How Users Move and Where They Drop Off
Track feature drop-offs and cross-tool movement patterns to surface the clearest UX and growth opportunities.
3 Critical Feature Discovery Gaps
Features that exist but are functionally invisible to users
Export / Download Not Measurable
297 users successfully created visualizations, but CDE logs currently do not capture download/export actions. We cannot yet distinguish true non-usage from missing tracking.
Recommendation
Add an explicit download/export analytics event first. Then evaluate discoverability and add a post-render CTA if observed usage is still low.
Spatial Search Rarely Applied
Of 14,007 EUI visits, only 222 (1.58%) opened the spatial search panel — and just 15 ever applied it as an active filter. Users open the feature but don't complete the workflow.
Recommendation
Add an in-panel tooltip explaining what spatial search does and why to apply it. The “Apply Filter” button may need to be more salient — consider auto-applying on config completion.
Opacity Panel Undiscovered
The opacity settings panel toggle was found by only 26 users in the entire dataset.262 total opacity interactions happened — all from those who stumbled on it manually.
Recommendation
Expose opacity controls in the main sidebar as an expanded section, not behind a collapsed toggle. Or add a first-use tooltip pointing to the panel after the user places a tissue block.
EUI Spatial Search — Drop-off Funnel
Of 222 users who opened spatial search, only 15 (7%) ever applied it as a data filter
The Entry Problem
Only 1.58% of EUI visitors (222 of 14,007) ever open the spatial search panel. The feature is valuable but not prominent enough to be discovered organically.
The Completion Gap
Of those 222 who open the panel, 104 (47%) configure an organ but only 15 (7%) apply it as a filter. Users explore the feature but don't understand or find the final “Apply” action.
Note on “Explored Results”
323 result views exceed the 104 who configured — likely users re-opening an existing search. Results are browsed but rarely committed to.
CDE User Journey — Where Users Go and Where They Stop
Flow of all 362 users who uploaded data through the CDE workflow
Direct Entry: 64%
233 of 362 uploaders bypassed the landing page entirely — arriving directly at the create page via bookmarks, external links, or navigation. Suggests the landing page is not the primary discovery path.
Config Drop-off: 18%
65 users (18%) abandoned at the configuration step — 5 axis selectors + cell type + parameters. This is the highest-friction step in the workflow. Defaults or smart suggestions could help.
Instrumentation Gap
Download/export interactions are not currently logged in CDE. Add an explicit event to measure whether users stop at visualization or continue to export.
Tool Co-movement Heatmap
Pearson correlation of monthly visits · Jan 2024 – Jan 2026 · KG dampened by 21 months of zeros pre-launch
Correlation Force Graph
Node size = total visits · edge weight = Pearson r · dashed = weak link · drag nodes to explore
Where Do Users Go Between Tools?
Directed flow of tool-switching sessions · arrow thickness = session count · % label = probability from source tool
KG Explorer is the dominant destination — 82% of FTU Explorer exits and 80% of CDE exits flow to KG Explorer. It acts as the knowledge layer users return to between tool sessions.
89% of RUI exits go to EUI — users who register tissue blocks immediately explore spatial context in EUI. This is the clearest sequential workflow in the entire suite.
KG ↔ CDE and KG ↔ FTU show strong bidirectional loops — users alternate between exploring the knowledge graph and drilling into cell or FTU data. Cross-linking these pairs more tightly would reduce friction in these natural workflows.
Cross-Tool Usage (Session-Level)
Do users move between tools in the same session?
Users strongly prefer staying in one tool per session. All cross-tool pairings have lift < 1 — meaning switching tools is less likely than random chance.
Most cross-tool sessions involve KG Explorer + CDE (34 sessions). Use lightweight cross-links in context rather than aggressive handoff prompts.
Who Actually Uses Multiple Tools?
2,052 users visited 2+ tools across their full history · top combo: CDE + EUI + RUI (554 users)
The top combo — CDE + EUI + RUI — is the “full atlas” power user workflow: data upload, 3D exploration, and registration in one session. These 554 users are your highest-value audience.
KG Explorer already appears in cross-tool combos with EUI and RUI despite launching only in Aug 2025. This confirms it's being adopted as a complementary lookup tool, not a standalone replacement.
Surface contextual links between the top 3 combos in each tool's UI. Even a 5% increase in the multi-tool rate would represent hundreds of higher-engagement sessions per month.
Cross-Tool Growth Opportunities
Derived from monthly co-occurrence patterns — tools that move together share audiences
Strongest correlation in the dataset. Both spike together during IU workshop events — they share the same workshop-attending researcher audience. Adding an FTU Explorer CTA inside EUI (and vice versa) could increase adoption of both with near-zero engineering cost.
Action: Add “Explore organ FTUs” button in EUI sidebar when a user selects an organ
CDE and RUI share a data-registration researcher audience who work with tissue-level data. CDE users who visualize cell distributions are likely the same users who register tissue blocks in RUI. Cross-linking them could improve both tool stickiness.
Action: Link from CDE visualization to “Register this tissue in RUI” workflow
Near-zero correlation means EUI and CDE attract separate user personas. EUI serves spatial browsers; CDE serves data-upload researchers. Attempting to cross-promote these two would likely see low conversion — focus cross-promotion budget elsewhere.
No action: Different audiences — don't conflate them in onboarding or marketing
KG Explorer peaked in October 2025 at launch excitement and has declined 45% since. Over its active window it shows r=0.64 correlation with RUI — suggesting its audience overlaps with RUI's power users. Targeted re-engagement of RUI users could reverse the KG decline.
Action: Add KG Explorer surface to RUI sidebar — “Explore the knowledge graph for this structure”
What's Working