Pages
Teams & Users — Strip Bites
Source: qdash /teams — c67 micro-viz bake-offs, 2026-08-15
The swarm chip is the driving form: bite-sized visual overviews inside the chip geometry we already ship. The metric switcher is the cognitive axis — and each lens gets the micro-visualization that best answers its question, chosen by bake-off below, not by default. The law: a micro-viz earns the slot only if it answers the lens's question better than the number beside it. Otherwise, ship the number.
1
Spend
The lens's question: how is the whole divided — and how is it distributed inside the team?
candidateShare barthe baseline — one number, drawn
ML-Infra$4.2k23% of fleet · 31 people
Payments-EU$2.9k16% of fleet · 24 people
Core-Banking$6003% of fleet · 9 people
Research$9005% of fleet · 5 people
Support-Ops$3002% of fleet · 12 people
Unassigned$4.1k22% of fleet · 61 unmapped
candidateTwo-slice micro-donutshare of fleet as a ring
ML-Infra$4.2k23% of fleet · 31 people
Payments-EU$2.9k16% of fleet · 24 people
Core-Banking$6003% of fleet · 9 people
Research$9005% of fleet · 5 people
Support-Ops$3002% of fleet · 12 people
Unassigned$4.1k22% of fleet · 61 unmapped
winnerMicro-swarmone dot per person, x = log spend, color = state
ML-Infra$4.2k23% of fleet · 31 people
Payments-EU$2.9k16% of fleet · 24 people
Core-Banking$6003% of fleet · 9 people
Research$9005% of fleet · 5 people
Support-Ops$3002% of fleet · 12 people
Unassigned$4.1k22% of fleet · 61 unmapped
Verdict — micro-swarm. The donut is the dataviz procedure's literal anti-example (a ratio wants a meter, and the sub-line already says "23%"). The bar repeats the sub-line in ink. Only the swarm adds what the number cannot: shape — Payments-EU's two red high-spend dots, the amber mass, the long quiet tails.
2
$ / person
The lens's question: is this team's burn typical or extreme against the fleet?
candidateScaled barthe baseline — longest bar = hottest team
ML-Infra$135per person · fleet median $75
Payments-EU$121per person · fleet median $75
Core-Banking$67per person · fleet median $75
Research$180per person · fleet median $75
Support-Ops$25per person · fleet median $75
Unassigned$67per identity · fleet median $75
winnerDot on the fleet rangemin / median / max ticks, the team as a positioned dot
ML-Infra$135per person · fleet median $75
Payments-EU$121per person · fleet median $75
Core-Banking$67per person · fleet median $75
Research$180per person · fleet median $75
Support-Ops$25per person · fleet median $75
Unassigned$67per identity · fleet median $75
Verdict — dot-on-range — the "progress slider" idea made honest: not progress, position-in-distribution. The bar ranks; the dot locates. Research sits past everyone at $180; Unassigned rides the same range at $67 per identity, amber. The median tick is the sentence "2.4× the fleet median" drawn instead of said.
3
Active
The lens's question: how much of the team is on right now?
candidateContinuous meterthe baseline — fill = running share
ML-Infra12/31running right now · 31 people
Payments-EU9/24running right now · 24 people
Core-Banking4/9running right now · 9 people
Research3/5running right now · 5 people
Support-Ops11/12running right now · 12 people
Unassigned3/61running right now · 61 unmapped
candidatePie / ringthe same ratio, rounder
ML-Infra12/31running right now · 31 people
Payments-EU9/24running right now · 24 people
Core-Banking4/9running right now · 9 people
Research3/5running right now · 5 people
Support-Ops11/12running right now · 12 people
Unassigned3/61running right now · 61 unmapped
winnerUnit dotsone dot per person, filled = running · meter fallback past ~12
ML-Infra12/31running right now · 31 people
Payments-EU9/24running right now · 24 people
Core-Banking4/9running right now · 9 people
Research3/5running right now · 5 people
Support-Ops11/12running right now · 12 people
Unassigned3/61running right now · 61 unmapped
Verdict — unit dots. People are countable — at n=9 a continuous fill fakes continuity, and "4 of 9" as four filled dots is read without arithmetic. The pie dies by the skill rule (a single ratio → meter, never a two-slice pie). Above ~12 the dots collapse into exactly the continuous meter — the same density threshold the focus card uses, so the system has one rule.
4
Risk
The lens's question: how bad — and is it one person or systemic?
candidateSeverity wordsthe baseline — "2 Crit · 1 Elev", no marks
ML-Infra1
1 Elev
open items · 31 peoplePayments-EU2
2 Crit
open items · 24 peopleCore-Banking3
2 Crit · 1 Elev
open items · 9 peopleResearchclear
clear
nothing open · 5 peopleSupport-Opsclear
clear
nothing open · 12 peopleUnassigned1
1 Crit
open items · 61 unmappedcandidateHeat celltint intensity = open count
ML-Infra1open items · 31 people
Payments-EU2open items · 24 people
Core-Banking3open items · 9 people
Researchclearnothing open · 5 people
Support-Opsclearnothing open · 12 people
Unassigned1open items · 61 unmapped
winnerSeverity glyphsone square per open item, countable
ML-Infra1open items · 31 people
Payments-EU2open items · 24 people
Core-Banking3open items · 9 people
Researchclear
clear
nothing open · 5 peopleSupport-Opsclear
clear
nothing open · 12 peopleUnassigned1open items · 61 unmapped
Verdict — glyphs. Findings are units, not magnitudes — three squares answer "one person or systemic" at a glance and each square is a portal to its finding. The heat cell re-introduces magnitude-by-color for what is a count of discrete things (and dies the moment two teams need comparing). Words stay in the sub-line; clear teams show nothing, quietly — calm chips make hot ones pop.
5
Agents
The lens's question: one tool each — or a stack?
candidateScaled ratio barthe baseline — bar length = agents/person
ML-Infra1.5agents per person · 31 people
Payments-EU1.3agents per person · 24 people
Core-Banking1.7agents per person · 9 people
Research2.4agents per person · 5 people
Support-Ops1.2agents per person · 12 people
Unassigned1.1agents per identity · 61 unmapped
candidateRatio on the fleet rangethe $/person treatment, reused
ML-Infra1.5agents per person · 31 people
Payments-EU1.3agents per person · 24 people
Core-Banking1.7agents per person · 9 people
Research2.4agents per person · 5 people
Support-Ops1.2agents per person · 12 people
Unassigned1.1agents per identity · 61 unmapped
winnerMicro-histogramx = agents per person (1…5+) · height = people
ML-Infra1.5agents per person · 31 people
Payments-EU1.3agents per person · 24 people
Core-Banking1.7agents per person · 9 people
Research2.4agents per person · 5 people
Support-Ops1.2agents per person · 12 people
Unassigned1.1agents per identity · 61 unmapped
Verdict — micro-histogram. The ratio hides its own question: "everyone runs 2" and "three people run 6" produce the same mean. The histogram is the distribution the ratio erases — ML-Infra's right-shifted mass is a tooling culture, Support-Ops's single tall column is a standard kit. The dot-on-range is honest but answers the wrong question (position, when the question is shape).
Composed
The strip, with every lens wearing its winner
Switch the lens; the micro-form follows. Same chip geometry throughout, every mark still a portal (hover any dot, square, or column). The portals' Auto lens composes on top — it picks the lens per chip, and the form follows the lens for free.
Teams
ML-Infra$4.2k23% of fleet · 31 people
Payments-EU$2.9k16% of fleet · 24 people
Core-Banking$6003% of fleet · 9 people
Research$9005% of fleet · 5 people
Support-Ops$3002% of fleet · 12 people
Unassigned$4.1k22% of fleet · 61 unmapped
every dot one person · the amber chip's cloud is the $4.1k, people-shaped
Principles
Form follows the question
| Lens | Question | Winning form | What it reveals that the bar hid |
|---|---|---|---|
| Spend | how divided + how distributed | micro-swarm | outliers and shape — the two red dots inside Payments-EU's $2.9k; the bar showed one length |
| $ / person | typical or extreme vs fleet | dot on range | position against the median — "2.4×" drawn; the bar only ranked neighbors |
| Active | how much of the team is on | unit dots (≤12) / meter | countable presence — 4 of 9 without arithmetic; the fill implied a continuum that isn't there |
| Risk | one person or systemic | severity glyphs | discreteness — three separate squares are three separate problems; any bar would have summed them |
| Agents | one tool or a stack | micro-histogram | the distribution the mean erases — same 2.3 ratio, opposite cultures |
The pie's obituary — a two-slice pie for a ratio is the dataviz procedure's literal anti-example: a single ratio wants a meter (or countable units). A multi-slice pie for share-of-spend says what the sub-line already says in words, worse. No lens's question survives contact with a pie; none ships. And the test, once more: reveal or don't render — every winning form above carries information the number beside it cannot; every losing form repeated the number in ink.