We were inspired by Teresa Torres’ breakdown of how 99designs accomplishes continuous discovery, so we wanted to share our own.
In the post, the 99designs team describes what their stack might be like in a perfect world. The perfect stack includes a research database where every team can view learnings and become aware of what customers have been interacted with and what they spoke about.
Sound familiar? That’s what we built in CustomerIQ.
Given this, we want to share how we automate the planning and data collection that leads up to submissions in our CustomerIQ workspace. Considering our own experience and that of 99designs, we think it’s near-perfect.
Before we dive in…
What is continuous discovery?
Continuous discovery is a product management approach invented by Teresa Torres focused on ongoing conversations with customers to better understand their needs, pain points, and goals. This continuous cycle of learning allows product teams to consistently collect new information and validate their assumptions, enabling them to build more effective and customer-centric products.
Challenges we want to solve for with our stack
Through past experience and a newfound desire to do discovery continuously, we knew we wanted to solve for three challenges
1. Build a process we can stick to
First we want to solve for inaction. We know we need to be continuously talking to customers but the effort that it takes to recruit and schedule interviews leads us to inevitably dropping the ball.
So we need to automatically get in front of certain customer segments with an invite to speak to us, then we need them to book a meeting whenever suits them (you see where we’re going with this).
2. Automate the synthesis
The second issue we want to solve is how long it takes to synthesize our conversations. We’ve found it takes 2-3 hours to do a thorough analysis of a 1 hour conversation. To do these conversations continuously we need to dramatically decrease that effort.
3. Maintain an insights database
The third issue we’re solving for is duplication of work or worse, annoying customers. We need somewhere to store everything we learn *and* who we spoke with and what we spoke about. The 99designs team mentions this as their ideal future state, so we’re happy to know we’re not alone.
Alright, here’s how we did it.
Core components and our stack
Tl:dr we’re using: Figma, Google Docs, Posthog, Hubspot, Zoom, and CustomerIQ. This sounds like a lot, but the core of the continuous process is managed by Hubspot, Zoom, and CustomerIQ. Let’s dive in.
Mapping out opportunity space
Before we send out requests to users we need to know who we need to speak with and about what. We identify that by identifying our current target opportunity. We use Figjam to build out a simplified opportunity/solution tree (shoutout Stefan Richter for the template) , and point to the opportunity we’re discovering
Then we use Google docs to plan our conversations (you can download the template here). We create a new guide every time we switch opportunity spaces. This way we can have a fruitful conversation without stressing about what to ask. We’ve already thought of the questions and are ready to gather some great, unbiased data.
On the product side we mostly target our own users. This way they have a ton of context into what space we’re solving for but we can also automate their interview invites. We primarily invite users to interviews via in-app surveys from Posthog. We use Posthog for other product analytics so it makes sense to use their survey feature too.
What’s more, this combination allows us to target specific user segments who will have more knowledge of our problem space.
We also send targeted emails through Hubspot but using data we capture from Posthog for specific segments.
When we send the in-app survey through Posthog it contains a link to our calendar to book a meeting. For this we use Hubspot, but Calendly is another popular choice. There’s nothing better than logging in in the morning and discovering a new meeting booked. Huzzah!
Prepped with our interview guide, we conduct and record the interview using Zoom. We have also experimented with meeting agents who record automatically but for some reason always revert back to the classic recorded Zoom.
Every interview gets loaded into our CustomerIQ folder we’ve aptly named, “Product discovery interviews.”
By far the best thing about this is we don’t need to take notes during the meeting, the folder takes notes for us.
Right after the interview is added we can watch the insights count stack up with all the valuable data we gathered during the call. Each note is tagged with a sentiment and readied for synthesis.
We have a couple “Views” setup in CustomerIQ where we’re analyzing the insights in the Product discovery folder from a few different angles. First we filter a view by insights with a negative sentiment, and use the Discover feature to identify themes. This tells us the key challenges facing the user. In a different view, we filter by positive sentiment and discover those themes. This tells us what they enjoy. We also share these views with Marketing as this helps us position our existing solutions.
It’s important to note this synthesis process only takes a few minutes. This gives us the confidence to have more conversations knowing the opportunities are automatically generated.
Collaboration and sharing
By this point we’ve usually identified 3-4 opportunities worth sharing so we add what we’ve learned to a doc in CustomerIQ and share with everyone on our team.
Bonus: Designing and running experiments
The stack above is how we conduct all our problem discovery but we did want to share a quick note on extending into solution discovery. Here’s how we do that:
We use Figma for everything on the design side (why wouldn’t you?!). We tend to create pretty high-fidelity prototypes that we can just hand over to a user (perfect world), or at least demo (still gets good feedback).
User testing prototypes
We’re exploring some prototype tools where we can watch the user interact with the prototype similar to session recordings in the product. The current front-runner is Maze. We’d love to know if anyone has any experience here - let us know on X (Formerly known as Twitter).
When we record these sessions and feedback we still use CustomerIQ for synthesis and summary, we just use a different folder. We like to separate this feedback from our problem discovery since they’re inherently two different sides of the coin.
Lastly, if you’re providing incentives for a user’s feedback we use Tremendous. We also like to tag that user in CustomerIQ so we can track when they’ve been paid. Something like, “Paid Sept 2023.”
Setup your insights database with CustomerIQ
We hope this helps! It sure helps us. Hopefully by now you understand the value in automating your discovery synthesis and building out an insights database. You can build yours totally free by creating a CustomerIQ workspace, we’d love to have you.