Before we reskin School Comparison, let's define what "better" means
A measurement plan for schools/comparison — our first real production page for the dataviz reskin — built from Vitaly Friedman's "How to Measure UX" and the open questions on design‑crit‑dataviz. Goal: pick a reskin direction because it tests better, not because it looks better.
The design-crit brief says the real problem isn't how charts look — it's that "many show a number with no context" and teams have to relearn each one. School Comparison is a good first production target because it has both chart categories we're least sure about: a KPI hero row (4 tiles, no context shown) and a color-intensity heatmap table (comparison depends entirely on people reading color correctly).
Friedman's warning: if we jump straight from the 70 sketches to "ship the one the room likes," we're designing pixels and hoping. The fix is cheap — test the current page first, so we have a number to beat.
Step 1 of the strategy. Run these on today's live page, before any sketch is chosen. This is the number every reskin direction has to beat.
| Task we give the administrator | What it actually tests | Why it matters here |
|---|---|---|
| "Which school has the highest average total time spent per day?" | Content success | Table encodes value by color intensity, not just number. Does color let them answer correctly, or do they have to read every cell? |
| "Is Madison Middle School's staff time higher or lower than Kennedy Middle School's?" | Content success | Two mid-range, similarly-shaded cells. If this fails, the heatmap scale doesn't have enough perceptual steps. |
| "What does '197m — Time Spent — This Month' actually mean: per student, per day, or district-wide?" | Missing context | This is the exact gap named in the project brief. If admins guess wrong here, the KPI hero redesign is the priority, not color. |
| "Show me how you'd compare two specific schools side-by-side." | Navigation / discovery | Tests whether Compare is findable at all — a navigation-success question, kept separate from content-success above. |
Step 2. Once "Heroes & KPI headers" and "Heatmaps" are narrowed to 2–3 directions each, don't pick with a vote — re-run the same loop.
Step 3. This is the bridge Friedman says is usually missing — connecting a design initiative to a business goal the org already tracks, through metrics we can actually measure.
Step 4. Applied to this page specifically, feeding the design-crit site's own "prove it beyond one page" step.
Step 5. This page is a live stress-test for exactly the open items on the design-crit site — all four tile hues sit on screen together here, which no single sketch shows.
Run a colorblind simulation on this exact screen (all four tiles + the table's blue intensity scale together, not in isolation) before locking tokens. No donuts appear on this page — good, consistent with the existing constraint.
- 1Do we approve running the 4-task baseline on the current page before sketching this page further?Decision: pending
- 2Who can get us 15–18 district admins — or is Customer Success our proxy for round one?Decision: pending
- 3Which 2–3 "Heroes & KPI headers" and "Heatmap" sketches do we carry into the re-test round?Decision: pending