Four Product Discovery Models: A Practical Map

All tech companies have to choose what to build, but approaches vary — from top-down projects to bottom-up research and experimentation. Your company is probably somewhere on this gamut, but it’s not always a deliberate choice. It’s important to realize that these are not the only two possible models, and that there isn’t necessarily just one right answer.  

In this article I’ll lay out the territory of product discovery using consultants’ favorite tool: the 2×2 grid.  

The Two Axes of Product Discovery

We can map product discovery along these axes: 

• Centralized vs. Empowered: Who’s driving the process? Is discovery owned by a central authority (e.g. leadership, a founder, or a select group), or are teams trusted to make local product decisions?
  
• Opinion-based vs. Evidence-guided: Are decisions made based on gut feel, consensus, and rank, or grounded in data, analysis, and experiments?
  

These axes form four quadrants, each with its own models, advantages, and challenges.

1. Centralized + Opinion-Based: Command-and-Control

In this quadrant the assumption is that some people are able to predict which products and business ideas will succeed and which will fail, which implementation is correct and which is wrong. 

The classic example is the traditional company performing what I call the planning waterfall —leaders or committees define strategies, roadmaps, and  projects, mostly based on sparse data and opinions; “delivery-teams” execute. 

But there are more modern takes on this theme. When I worked at Microsoft in the 2000s we practiced classic product management, where PMs were entrusted with roadmaps, project plans, and detailed specs. We created those mostly based on our own judgement and the opinions and feedback of our colleagues and managers. I still see this form of product management practiced widely today.

In recent years, some successful silicon valley entrepreneurs have been promoting the concept of Founder Mode (aka founder-led companies). The core belief is founders are unique, visionary people that should run their companies in a radically centralized way — essentially making or approving all product decisions. Empowerment, A/B testing, and other modern practices are considered harmful and befitting of “professional managers”. The inspiration for founder mode?  “Steve Jobs” (I’ll touch on this myth later in the article).

There’s a kernel of truth here — leaders should indeed lead and guide the company, and “democracy” (which the founder-mode people seem to fear the most)  doesn’t work for product development. Still, as I’ve written in the past, trusting all decisions to a select group of people is very risky: 

This centralized approach to decision-making may be better for consistency and alignment, and may feel less risky, but it can, and often does, lead to bottlenecks, slow decisions, high meeting overhead, and disempowered, demoralized employees. Worse, relying on the snap-judgment of busy people that may be optimistic and self-confident, but not always well informed, is hardly a recipe for great decision-making.

2. Empowered + Opinion-Based: Creative Chaos

It’s been shown time and again that rank-and-file employees, especially engineers, are able to come up with some of the best product ideas — Gmail, Twitter, and Slack are famous examples. The aim in this quadrant is to tap into this creative power by allowing individuals and teams to pursue product ideas without having to go through rigorous validation or approval processes. 

This approach was quite trendy in the past. Google is famous for instituting the 20% project policy and generally encouraging bottom-up innovation. Some projects graduated into betas and full-fledged products, but most were sunset due to lack of usage. This “throwing stuff at the wall” approach helped Google create an ecosystem of web services that attracted loyal users and greatly accelerated growth. 

But the Creative Chaos approach is not without its disadvantages. The company has to be merciless at killing ideas that don’t work  (“fail fast” was one of Google’s mantras as demonstrated by its extensive graveyard) or suffer an excess of incompatible apps and services. Creative chaos works when you have entrepreneurial, well-informed  employees, and an environment that allows bottom-up ideas to compete for funding with top-down ideas. Both are not true for most organizations. It’s also harder to manage and align this form of product discovery, and it may distract from the mission and strategy of the company. 

For these reasons creative chaos is far less popular today. When I joined Google in 2010, the senior leaders were already shifting the culture towards top-down big-bet projects. The inspiration?  “Steve Jobs” (wrong again as I’ll explain shortly). But the spirit of bottom-up intuitive product discovery still lives in Google and in many other companies through side projects, hackathons, internal startup incubators, and innovation labs. It’s a powerful product discovery pattern, but rarely the main one.  

3. Centralized + Evidence-Guided: Benevolent Dictatorships

“If we have data, let’s look at data. If all we have are opinions, let’s go with mine.” — Jim Barksdale, co-founder of Netscape

The leaders in this quadrant are very much in charge, but they acknowledge they are fallible and wish for their assertions to be tested. Evidence weighs heavier even than manager opinions. 

Judging by his words and actions, Steve Jobs believed in this form of product discovery: 

“A lot of people will tell you that I had a very strong opinion and they presented evidence to the contrary and five minutes later I completely changed my mind, because I’m like that. I don’t mind being wrong and I’ll admit that I’m wrong a lot. It doesn’t really matter to me too much, what matters to me is that we do the right thing.” — Steve Jobs (video)

These are not just empty words: Jobs initially didn’t believe in multitouch, was against  Apple developing a phone, and did not want to launch an app store, but he allowed others to convince him otherwise. All of these were bottom-up ideas that Jobs didn’t initiate, but he came to accept.  

The benevolent dictator model offers a powerful mix of human judgment and evidence-guided decision, which is why it is used at Meta, Amazon, and other bigtechs; still it’s not applicable to most companies:

• Relies on having extremely capable senior leaders that are well informed, technology and business savvy, yet humble in the face of contradicting data
• Does not scale a— the leaders can be closely involved only with a select few projects (known in Facebook as “being in the eye of Sauron”). Everything else has to be operated in a different way
• Still puts all the decisions in the hands of a few powerful people which may lead to false positives (the Metaverses, Amazon Fire phone) and false negatives.  

4. Empowered + Evidence-Guided: The Product Operating Model

This approach is considered the gold standard by many product thinkers, consultants, and practitioners. It’s heavily inspired by the scientific method, modern business management theory, psychology and sociology. 

The product operating model (a term coined by Marty Cagan) calls for leaders to  develop a strategy and define concise measurable goals. Product teams, with help of managers, translate those into tactical goals and then conduct research, evaluate and test multiple ideas, and build and launch the ideas that move the needle. There’s room for bottom-up goals and bottom up ideas. Top-down big bets exist, but they start small, and are subject to the same validation as other ideas. All decisions are made using a combination of human judgement and evidence. 

There are clear benefits to this approach. Companies that employ the product operating model can drop their bad ideas early and double-down on the ideas that work (which can truly 10x outcomes). Leaders at all levels are liberated of the burden of having to tell people exactly what to do. The quick iteration mode of build-measure-learn is much more engaging and rewarding for engineers, designers, and product managers. Leaders find they can achieve more having  value creation teams rather than feature factories. 

The challenges of this approach are also substantial. Empowerment and evidence-guided development go against long held beliefs and practices. The full product operating model requires the organization to uplevel its strategy, goals, research, dev infrastructure, and culture. In practice, many companies get stuck in some hybrid of the new and old.   

It’s Not That Simple 

I’m not a big fan of 2×2 matrices because the world rarely splits neatly into four clear-cut cases. The same is true for product discovery. Many companies are operating in a hybrid mode that spans across the grid. Take Amazon for example — teams are empowered to come up with ideas and run fast local experiments, but on bigger matters they have to go through the chain of command using the rigorous PR/FAQ review process. Most big initiatives still come top-down and managers keep a close watch on progress. 

Gen-AI paradoxically is reinforcing all beliefs. Tech bros see the AI sea change as a reason to double-down on their founder-led worldview — how else can they “act with urgency”? Some believe that vibe-coding will create another wave of Creative Chaos where anyone can make changes to the product (good luck with that). Top-down traditionalists see AI as a way to amplify the output of their feature factories. Evidence-guided development enthusiasts like myself believe that AI will help us overcome human biases and make us better at observing the world and testing our ideas.  It’ll likely take years to discover the real impact on product innovation. 

For now, I hope this model will be a useful conversation starter — where are you today on these  two axes and where do you want to end up (tip: if you do nothing you’ll likely slide towards Command and Control, that’s just how it is). As always, if you found this article useful please share with the people that ought to know and take action and with your broader networks.