What we learned about chart configuration
1. Chart configuration controls are not discoverable or self-evident
Users do not immediately recognize where or how to configure charts. Critical actions are hidden behind low-affordance icons and secondary UI elements, causing users to search rather than act with confidence.
Evidence
- Cursor gravitation toward the top-right area without decisive action.
- Hesitation and back-and-forth movements before opening settings.
- User quotes expressing uncertainty about where to make changes.
Insight
Users expect configuration to be closer to the chart itself and more visibly actionable.
2. Users lack confidence that their actions produce the intended result.
Even when users interact with controls, they are often unsure what has changed or whether the system behaved as expected.
Evidence
- “I think this changes the colours, but I’m not totally sure what it’s grouping by.”
- Insufficient feedback after configuration changes (heuristic).
- Users abandoning settings mid-interaction.
Insight
The interface does not adequately confirm cause-and-effect, leading to trial-and-error behavior.
3. The settings panel creates unnecessary cognitive load for frequent tasks.
Common actions (e.g. changing axes, grouping, or colours) require navigating a dense settings panel where all options appear equally important.
Evidence
- “There are a lot of options here… it is a process of trial and error.”
- High cognitive load identified in heuristic evaluation.
- Repeated interactions and drop-offs in Amplitude.
Insight
The system optimizes for configurability over usability, making basic tasks feel heavy and time-consuming.
4. Users’ mental model favors direct manipulation, not indirect configuration.
Users expect to interact with the chart, not configure it from a separate control surface.
Evidence
- “I expected to be able to change things directly on the chart.”
- Users clicking around the chart area before opening settings.
- Gear icon not perceived as the primary entry point for chart changes.
Insight
There is a mismatch between the product’s configuration model and users’ mental model of “editing the chart.”
5. Chart ↔ table switching feels like a disruptive context change.
Switching views breaks continuity instead of feeling like an alternative representation of the same data.
Evidence
- Too many clicks to be able to toggle between a chart and a grid.
- Inconsistent interaction patterns between chart and table views
Insight
Users expect chart and table views to share configuration, filters, and mental context.
6. Lack of visual hierarchy obscures what matters most.
Because all controls appear visually similar, users struggle to identify interactive elements and the most common or critical actions are hidden.
Evidence
- Underused but important controls (Amplitude).
- Low affordance of key actions (heuristics).
- Users overlooking primary configuration options.
Insight
Without a clear hierarchy, users must read and interpret instead of recognize and act.
7. Friction compounds across the workflow, not just in isolated moments.
What appears as “minor” discoverability issues accumulate, making routine analytical tasks feel slow and effortful.
Evidence
- “I find it time-consuming completing basic tasks such as changing the X and Y axis.”
- Repeated hesitation and backtracking in session recordings.
- Multiple methods pointing to the same breakdowns.
Insight
The experience doesn’t fail catastrophically, it fails gradually, eroding efficiency and trust over time.
Synthesized insight
The chart settings experience is powerful but opaque.
Users struggle not because the system lacks capability, but because controls are hidden, feedback is unclear, and interaction patterns don’t align with users’ expectations of direct, visual manipulation. This leads to hesitation, trial-and-error behavior, and a sense that even simple tasks require unnecessary effort.
