10 experiments you can try to improve discovery
Want to improve your discovery? Here are 10 things you and your organisation can try to make your discovery even more valuable, based on a recent retrospective of discovery.
Experiment 1: Can’t interview? Don’t start.
Data/research suggests that
- teams can wait for weeks to carry out their first interviews with users when they have to develop a route to users. This means that the first few weeks of discovery may have limited return on investment and the final weeks are over-loaded with interviews and do not have enough time for the team to draw conclusions before discovery ends
So if we try
- starting discovery when the first interview can take place within 3 days
And measure
- time taken for first interview
- time taken to complete discovery
- % of discoveries requiring extensions
- confidence in discovery insights
We should see this change
- better insights from discovery (with ‘better’ defined as team and stakeholder confidence in these insights)
Experiment 2: Subject matter experts are invaluable
Data/research suggests that
- subject matter experts being a core member of a discovery team increases the value of the discovery (and may reduce the time taken to complete discovery)
So if we try
- increasing the involvement of subject matter experts with the core discovery team (where this is useful)
And measure
- % of discovery team ceremonies attended by stakeholders
- time take to carry out first user interview
- amount of time spent with users during discovery
- % of discoveries that require a deadline extension
We should see this change
- greater confidence in outcomes of discovery (and potentially shorter discoveries).
Experiment 3: Don’t be too digital
Data/research suggests that
- discoveries carried out by ‘Digital’ teams will tend to skew discoveries towards digital solutions (i.e. in-house software)
So if we try
- Digital ‘plus’ teams (better mix of ‘the business’/policy/analytical services, etc and ‘digital’)
And measure
- % of discoveries that lead to digital solutions
We should see this change
- greater balance of digital/non-digital solutions
Experiment 4: Lean Startup over Scrum
Data/research suggests that
- ‘pure’ Scrum is not a great framework for discovery and alpha because it expects requirements and value to be defined up-front
So if we try
- a hypothesis-driven approach as described in The Lean Startup
And measure
- % of discoveries that do not continue to alpha
- % of discoveries that pivot the problem, or user, or both
- % of discoveries that recommend pausing until conditions are right for alpha
- % of discoveries that suggest that ‘digital’ isn’t the best, or only solution space to explore in alpha
We should see this change
- data-driven decisions in discovery; teams able to share the hypotheses how they’ve tested to draw conclusions as to the value of a problem being solved.
Experiment 5: Share discovery guidance
Data/research suggest that
- dysfunction can appear during discovery due to a lack of shared understanding of the value of discovery, and the approach to discovery
- members of discovery teams may not have the same understanding
- discovery teams and their Service Manager may not have the same understanding as ‘the business’/stakeholders
So if we try
- sharing discovery guidance
And measure
- % of discovery team members, Service Managers and stakeholders that can consistently describe the value and approach to discovery
- stakeholders/’the business’ satisfaction with discovery
We should see this change
- teams encounter less dysfunction due to uncertainty around the point of discovery and the approach to discovery, and stakeholders’ expectations around discovery being met more consistently.
Experiment 6: Don’t reinvent the wheel
Data/research suggests that
- some of the core concepts of discovery and alpha (new to government) share more with project management concepts (familiar to government) than we expect
So if we try
- highlighting similarities between agile development and project management
- learning from project management
- sharing guidance on when project management is the best approach, and when working with agility is the best approach
And measure
- % of ‘work’ that makes an informed choice in the best model (agile or project) based on the conditions in which it’s working
We should see this change
- the right approach is taken for the right ‘work’ because there is greater shared understanding of how the approaches work and the value of their differences.
Experiment 7: Show me the money
Data/research suggests that
- discovery teams would like greater accountability
So if we try
- making discovery teams aware of their budget and the rate at which they are spending it
And measure
- the average cost of a discovery
We should see this change
- average cost decreases
Experiment 8: What’s a pre-discovery?
Data/research suggests that
- pre-discovery has become a common stage of development but has yet to be clearly defined so is used inconsistently
So if we try
- defining the value of a pre-discovery stage, and deciding if it should exist, and (if ‘yes’) applying it correctly
And measure
- number of pre-Discoveries that take place in the future versus in the past
- % discovery teams reporting that a pre-discovery helped their discovery
We should see this change
- pre-discovery of more use to discovery teams Or
- pre-discovery of more use to portfolio Or
- pre-discovery no longer used Or
- ???
Experiment 9: Clearer briefs
Data/research suggest that
- discovery teams can spend several days (sometimes over a week) clarifying the brief for a discovery
So if we try
- a consistent approach to discovery briefs
And measure
- average time spent by a team defining the brief
We should see this change
- teams begin discovery sooner
Experiment 10: Run regular discovery retrospectives
Data/research suggests that
- the questions used to carry out the retrospective that led to these hypotheses worked well
So if we try
- team/organisational discovery retrospectives using similar questions
And measure
- responses
We should see this change
- improvement in discovery.
Notes
- note 1: I wrote a post about product roadmaps in which I highlighted a format for hypotheses shared by Jock Busuttil and have used that format for this post
- note 2: it’d be great to hear from anyone who tries any of these hypotheses or has suggestions of their own, say hello on Twitter or share suggestions on Github.