The Product Discovery Process

So, let’s understand what product discovery is and how to do it well.

What is Product Discovery?

Product discovery is the process of understanding your users’ pains and coming up with viable solutions to cover them.

This process is usually very different from how your stakeholders imagine it. For many of them, it’s about creative thinking and coming up with a list of cool feature ideas once, then building the product based on that list.

This thinking is wrong in 2 key areas:

• Discovery is a never-ending and iterative process. You don’t finish discovery, you finish a cycle of it and start the next one.
• Discovery is not about creatively coming up with cool ideas. It’s about understanding user needs, bringing solutions that can fix them, then validating that these solutions actually work.

Another common misconception is about the relationship between discovery and delivery. Many stakeholders see them as two separate phases that come one after another. In reality, effective product development teams do discovery and delivery in parallel. While your team is delivering the features you discovered 2 months ago, you’re running a discovery for the next set of features.

Visual:

• “Discovery vs. Delivery” split-path diagram. Visualize this idea: teams do discovery and delivery in parallel. While your team is delivering the features you discovered 2 months ago, you’re running a discovery for the next set of features.
• Alt: “Parallel tracks showing continuous discovery and delivery.”
• Caption: “Product discovery isn’t a one-time event — it runs alongside delivery.”

Even if you’re in a market where waterfall development is the norm and discovery comes before delivery, it’s always important to pay all of your attention to discovery, as its quality will directly affect product success.

Need expert help selecting the right tool?

With one-on-one help, we guide you to your top software options. Narrow down your software search & make a confident choice.

Get free tool advice

Why Discovery Is a Competitive Advantage

The short answer would be like this. With discovery, you understand your personas and user needs better and come up with effective solutions. Having effective solutions makes you a strong competitor in the market as you quickly acquire and easily retain customers.

For the long answer, let’s look at Marty Cagan’s Inspired. In the book, Marty points out that effective discovery can help you mitigate 4 types of risks:

• Value Risk refers to the chances of people not using your product because it's meaningless to them.
• Usability Risk refers to people struggling to use your product.
• Feasibility Risk refers to your team being unable to build it.
• Viability Risk refers to the product helping your business survive and thrive.

Assuming that you can mitigate all of these risks and your competitors fail at one or more of them, you are getting a strong competitive advantage and the chance to take a bigger piece of the pie.

Tiny Speck is a great demonstration of this idea. It was a game development company that built an MMORPG online game that failed miserably. The problem - they built a game by simply copying ideas from others (a.k.a. Factory Trap) without clearly understanding the gamer base's needs.

The company soon killed this game and released its internal communication tool instead. Yup, that was Slack.

So, the same company is also a great demonstration of product success using rigorous discovery. Unlike their game, the Slack team was running endless iterations of discovery and had a very clear understanding of what their user' needs were.

4 Key Questions of Product Discovery

If we want to understand the essence of product discovery without getting lost in the details, we can look at these 4 questions that it tries to answer.

• Is there a real problem? That’s right, sometimes you risk creating a solution in search of a problem. Hello 90% of all ChatGPT-powered startups out there ;)
• Can we solve it? Sometimes the problem is in the complexity of the solution. S.T.A.L.K.E.R. is a Ukrainian video game series that faced this problem. They wanted to build A-life, their AI system that fully simulated the lives of NPCs and animals in the game. But it was just too complex a problem to solve.
• Should we solve it? Sometimes the problem is there, but it’s not big enough to solve with a product. Google Glass is a great example of it. Yes, it was cool, but having no Google Glass wouldn’t make life less convenient for anyone.
• Can we deliver it effectively? This question is more about the operational side of product development and your ability to scale and support your users. YouTube, during its early days, was very close to failing because of its inability to scale.

What you essentially do during product discovery is use different tools and frameworks to obtain insights, letting you answer these questions.

Frameworks to Structure Discovery

One could argue that a seasoned product manager can do discovery without using any playbook or framework. That’s right. But there’s a reason we have discovery frameworks. They let you bring structure and predictability into your discovery process.

In terms of the frameworks themselves, let’s look at these 5:

Dual-Track Agile: This is the process of having two parallel processes in Agile development. You have the discovery track that creates feature ideas and the delivery track that designs and builds them.

ProductBoard and Jira are usually the tools I use to manage this framework.

Opportunity Solution Tree: We will talk a bit more about this later on. But, in short, this framework lets you create a tree where the root is the core outcome. Out of this root, you create opportunities to solve pains as branches. Then, for each opportunity, you create branches of possible solutions.

Miro has great opportunity solution tree templates that you can use.

Discovery Sprint: This framework is inspired by Google’s brainchild, the design sprint. It allows rapid iterations on solutions. You get around 2 weeks to run user interviews, come up with a solution, build the prototype, and test it.

Continuous Discovery: Unlike the previous three, where you have iterations with a clear start and end, continuous discovery has no end. You simply run endless rounds of interviews with your target audience, continuously create solutions, and test them.

Here’s a side-by-side comparison of these frameworks.

The selection of the framework itself will depend on you. Not every team will benefit equally from the same framework. Use the table above to choose one that aligns with your team’s size, pace, and tolerance for uncertainty.

Unlock the Product Vault

We’ve collected the goods — AI prompts, exclusive deals, and a library of resources for product leaders. Unlock your account for access.

Phone

This field is for validation purposes and should be left unchanged.

Name*

First name Last name

Business email*

This field is hidden when viewing the form

Free Account Module

Bottom BarLanding PageFooterInlinePop UpSidebarSSO

LinkedIn profile

Seniority*

Select an optionExecutive / C-suiteVPDirectorManagerIndividual ContributorOther

Company's AI adoption maturity*

Select an optionExploring opportunitiesRunning pilotsScaling adoptionEmbedded in the business model

Roles & Team Collaboration

There’s a common misconception (and a huge pitfall) that product discovery is only about product teams.  Although it is mainly an activity for product teams, a great discovery process will involve other teammates and stakeholders, too. Here is how each team member will contribute to the discovery:

• Product Managers: They lead the discovery process, align teammates, and prioritize solutions and insights.
• User Experience Design: They take part in user research, create design concepts, clickable prototypes, and run usability testing.
• Development: They assess the feasibility of solutions and point out product delivery risks.
• Stakeholders: They give the high-level strategic and business context.

As we can see, the participation of each teammate is important. So, if your leadership asks you if the scrum team should participate in the product discovery process, your answer is a definite yes!

The Product Discovery Process

Understanding the needs of your customers and coming up with validated solutions for them reminds us of a funnel where the top layers are much larger than the bottom ones.

This is indeed the case in product discovery. You start with a bunch of insights, complaints, and comments from your target audience and end up with only a couple of validated solutions that you know will create value for your users.

Here’s what it looks like.

Now, let’s break down the process of product discovery a bit more and understand the intricacies of each of its elements.

Generally, the product discovery process consists of these 10 steps.

Now let’s look at each step in detail.

1. Review Your Strategic Direction

Good features cover user pains. Great features do the same while following your product’s greater strategic direction. Before you begin product discovery, it is important to revisit your strategy document and remind yourself of where you want to reach in the long run and what direction you want to take.

This will help you focus your discovery efforts on solutions that are aligned with your strategy. Otherwise, you risk focusing on features that will get you nowhere in the long run. 

2. List Your Assumptions

Another important aspect of discovery that you need to handle before communicating with users is your assumptions. The idea is that you’re most likely making a lot of assumptions about your market or users, and it’s important to document them to make sure that your research is not relying too much on the more bold assumptions of yours.

For this, you can build an assumption map.

On the map here, you can rely on the assumption on the top left side and be wary of the assumptions on the top right.

3. Talk To Your Users

Without this step, product discovery is just guesswork. Talk to your users—or risk building the wrong thing entirely.

There’s a wide variety of ways you can talk to users:

• Visiting them (e.g., going to a hospital and interviewing doctors).
• Meeting them at trade shows (e.g., talking to tech reporters at CES).
• Hopping on Zoom calls with them.

For the last one, you can use various methods such as research participation platforms, reaching out to your user base, or reaching out to your target audience on LinkedIn using its recruiter account.

4. Analyze Insights From Users and Data

Beyond user interview summaries, check your product analytics dashboard and the feedback your support and sales teams have gathered. That way, you’re pulling insights from multiple sources. And rather than reading through endless summaries or transcripts, you can have an LLM process it all for you.

Here’s a sample prompt you can use:

You are a senior software product manager with 10 years of experience. You are great at analyzing the transcripts from your sales team and identifying key information that will be useful for your product research.
Here is the sales call transcript
{{sales_transcript}} 
Based on the information in the sales transcript, please identify the following information.
Please identify the following:
1. Top pain points: refer to the most prominent pains and challenges that the user is experiencing with current tools and processes.
2. Why are they interested in you: refers to the reason the company is looking for a new tool now.
3. Product feedback: refers to the feedback that the company has given to the sales team about {{your_product_name}} and its features when they have watched the demo.

After looking at the feedback and pairing them with user interview insights and analytics, you will start seeing trends and common themes appearing.

5. Identify User Pains Within Your Findings

Identifying pains is not always that straightforward. Yes, there are cases when 80% of all users told you that a certain part of their work is a pain, but it doesn’t necessarily mean that it’s a problem worth solving. You should also understand the “intensity” of that pain.

Imagine that 70% of your social media platform users say that they are annoyed by the number of notifications and the fact that you cannot mark all of them as seen.

When you pair their feedback with analytics - showing that 90% of them have 2-3 notifications a day on average, you might gauge that this complaint is not "intense" enough to prioritize compared to other items on your list.

6. Come Up With Solutions For These Problems

This step is when you come up with potential solutions to the problems you have identified as a result of your customer interviews and feedback analysis.

You can just sit down with your design and development teams and brainstorm. But, to make sure that your solutions are directly tied to a larger business goal or outcome, you can use the Opportunity Solution Tree framework by Teresa Torres. Here’s what it looks like for the Spotify users’ pain of “I don’t know what to listen to”.

Through the tree framework, we saw that removing the “what to listen to” barrier could boost daily active users (DAU), making it a high-impact problem to solve.

7. Test Your Solutions With Prototypes

Analyzing user insights and turning them into a list of potential features is just the starting point. Product discovery is grounded in the scientific method, which means you should test and validate ideas before adding them to your product roadmap.

You can validate solutions at different stages of their lifecycle — from simply describing the idea to a user during an interview and gathering feedback, to generating basic mockups, to asking focus groups and gathering user feedback on early prototypes. You could go so far as to build and launch a minimum viable product (MVP).

Early-stage validation, such as discussing the idea in an interview, is fast and inexpensive, but the quality of customer feedback can be low since users may struggle to fully understand the concept.

In contrast, testing with an MVP provides richer, more accurate feedback because users can interact with the actual functionalities of the product. However, this approach is slower and more costly due to the development work involved.

Considering this tradeoff, my favorite validation method is clickable prototypes, as they are both relatively cheap to make and show real users an actual UX design of your new product or feature.

To speed up prototype creation, you can use Figma AI or the built-in AI features of other UX design tools.

8. Document Test Results

Once your test runs are complete, compile all your learnings and results into a single document. Group them by the solution and the hypothesis being tested.

This way, when it’s time to evaluate a solution, you’ll have everything you need in one place to make an informed decision.

9. Validate or Invalidate Your Solutions

This is the point where you look at all the findings from your tests and decide if:

• You consider the solution validated and add it to your existing product backlog.
• You understand that the solution is on the right path but needs a rework. Then you take it back to the ideation phase to fix based on test results.
• You invalidate the solution as it does not solve the user problems it was supposed to solve.

All three outcomes are expected. When you invalidate a product, you should consider it a huge success, as you just avoided building products and wasting time and resources for something that does not solve your users’ pains.

10. Build The Validated Solutions And Start Over

The final logical step of the product discovery process is adding your solutions to your backlog, prioritizing, and building them.

However, it’s really important to understand that you don’t really stop product discovery here. Instead, you wrap up this iteration and immediately start the next one. Successful products are those where the discovery process never ends.

Here, teams are able to make the right product decisions based on their learnings from user feedback, usability, and A/B testing.

Tools for Product Discovery

Here are some tools worth exploring to help you move through product discovery faster and get better results, organized by the phase you’re in.

• User Research: Hotjar for session recordings, UserTesting for usability tests, and Figma for user experience design and clickable prototypes.
• Journey Mapping and Visual Collaboration: Miro is the universal tool here. You can also consider trying its emerging competitor - FigJam or other Miro alternatives.
• Competitor Research: Similarweb for competitor traffic channels, Builtwith for tech stacks.
• Product Analytics: GA4 for channels and traffic. Amplitude for behavior and key product metrics. For heatmap software, HotJar is a popular option.
• Tracking and Prioritization: RICE scoring model for lightweight priority-setting, Aha! or ProdPad for roadmapping and product idea storage.

Those are the tools I use and recommend to my peers. For more options, you can also take a look at our curated list of product discovery tools.

Also, if you have noticed, I did not mention any AI solutions here. That’s because I want to talk about it separately and in more detail.

How AI Is Transforming Discovery

Discovery has probably been the product management area where AI has added the most value. The reason is that LLMs are great at the types of tasks you would commonly find in product discovery. Specifically, transcription, summarization, and extracting insights.

We covered the impact of AI on product discovery in one of the episodes of our podcast. Our guest, Craig Watson, walked us through his own experience and shared a lot of valuable insights.

Specifically, he points out that AI has been a great help to product managers in the following areas:

Transcribing and summarizing user interviews: PMs don’t have to take notes during the call and listen to the recordings afterward. All they need is the AI bot to be present in the call and record it. The bot will then turn the recording into a transcription and summary - highlighting the most important findings from that interview. Great tools for this are Dovetail, Krisp, and GreatQuestion.

Detecting repeatable patterns in the qualitative data: It’s very time-consuming to listen to all interviews and find pains or processes that repeat. LLMs can handle this task very well and go through hundreds of interviews in search of patterns. Dovetail is the best at this, from my experience.

Feedback clustering, counting, and scoring: Again, it will take hours or even days for product teams to manually convert qualitative data into quantitative data. LLMs need mere seconds to count and see which feedback is the most common in the data. They can also do scoring using the weighted scoring model or RICE.

Despite all of the value you get from these automations, Craig still recommends not relying too much on AI when performing discovery.

“AI won’t replace your intuition — it’ll help you make faster, evidence-backed calls.”

I agree with him. Let the AI do the manual work for you. But don’t trust the decision-making of the AI without checking it yourself first.