Continuous Discovery Research: Quality, Process, and?Outcomes

In our early October 2023 free webinar, I looked at what continuous discovery often is, and what it should be. How can we plan continuous cycles of research so that we are frequently delivering actionable insights and strategic suggestions?

We looked at some diagrams from Teresa Torres and Tim Herbig, and applied critical thinking to what they recommend for or against.

The webinar is a great watch, but here’s a very short summary in article form.

What Is Continuous Discovery?

• Frequently gather insights into target audiences to inform product decisions.
• Shift teams from OGA (opinions, guesses, and assumptions) to EDK (evidence, data, and knowledge).
• Reduce our failure rate. Right now, people seem to think high failure rates are cool and business as usual. But if we did better discovery work and better understood target audiences, we would fail much less often.

Avoid Common Pitfalls

• Research rigor and standards. It can seem faster and more “democratic” for anyone and everyone to conduct research. But there isn’t a great reason to have people who are poor at research planning, execution, or analysis doing research. Flawed or invalid research can set us up later for failures.
• Running surveys or asking people what they want when we should be observing tasks and behaviors.
• Speaking with the same customers repeatedly skews findings since they are savvy. They know how to get around your awful system, and might not ask for — or know — what they really need. Recruit new people and target audiences who aren’t customers.
• “A few” weekly customer conversations aren’t enough (and might not ask the right questions). A few weekly customer conversations are more like a slow-motion survey. If you really want to understand a target segment, qualified Researchers will need to observe and speak to more than “a few” people. The best practice is 8–12 people per target segment, plus people with disabilities, diagnoses, and conditions. If you have three target segments, you should meet 30+ people to ensure that your participants are good representatives of your target populations.
• Don’t expect customers to understand their own problems or innovate for you. It’s your job to understand them, including insights they don’t even have about themselves. Great Researchers will notice habits and problems customers don’t even know they have.
• Approaching discovery work with a hypothesis or idea puts you in a box, biases you toward validating your idea, and makes you more solution-focused than you should be. Discovery is about discovering more about users, contexts, and behaviors. It’s about understanding tasks, mental models, and unmet needs. Making discovery about “discovering” people who like your idea, or discovering how you can push people to do what the business wants are two of many reasons we see so many product and experiment failures later on.
• Speed over quality. Going fast is nice, but not when we consider risks, the value we are trying to create, and how often we get it partially or completely wrong. Quality insights take time, but will set projects and decisions up for success.

Cadence of Continuous Discovery Research

When we care about the quality of the research work and the outcomes it creates, then our discovery cadence is more about how great our work can be versus how fast it can be. We’re OK with it going a little slower than a random, arbitrary, unreasonable speed standard invented for Engineers.

• Have at least two Researchers per project. This speeds up the project while also making sure that someone always has 100% knowledge in case someone quits, is laid off, or goes on vacation or medical leave.
• Consider how long it takes to do great research. Generative research takes around 6 weeks to do well, sometimes longer. Evaluative testing might happen in a week, but often needs two weeks or longer. My current job has a large number of diverse target audiences. To make sure they are all represented in our research, our generative project is expected to take five to six months. Our evaluative cycles are taking around seven weeks each.
• Do you want insights every sprint? Every week? Then hire the number of teams you need to create that cadence. If you have mostly 6-week research projects, and you want fresh knowledge each week, then you need 6 teams working concurrently. That way, each week, one team is ready to present their findings.
• Establish cycles and a cadence based on business needs, not arbitrary time blocks. What do you need to learn? We know you want to learn it yesterday, but it will be better to learn the right things the right ways, even if that takes time. The companies you admire the most do it this way.

Make the Investment

• Proper staffing and time for discovery provide substantial ROI through risk reduction. Strategies, decisions, priorities, and products can be guided by good evidence and knowledge.
• Remove the risk of working from guesses and assumptions. We’re wrong too often. Our high A/B test, experiment, and product launch failure rates tell us this.
• Continuous insights prevent wasted effort on features users don’t need or want. But only if our research is solution-agnostic. If our research tries to prove our idea is good and people want it, we are still in the feature factory hamster wheel.
• Do the math. Removing days of a team’s rework, root cause analysis, fixes, failed experiment cycles, etc. pays Researcher salaries. Avoiding a single project failure pays Researcher salaries.

Conclusion

Continuous discovery is a long game requiring patience and proper resourcing. Rushing through superficial sessions leads to non-actionable “insights,” false “validation,” or incorrect “data.”

With the right research talent and realistic cycles, research provides ongoing value by empowering evidence-based decisions and reducing risks and waste.

Connect with us or learn more:

• CustomerCentricity.com — our website
• youtube.com/@cx-cc — YouTube channel
• patreon.com/cxcc — online community and free mailing list