Co-Creation User Research · Community
Fan engagement beyond listening
Led the qualitative study, Co-Creation, to generate new thinking with Podimo users rather than evaluate existing features, uncovering the moments, motivations and unmet needs that sit behind how people engage with content. Partnered with a data scientist on a quantitative follow-up, a Community Feature Hierarchy study, and together we built a mixed-methods case that reframed the roadmap around 3 shifts: episode - and host level and listener engagement. That combined evidence directly shaped feature prioritization with cross-functional stakeholders across product, design and data science teams.
The starting point. Currently Community on Podimo is a generic, platform-wide layer that doesn't reflect yet how users actually experience different shows. Past feedback signalling resistance to social media-style features made it clear the current model needs to be challenged.
Why co-creation as a research method. Rather than evaluating existing features, we wanted to generate new thinking together with users: letting them define what participation could look like on their own terms, unconstrained by what already exists. The goal is to surface new models, not validate old ones.
Read every finding as directional. All participants were unusually loyal, high-intensity users. Every finding should be read as directional evidence from the most engaged end of the user base, not generalisable to the median user.
When does community add value — and when does it get in the way?
TODO — add copy: the business + user problem, and what you got wrong.
TODO — add copy: your scope, influence and what you initiated.
TODO — add copy: the decision you took, before the solution.
Type. Co-creation (exploratory/generative — not evaluative user research). Open-ended by design: the goal is to discover possible directions, not to confirm a predefined solution.
Four phases per session. Relationship to the show they love · co-creation sketching · reflection · concept wireframes.
On the wireframes. These were shown only once a participant's own sketching had already moved there, or briefly at the end — the purpose was not to validate features, but to see whether users would arrive at similar concepts by themselves.
Research Ops learnings: navigation friction repeatedly crowded out the community conversation. Passive engagement surfaced as something that may be undervalued in community thinking. And it is genuinely hard for existing users to ideate into novel community territory from their own experience — the most original ideas came from the participants with the most cross-platform exposure.
Insight presentation to the Community team, structured as an executive summary, the reasoning behind the method, who we talked to, insights across four chapters, and an appendix routing out-of-scope signals to the teams that own them.
Fan identity is host-mediated, not community (peer-to-peer) mediated. The core attachment is to the hosts as people, not to the show as a brand. Features that foreground host presence carry more perceived relevance than features that foreground peer interaction.
Community = extension of the episode, not a separate destination. The window during or immediately after an episode is when community engagement feels natural. From the moment a feature requires navigating away from the player, that window closes before it becomes accessible.
Most people we talked to want to observe, not participate — and that’s real engagement. Reading without posting is community; knowing others are listening is community. The gap between lurking and contributing is one big jump, and most people don’t take it.
Navigation must improve first. We can’t enrich community without fixing how users move through the app. Navigation is upstream of everything else.
TODO — add copy: what you would do differently.
When I’m in the episode, anything related to the episode I’m engaged with. If it’s outside... I don’t care.
Which community features actually earn their prominence
Community grew feature by feature — a player redesign inspired the hierarchy question. This deep dive asks which community features actually earn their prominence, and provides suggestion to re-rank them.
Community grew feature by feature. Comments and reactions first, then polls, then a community feed, with more still shipping. Features coexist in the same places but also in separate placements, so the question became a systems one rather than a placement decision.
Why it matters. Prominence is zero-sum: every hero slot given to one feature is attention taken from another. Placement drives behaviour, and not all engagement is equal. Getting it wrong compounds — teams keep investing in the wrong things while the real driver of retention stays starved.
So: which community features actually deserve prominence?
1. Does it matter? — value signals. Reach, stickiness, churn impact and contribution to the north star metric.
2. Is the signal real? — rigor. Controls for listening depth, tenure, plan and market; controls for overlap between the community features themselves; robustness across multiple cohorts and windows.
3. Context. Growth trend, and each feature mapped to the surfaces it currently occupies.
4. Not here. The WHY — qualitative, supplied by the co-creation study above. And causal proof, which would need a randomised test.
A feature earns a high place in the hierarchy only when it clears the value signals AND survives the rigor checks — the climb is what's popular → what independently retains → (next) proven cause.
Limitations carried into the readout: a single point in time rather than a trend, one market group excluded, and an observational design — a strong association, not proven causation.
Realign prominence with value: elevate comment-reading, keep polls as a growth play, and make the feed content earn its real estate. Before we re-rank anything: a feature only deserves a slot if it serves that slot's core job better than what's there today — not just because it performs well on its own.
Reading comments is the community engine. Viewing comments came out first on every axis — widest reach, stickiest week to week, largest contribution to the north star metric — and after controlling for real listening depth, it was the only community feature with an independent retention effect. The others lost their apparent protection once the engaged listening they co-occur with was accounted for.
The feed has a stickiness problem, not a reach problem. People find the feed — discovery isn't the issue. What they find once there isn't landing.
Polls have a small and fast-growing audience, but are not an independent lever. The effort-versus-impact tradeoff is less favourable than comments or reactions.
I think the thing that you need to do when the podcast is finished is interact with it, place a comment — I noticed that in this view I cannot see the comment section.