How to lead product “discovery”

danramsden
9 min readSep 15, 2023

--

Summary: Defining discovery as a structured process to increase confidence can help you decide how to direct discovery projects. And adopting a double-loop learning process during these projects is more likely to lead to genuine discoveries.

Last month I shared a critique of some ways people describe and structure design processes. I also shared a tool I’ve developed to direct “discovery.” I wanted to share more about that tool and talk about why I think confidence is a good measure of success for your “discovery” processes. Creative leadership is about deciding and directing what to do next. That can be difficult in “discovery” and sticking to a formulaic script can introduce inefficiency, despite the false confidence it might provide. Defining discovery as a structured process to increase confidence can help you decide how to direct discovery projects.

Good to great

In Good to Great, Jim Collins introduced the idea of a reinforcing loop being the engine of effective organisations. Reinforcing loops exploit a set of aligned factors where outputs feed inputs. Collins used this idea to talk about how breakthrough organisations don’t achieve greatness through a single act of inspirational disruption, but through careful alignment of activities. The alignment acts like the counterweights on a flywheel. Once a flywheel is moving, the same amount of input energy creates increasingly fast rotations. Organisations can use strategic alignment and exploit network effects or economies of technology or scale to create a reinforcing loop to achieve their objectives more easily.

For Amazon, their flywheel is built around excellent customer experience. This attracts customers (traffic), and the vast potential customer base attracts sellers. More sellers expand the range of available products offered to customers. The comprehensive inventory and high level of service creates a reinforcing loop. With economies of scale, Amazon can reduce costs, which in turn can be passed on to consumers which adds to customer satisfaction and continues to drive the flywheel. All this alignment leads to Amazon’s goal — growth.

Service design can train us to be on the lookout for loops and system dynamics. But too often, ‘discovery’ is framed as a linear process to validate a hunch or find a “problem” for a solution that someone has already fallen in love with. It can also resemble a waterfall process, whereby we discover something and then exploit it — like explorers did in the past. Implementing and testing is obviously necessary. But viewing discovery as a “snapshot and act” or “freeze and unfreeze” approach to change is far less effective than directing it as a continuous process of building confidence. Most change is about momentum and most competitive advantage is about incremental improvement and non-replicable integration. Even if you’re directing discontinuous innovation, considering the accumulation of insight and confidence as the real objective of discovery can help you as you direct teams and collaboration.

Increasing confidence through design

Design is a dialogue between a situation and imagined (and then actual) interventions. We spend time understanding a context and use our imagination to dream up outcomes we’d prefer and factors that could move us closer to our desired situation. It’s not linear, but a series of progressively concrete comparisons between the problems you identify and the solutions you try to create. Discovery is like the back and forth of a good conversation. There’s a repetitive echo as we make progress from problems towards solutions.

But if you don’t just move left to right and broader and narrower during discovery, as depicted in the double diamond, how do you direct teams and resolve complexity? I think you decide by answering questions… My ‘design pyramid’ describes mindsets and skillsets which teams can adopt to orient themselves and navigate through discovery projects. At the highest level it describes two forms of “making”… sense-making is about observation, measurement and synthesis and it transforms data into information, knowledge or wisdom. Difference-making intervenes and disrupts a situation to change it. These two thinking styles answer two types of questions — sense-making gives us an understanding of the situation. Difference-making identifies and invents alternatives to answer “how could things be?”

Below this level of the model I describe four broad activities: identify, create, experiment and evaluate. Each of these includes more granular activities or areas of focus. I carefully arranged the labels onto the three-dimensional shape, so the arrangement helps teams intuitively understand relationships between activities. Moving between the modes helps to “balance” activities and provide momentum. The three-dimensional arrangement helps you identify what to do next.

I’m sticking with the three-dimensional shape, it has lots of advantages. But I could have described my approach as a flywheel. Great discovery projects begin by identifying or articulating a problem or opportunity, they create imaginative responses and then through experimentation and evaluation test and optimise for positive outcomes. In an ideal world the evaluative efforts within discovery help us identify further opportunities (or gaps in our understanding) and we start the process again, all the time increasing our confidence that our activities are generating more momentum towards our organisational goal.

Everyone has got a plan until they’re punched in the face…

Scientific management suggests that efficiency can be maximised by following the perfectly optimised linear process. But this only works when you have high levels of stability and predictability of method and outcome — you know how to do something and what it will look like when you’re finished. Innovation, in fact most strategy and design involve higher levels of volatility and unpredictability. Planning a linear discovery process ignores this.

We can’t determine the “perfect process” for a discovery project at the beginning because we don’t know what discoveries we might make. Imagine sticking to a linear plan for an expedition when you encounter the unexpected — your potential for death increases… we need to cross a river… we need to climb this rock face…. we need to evade this bear… as we discover more, our mental picture of the idealised path will change — we can’t anticipate all the opportunities and barriers we’ll discover. In his great book on planning, Peter Morville uses the example of how the most skilled players of Tetris actively rotate the shape while they mentally explore different arrangement and placements, rather than planning and then rotating. Discovery requires the same type of mental agility. The combination of action, reflection and planning creates better outcomes.

Discovery is the perfect practical example of the need for reflection-in-action and the value of double-loop learning. Chris Argyris described the concept of double-loop learning: “an individual, organisation or entity is able, having attempted to achieve a goal on different occasions, modify the goal in the light of experience or possibly even reject the goal”. Single-loop learning is repeatedly attempting to solve the same problem, with no variation of method and without ever questioning the goal. Applying linear planning and process to discovery encourages single-loop learning — you might discover a solution, but it’s more likely to be a solution to the wrong problem. A double-loop approach is more likely to lead to genuine discoveries. It turns out that discovery is more about “double loop” than “double diamond”.

A double-loop learning approach incorporating reflection-in-action updates your mental model. It encourages real-time re-evaluation and (re)framing. You learn through doing, which is exactly what discovery is.

Discovery combines practice and process and is about progress, not perfection

Practice: Planning and strategy define actions to move towards a goal. But discovery is about identifying the goal, as much as it is about moving towards it. The usual tricks of planning don’t work. But, if we approach discovery like other agile practices, in short sprints, the outcomes of which are increasing levels of confidence then we get the efficiencies of a process with the benefits of a practice-led, learning approach.

The early stages of this type of approach can feel like the opposite to progress, as all you’re uncovering are questions, not answers. But this is because you’re discovering the unknown unknowns that have the potential to disrupt your activities when you try to act at pace or at scale… remember this and you’ll realise that even if answers take time to come, you’re still always increasing confidence with well-led discovery. Discovery can feel like a jerky, zigzagging activity with lots of “pivoting” and changes of direction. Reframing discovery as a learning activity to increase confidence (and ultimately pace and scale) makes this potentially dispiriting experience reassuring.

Process: My pyramid and the process around it encourages teams to focus on one or a small number of unanswered questions and find the cheapest reliable method to generate an answer to increase confidence. It also — most importantly — normalises asking questions.

Increasing confidence will usually follow the same loop — recognise a question you can’t answer, create or locate data or possible answers, experiment by testing these against the hypothesis, and then extend the consequences of what you’ve learned to the next biggest context — giving you the ability to re(frame) or update your mental model to exploit a big opportunity or address a more significant problem. Begin again. Directing discovery now becomes about identifying the most important assumptions and questions. As a rule of thumb, the most important questions are usually sense-making questions, and we use “difference making” to generate answers where the data doesn’t already exist. We can prioritise questions by working through processes like the “five whys” to identify dependencies and pursue multiple “cheap” methods for generating answers to make progress.

Progress: Thinking of discovery as the identification, prioritisation and answering of questions creates the same sort of feeling of inevitability that we see in a good flywheel –one activity feeds the next and increases momentum. If our goal through discovery is to increase our confidence, then unanswered questions are friction in the system. When a team is struggling to know what to do next, the answer is: audit assumptions and unanswered questions and then set about answering them. Getting an answer will usually mean stepping through the loop to answer the question in front of you…

For each of the “activities” on the discovery pyramid you can ask questions:

  • Are we clear on the objective — not just the objectives of the discovery process, but also the objective(s) of the organisation. Do we know why we are doing this and how we will measure success?
  • Have we consulted the most useful data?
  • Do we need to generate more data through formative research?
  • Have we had enough ideas?
  • Have we iterated on those ideas enough?
  • Can we describe the ideas in a way that will enable us to make decisions about them?
  • How will we test whether an idea achieves its intended purpose?
  • What comparable ideas could we use to better understand our own?
  • Why would we do this — does it fit our strategy and integrate within our organisation?
  • What harm might we do?
  • What technologies would we rely on?
  • Does this scale?
  • Are we ready? What might stop us?
  • How will we measure the impact?

These are just a selection of the questions that the pyramid prompts. They provide a checklist that can help direct a discovery project — if you can’t answer a question, you must make a choice about whether the answer is worth generating and what techniques you could use the get that answer.

Usually, finding answers and making progress in discovery is about moving between modes and generating momentum by resolving ambiguity… for example, the evaluative question of “what harm might we do?” might prompt a team to either return to identifying useful data to model impact, create an experiment to test assumptions or create and iterate to mitigate harms. A skilled leader will look for the cheapest reliable answer, and they won’t discount the option of pursuing multiple answers by combining the efforts of different disciplines across the modes. In my discovery toolkit I’ve created a number of these prompts and “plays” to pose questions and propose strategies for finding answers.

If you’re interested in learning more about the discovery pyramid toolkit which includes templates, canvas, playbook and facilitators guide please express your interest here.

--

--

danramsden

I'm a Creative director at the BBC. I like words, design, data and magic. These are all my own views (apart from retweets. I borrowed those to look clever.)