You have run a bunch of JTBD Switch Interviews, but now what? This post shows you how to turn them into actionable insights.
Everyone focuses on the skills needed to run a Switch Interview, few take the logical next step and analyse a set of interviews to get the real insights.
As Olivier Diekmann said to me in a LinkedIn conversation – this is like buying the ingredients for making a soup, putting them all in a pot, and refusing to turn the stove on…
Method One - Bob Moesta's original approach
One thing you’ll notice when you run more than 5 or 6 interviews is that you start to hear common themes. In our research project we heard a few:
- Some students had their own social media scroll stopped by a visual, then wanted the ability to do that to others
- Other students used to rely on graphic designers to create polished visuals, but now saw an opportunity to make more money by learning the skills themselves
- And there was a group who loved the feeling of drawing and felt it would be a healthy hobby for their future
After you have finished the Interviews phase of your research, you could simply:
- Review each and every interviewee and capture the essence of their purchase – what was the main reason they made the purchase?
- Open a Miro board, and drag interviewees who had similar reasons together into groups
- Drag interviewees with different motives away into separate groups
Once you have 3-5 groups where there is a common story, common reasons to buy, then you have “Jobs-to-Be-Done”. Each groups buys for a distinct set of reasons.
Pros and Cons of this method to form Jobs-to-Be-Done
The big benefit of this is that it’s a relatively simple, low-tech and accessible way to analyse the interviews you have. All you need is a table with some pieces of paper with names written on, and enough time to think this through in depth.
The drawback to this method is that you really need to have deep empathy and a grounded understanding in Jobs-to-Be-Done to know what you’re looking for.
- When I first started I could see common themes from my interviewees – they lacked the time to learn, they struggled with knowing what to learn etc – but I couldn’t see the common patterns as I had only just learned what JTBD was all about
- As you go through a few projects this becomes easier to do, but I still prefer the next method because it’s more data-driven, and easier for a beginner to understand
Method Two - Ryan Singer's analysis with Ward's Method
Ryan was kind enough to walk me through how he had set up an application to analyse Jobs-to-Be-Done. His method uses a clustering technique called Ward’s Method to separate interviewees into distinct groups.
- Instead of looking at the Big Picture, we dig deep into each individual Push or Pull statement made by an interviewee
- We still use card sorting – dragging statements that are more similar together and those that are less similar apart until we have different groups
- When we have individual piles of Push or Pull statements that fit together, we give them a group-level label
From common statements to clusters
In the research for Janis’ project, we ended up with 16 distinct groups of statements for Pushes and for Pulls.
The next step is to walk through each to get to a point where you can say “Yes” or “No” to the question “Did this reason cause this interviewee to buy?”.
- The end-state coding grid looks like the image below
Each row is an interviewee, and each column is a possible reason. The grid shows which reasons mattered to which people. If it mattered, code “Yes” (coloured light green) or whether this reason was irrelevant for this interviewee (No, coloured red).
Armed with these coding values we can use Ward’s Clustering Method (available in Python libraries – and also in other libraries too) – to drive out the clusters.
- Each interviewee starts out as a separate entity – they are 15 unique stories
- The algorithm then seeks to merge stories one-pair at a time, by grouping similar interviewees together (those who agreed with the same statements are merged first)
- Each merge is chosen to make the smallest possible increase in difference within the new group.
The outcome ensures you end up with the most natural clusters – stories that fit tightly together because they’re driven by the same core struggle or goal.
Given the number of interviews we are running, and the level of abstraction we have chosen for the statements, we aim to create 3 groups, 4 groups and 5 groups using this method.
- The final choice is based on reviewing the groupings and asking whether they make sense
- Can you tell a different story with 3 vs 4 vs 5 groups?
- Are you separating people that share a common story (if so, go up a level and have fewer clusters)
- Does it make sense that the individual interviewees have been joined together in a cluster?
Pros and Cons of this method to form Jobs-to-Be-Done
Obviously this requires a lot more work:
- First you sort the individual statements and label them at a group level
- Then you code each interviewee as a yes or a no for that statement
- Finally you need a way to shove all this data through a Ward’s Clustering algorithm so you can analyse it
I think the benefit of this approach is:
- Forcing you to think more deeply about the underlying motives that cause a purchase to happen (or not)
- Being certain that the difference within each group is minimised, and the difference between groups highlighted (so we can be deliberate in targeting different groups with the messages they need to hear)
- Having a mathematically sound rationale for forming the Jobs-t0-Be-done (important for some stakeholders so they trust the methodology).
What's next?
Once you have a set of clusters and you’re confident that the analysis is solid, you can move on to detailing out the Jobs-to-Be-Done. I’ll share more about that approach in our next post, so you can see:
- How to see the differences in demand between groups
- What to do with those insights to make better decisions about your own offering
- How we eventually map the demand discovered in the market to your supply-side offer, so we can work out what to change and what to keep to delight customers