Synthesis and Research Repositories for UX Researchers (2026)

From raw data to insight: synthesis as a craft

Synthesis is the work between data collection and a finding the team can act on. New researchers underestimate it; senior researchers spend more time on synthesis than on fieldwork. The craft has a sequence:

1. Capture cleanly during fieldwork. Recordings or transcripts are the source of truth; sketch notes are the index. Tag each session with participant pseudonym, segment, date, and study ID so it is queryable later.
2. Open coding. Read or watch the raw material and tag every fragment that strikes you with a short descriptive label — a phrase, not a category. The Indi Young approach (indiyoung.com) treats this as listening for the participant's reasoning, not for product validation. Codes accumulate; do not pre-build the taxonomy.
3. Axial coding. Once open codes accumulate, group them into higher-order categories and look for the relationships between categories. This is where patterns emerge: causal chains, contradictions across segments, conditions under which a behavior shows up.
4. Affinity at the right altitude. Cluster the categories into themes a non-researcher can hold in their head. Three to seven themes per study is the right zone; more than that and the audience cannot remember; fewer and the synthesis collapsed important nuance.
5. Name the insight. An insight is a sentence that explains user behavior in a way that predicts what the user will do next. "Users get confused by the pricing page" is a finding. "Users abandon the pricing page when the unit-of-billing does not match how they conceptualize their own usage" is an insight. The latter changes the roadmap; the former generates a tooltip.

Erika Hall's Just Enough Research (muleshq.com) is the canonical discipline reference. Hall's argument: more data does not produce more insight, and most teams collect more than they synthesize. The senior researcher's move is to spend collection effort proportional to synthesis capacity. Tomer Sharon's It's Our Research (tomersharon.com) reinforces this with an organizational lens: synthesis is where research credibility is won or lost, because the synthesis is what stakeholders actually consume.

Affinity mapping at scale, remote vs in-person

Affinity mapping is the workhorse synthesis method. The mechanics are simple: write each observation, quote, or code on a sticky note; cluster notes that feel related; label the clusters; iterate until clusters are stable. The Nielsen Norman Group primer (nngroup.com/articles/affinity-diagram) walks through the canonical method.

The medium changes the practice:

• Physical sticky notes on a wall. Still the gold standard when the research team is co-located. The kinesthetic act of moving notes by hand activates pattern recognition in a way no software replicates. The downside: the artifact is ephemeral; you photograph it, the participants see only the photo, and re-clustering after a break is friction-heavy.
• Miro or FigJam. The distributed default. Both tools support sticky-note clustering, lasso selection, frame-as-cluster, and tagging. Miro has stronger templates and an established UXR template library; FigJam integrates more naturally with Figma design files. Pick by where the rest of the team already lives.
• Inside a research repository. Dovetail and EnjoyHQ both support tagging and cluster-by-tag. The advantage: the affinity map is queryable later, and the same evidence can roll up into multiple themes. The disadvantage: clustering is more rigid than free-form sticky notes; some patterns only emerge when you can move things spatially.

The senior practice in 2026: do the first messy pass in Miro or FigJam (or on a wall) where spatial arrangement is fluid, then encode the resulting themes and supporting evidence into the repository as the durable record. The exploratory medium and the durable medium are different tools because they serve different purposes.

A few field-tested rules. First, write notes verbatim from the source — not a paraphrase, not an interpretation. The cluster's meaning lives in the language the participant used. Second, resist labeling clusters until the third or fourth pass; early labels lock in a frame and bias subsequent clustering. Third, time-box ruthlessly; affinity mapping expands to fill any time you give it, and the marginal insight per hour drops sharply after the first focused session.

Research repositories: tooling and reuse

A research repository is a system of record for evidence, themes, and insights. Without one, every study is a one-shot artifact that lives in a Drive folder and decays. With one, the research function compounds: insights from study three inform the design of study seven; a question from a product manager can be answered against five years of accumulated evidence.

Kate Towsey's Research That Scales (rosenfeldmedia.com/books/research-that-scales) is the canonical reference for the operating model. Towsey treats the repository as one component of a ResearchOps practice that also includes participant recruitment, study templates, taxonomy governance, and findings democratization. The tool matters less than the governance.

The 2026 tooling landscape:

• Dovetail. The market leader for dedicated UXR repositories. Strong transcript ingestion, tagging, AI-assisted theme detection (with the caveat that AI suggestions are draft, not truth), and insight-as-first-class object. The Dovetail blog (dovetail.com/blog) publishes practitioner-quality essays on synthesis methodology and repository governance.
• EnjoyHQ (now Great Question). Repository plus participant-recruitment platform; strong for teams that want one tool spanning ResearchOps and synthesis.
• Notion as a repository. Works at small scale with disciplined database design: a Sessions table, an Evidence table linked to sessions, a Themes table linked to evidence, an Insights table linked to themes. The friction is governance — without an owner, the structure decays. The Towsey book documents the model in detail.
• Coda as a repository. Similar trade-offs to Notion; stronger at relational queries and dashboards, weaker on rich-text ingestion of transcripts. Better for repositories that need to roll up to leadership dashboards.

The key design decision is taxonomy. A flat tag soup becomes unsearchable past 50 studies. A rigid hierarchy fails the day a new product line launches. The pragmatic middle: a small set of structural tags (segment, product surface, study type, date) plus a curated thematic taxonomy that is reviewed quarterly. Towsey calls this the "governed tag set" and treats it as a librarian's job, not a researcher's afterthought.

The reuse payoff is real. A repository with two years of governed evidence lets a researcher answer questions like, "have we heard this concern from enterprise customers in any prior study," or, "what do our highest-engagement users say about onboarding across all the studies we have run." These are meta-analytic questions, and they are the questions executives actually want answered.

Communicating findings that change roadmaps

A finding that nobody acts on is a research artifact, not research impact. The senior UXR move in 2026 is to match the communication format to the audience and the decision the team is making. The same insight has multiple shapes:

• The one-pager. For executives and product leadership. Single page, top of the page is the insight in one sentence, then three to five supporting bullets, then the recommended action. Nobody reads the third page; do not write a third page.
• The video highlight reel. Two to four minutes of stitched participant clips that ground the insight in the participant's actual voice. This is the format that changes design teams' minds the fastest, because designers respond to user reality more than to researcher abstraction. Tools like Reduct.video, Dovetail's clip feature, and EnjoyHQ's reel builder make this practical.
• The deck. For design review or stakeholder readout. Structured around three to seven themes; one slide per theme; each slide pairs the theme with a primary quote and a recommended next step. Anti-pattern: dumping the affinity map into the deck as evidence of work.
• The Slack thread or async write-up. For distributed teams and ongoing programs. Lower-ceremony, higher-frequency. Indi Young's writing on listening sessions (indiyoung.com) advocates for sharing in-progress observations rather than waiting for a polished readout, because in-progress sharing keeps stakeholders engaged with the work as it unfolds.
• The repository entry itself. The durable record. Title is the insight, body is the evidence, tags are the taxonomy, links are to source sessions. Even if no other format is produced, the repository entry must exist, because that is what makes the insight reusable.

The async-vs-sync choice matters. Synchronous readouts (the deck, the meeting) work when the team needs to align in real time on a decision. Asynchronous formats (one-pager, repository entry, Slack thread) work when stakeholders are distributed or the insight is ongoing. The cheap mistake is to schedule a 60-minute meeting to communicate a finding that should have been a one-pager link; the more expensive mistake is to write a one-pager when the team genuinely needs to argue about implications and a meeting was the right tool.

Across formats, the discipline is the same: lead with the insight, ground in evidence, recommend an action. Tomer Sharon's It's Our Research (tomersharon.com) frames this as research credibility — every readout that does not lead with the insight burns trust. The repository preserves the work; the communication format moves the roadmap. Both are the job.