
  • Delegating all the conversations to one person.
  • Starting from a solution and looking afterwards for a problem to justify it.
  • Trying to do it perfectly and not doing anything.
  • Talking to customers once a quarter and calling it continuous.
  • Not connecting interviews to an outcome and losing the thread of the tree.

For the logic of small experiments that test assumptions, read Lean UX. To place discovery within an agile team’s process, review UX in agile teams. And to understand how discovery is organized in phases, the double diamond offers a map.

Sources

Frequently asked questions

What is continuous discovery?
It is the practice, described by Teresa Torres in her book Continuous Discovery Habits (2021), of having at least weekly contact with customers by the team that builds the product, with small research activities aimed at a desired outcome.
What is the minimum habit of continuous discovery?
According to Torres, ‘at a minimum, weekly touch points with customers’, by the team building the product. That means talking to customers every week, even briefly. Weekly interviews are the central habit, because they uncover needs, pain points and desires.
What is the opportunity solution tree?
It is Torres’s visual structure that starts from a desired outcome and branches into customer opportunities (needs, pain points and desires) and candidate solutions, which are then validated with assumption tests.
How much time does continuous discovery take?
Torres suggests starting with one small habit: a single weekly conversation of 20 to 30 minutes with a customer. With practice, that rhythm compounds and becomes part of the team’s routine.