Saffron design system community and enablement
From drop-in to destination
How I redesigned Saffron's community sessions so designers had a reason to come back, and a way to teach each other.
A few Saffron terms, so the rest of this makes sense
This case study assumes some familiarity with Saffron, Thomson Reuters' internal design system. If that's new to you, here's what's worth knowing going in.
A drop-in format was training the community to be passive
Designers working with Saffron across Thomson Reuters product teams had inconsistent, unpredictable learning opportunities, and it showed up in the numbers. Only 52% of the community was being reached by TAPAS sessions, against an 80% goal. Only 50% of participants said attending Saffron sessions made them more efficient in their day-to-day work, against a 75% goal.† And the average Office Hours session drew just 6.5 non-Saffron participants.
The Office Hours format was open drop-in by design, so sessions were only useful to whoever happened to have a question that week. There was no advance agenda, no defined reason to attend, and no expectation of returning. Designers without an active question had no reason to show up, so the community stayed passive: receiving information when they happened to attend, rather than building a relationship with the sessions over time. Recaps were occasional and advance notice of content was rare, which meant there was no asynchronous layer for anyone who missed a live session either.
The deeper issue was structural, not content related. The format gave designers no participation scaffolding: no predictable agenda, no defined attendee role, no reason to come back. That created a community dependent on the Saffron team for reactive, one-on-one support rather than one capable of learning from, and teaching, itself.
That mattered beyond the room. A community that depends on the Saffron team for one-on-one support doesn't scale: the support load grows with every new product team, while the Saffron team doesn't grow to match it. Attendance was the clearest proxy I had for something harder to see directly: whether the community was starting to own Saffron together, or continuing to depend on us one question at a time.
A satisfaction survey confirmed what the attendance numbers suggested
In H2 2025 (the second half of the year), a satisfaction survey of the Saffron Design Ambassadors program asked participants directly about the Office Hours format. Ambassadors said they wanted more structured, predictable content, and clearer pathways for sharing their own work with the broader community, rather than only receiving information from the Saffron team.
That finding, paired with the attendance and reach numbers above, gave me an evidence base for two things: that the existing format was not serving its audience, and that the fix was about structure and reciprocity, not just better content. Ambassadors did not simply want more sessions; they wanted a predictable shape to each one, and a real way to teach the community themselves rather than only sit in on Saffron-led material.
| Signal | Result | Goal |
|---|---|---|
| TAPAS reach | 52% | 80% |
| Efficiency gains reported by participants | 50% | 75% |
| Average non-Saffron attendance per session | 6.5 | — |
Trading an open drop-in for a rotating, participatory format
I proposed replacing the open drop-in format with a rotating agenda structure, one that gave designers a reliable, predictable reason to attend and to actively participate, not just listen. Four decisions carried that forward:
Launched Show & Tell as a structured peer-presentation platform
Rather than knowledge flowing one way from the Saffron team, Show & Tell gave community members a standing slot to present work-in-progress to their peers. This shifted knowledge-sharing from something the Saffron team did to something the community did, for itself.
Replaced the open drop-in with a rotating agenda
Sessions now rotate between TAPAS education, Show & Tell, and open Q&A. Designers have a reason to attend even without an active question, because they know what a given week will cover.
Introduced advance agendas and AI-assisted recaps
An agenda posts the day before each session, naming the presenter and topic. A recap goes out within 24 hours, with links to notes and the recording. Together these build a consistent asynchronous layer for anyone who cannot attend live, something the old format never had.
Split off Ambassador Strategic Sessions
Ambassador-specific discussion moved into its own monthly Strategic Session, freeing the Collaboration Sessions to focus on the entire designer community rather than splitting attention between two audiences with different needs.
How this maps to Saffron's 2026 objectives
| CXO objective | Target metric | In plain language |
|---|---|---|
| Deliver valuable, well-attended sessions that increase Saffron awareness | 10 to 15 consistent community members per session by Dec 31, 2026 | Each session reliably draws 10 to 15 designers, not necessarily the same people every time, just a consistently healthy room. |
| Build a community of designers who return consistently and engage actively | 5 to 8 participants at 50%+ repeat attendance (Q1 milestone); 10 to 15 by Dec 31, 2026 | 10 to 15 designers outside the Saffron team showing up regularly, meaning at least half of each quarter's sessions. |
What changed by the end of Q1
General attendance: on track
Q1 averaged 12.2 non-Saffron participants per session across 10 sessions, an 88% increase over H2 2025's average of 6.5, and squarely within the 10 to 15 year-end target. Nine of ten sessions cleared the 10-person threshold; the only miss was January 8, the final session under the old format before relaunch. Attendance also climbed within the quarter itself: monthly averages were 11.0 in January, 13.3 in February, and 12.7 in March.*
Repeat attendance: narrowly missed, trending the right way
Four non-Ambassador participants sustained 50%+ attendance in Q1, one short of the five-person milestone target. One participant led at 80% (8 of 10 sessions); three more reached exactly 50%. A fifth participant reached 40%, and seven others attended three to four sessions, all building momentum in the back half of the quarter.
The sessions themselves became the primary peer-learning signal
The most significant knowledge transfer in Q1 did not happen in the support channels, it happened in the Collaboration Sessions. Every Show & Tell featured a non-Saffron designer presenting live work:
- January 22: a participant presented a wizard and stepper implementation for billing guidelines, surfacing a documentation gap that generated four Saffron action items.
- February: eight community members presented across two sessions, covering component library consolidation, non-blocking file upload patterns, an accordion-as-wizard alternative already in production, a copy-to-clipboard toast pattern, alert stacking solutions, multi-level AI activity log navigation, and persistent sidebar patterns, several of which directly informed Saffron roadmap and Ambassador discussions.
- March: community members shared deterministic AI workflow patterns and human-in-the-loop override standardization across multiple products, plus live AI workflow tools including content generation chains and UX quality assurance (UXQA) automation, two of them presenting without being prompted.
The Saffron team facilitates while the community leads the content. The Q1 2026 Ambassador quarterly report independently described the sessions as "an effective peer learning forum," which lines up with what the session notes show.
Between sessions, the channel changed shape too
The meeting chat evolved from occasional, ad hoc updates into a structured community space, with a recap inside 24 hours, an advance agenda the day before, and consistent engagement in between, both in the sessions' own meeting chat and in the Design Support channel, a Microsoft Teams channel where the community asks Saffron-related questions between sessions.
| Signal | H2 2025 | Q1 2026 |
|---|---|---|
| Post-session recaps | Occasional, ad hoc, one or two bullets | Every session, structured three to four points, linked to notes and recording |
| Advance agendas | Rare | Every session, posted the day before, naming presenter and topic |
| Reactions to recaps | Present but inconsistent | Consistent across all sessions, one to seven reactions per message |
| Between-session community messages | Not observed | Confirmed, multiple instances |
| Community contributing knowledge back | Not observed | Confirmed; consumers answered a Saffron team question unprompted in April |
In the Design Support channel, Q1 2026 threads reflected a higher community baseline than H2 2025: pattern gaps and component state edge cases rather than foundational questions. One community member, directed to a session recording, self-resolved a question and went on to submit an enhancement request unprompted.
In their own words
I attend the sessions whenever I'm not double booked and catch up on recordings when I am. I'm learning so much here. Katarzyna Ginalska, Reuters Designer · April 2026
This harkens back to our Lunch and Learns from a couple of years ago. Really enjoyed this and learned so much. Billy Gaylean, TR Product Designer · April 2026
Why this matters more than a headcount
Attendance and repeat engagement are proxies. What Saffron actually needs is a community that feels enough ownership over the design system to contribute back to it, and to adopt it consistently across product teams without needing the Saffron team in the room every time.
The two CXO objectives behind this work, drawing 10 to 15 consistent attendees per session and building a community that returns and engages actively, are not ends in themselves. They are leading indicators for two things Saffron actually needs: designers who contribute back to the system, and adoption that holds up consistently across product teams without the Saffron team hand-holding every one of them.
That's the shift I flagged going in: attendance was only ever a stand-in for whether the community was moving from consuming Saffron to owning it, and ownership is what turns a support relationship into a contribution relationship.
Q1 already shows early signs of that shift. Show & Tell presentations on component library consolidation, non-blocking file uploads, and alert stacking patterns fed directly into Saffron roadmap and Ambassador discussions, so the community was not just learning Saffron, it was shaping it. One community member self-resolved a support question from a session recording and then submitted an enhancement request without being asked, exactly the kind of unprompted, two-way contribution the old drop-in format never produced.
If that pattern holds, and Q2 is the real test of whether it does, the payoff is not a bigger meeting. It's a community that adopts Saffron more consistently because they understand it more deeply, and that contributes patterns, fixes, and feedback back into the system because they have some stake in it. That's the difference between a design system a handful of people maintain for everyone else, and one the whole design community is building together.
Q2 is the real test of whether this holds
Q1 delivered 12.2 average non-Saffron participants, already inside the 10 to 15 year-end range, with attendance still climbing on March 31. Two questions will confirm or revise that story:
- Attendance sustainability: does growth hold in Q2, countering the argument that some of the Q1 lift was relaunch novelty? Do the three near-threshold participants cross 50% repeat attendance, which would effectively meet the repeat-attendance target across H1 (the first half of the year)?
- Community depth: are Ambassador-led peer contributions in the Design Support channel strengthening? Do pure participant-to-participant support instances start to appear on their own, without a Saffron team prompt? Is Show & Tell engagement carrying into between-session behavior?