Building the first dashboard

1. CURRENT cost-per-unit  (big number)
2. TREND over time         (line chart, 12 months typical)
3. HISTORICAL comparison    (this quarter vs last)
4. YEAR-OVER-YEAR           (if data available)
5. ANNOTATIONS              (notes on key events affecting the metric)

[BIG NUMBER]                              [TREND SPARKLINE]
$11.83 / paying user                       ↘ down from $13.20 (3 mo ago)

[LINE CHART: 12 months]
   $14 -│      ████████
   $13 -│              ██████
   $12 -│                    ████████
   $11 -│                            █████
   $10 -│
   $9  -│
       └────────────────────────────────────
       Jun  Jul  Aug  Sep  Oct  Nov  Dec  Jan  Feb  Mar  Apr  May

[COMPARISON]
  This quarter vs same quarter last year:  improving 8%
  Quarter-over-quarter:                     improving ~10%

[ANNOTATIONS]
  Mar 15: launched new feature (cost up briefly)
  Apr 30: optimized data pipeline (cost down 6%)
  May 12: scaled to support EU expansion (small increase expected)

DATA REFRESH:
  Daily: dashboard reflects yesterday's data (24-hour billing lag)

REVIEW CADENCE:
  Weekly: review at team meeting (quick scan; flag anomalies)
  Monthly: detailed analysis with finance + leadership
  Quarterly: deep variance analysis; re-baseline forecast

KEY EVENTS to annotate:
  Major feature launches            (cost spike during ramp-up)
  Cost optimization sprints          (cost drop)
  Scaling events                     (Black Friday, EU expansion)
  Acquired company integrations      (one-time noise)
  Pricing model changes              (rate-card or commitment renewals)
  Methodology changes                 (numerator scope revisions)

WITHOUT annotations:
  Six months later, no one remembers why cost spiked in March
  Future-self misreads the trend
  Conversations stuck on "what happened in week 12?"

WITH annotations:
  Trend is interpretable years later
  Lessons captured in context

WEEK 1:     Set up dashboard; first data populates
WEEK 2-3:   Team starts reviewing in weekly meetings
WEEK 4-8:   Discussion patterns emerge:
              "Cost-per-MAU is up: investigate"
              "Why did cost drop in week 6?"
              "Annotation explains the drop"

WEEK 12+:   Mature usage; routine review
              Team trusts the number; uses it for decisions
              Annotations accumulate; trends become richer

INTERNAL AUDIENCES:
  Engineering: cost trend awareness in team metrics
  Finance: unit economics for financial reporting
  Leadership: business efficiency narrative
  Product: feature ROI per user

EXTERNAL AUDIENCES (some companies report externally):
  Investors: unit economics as efficiency proof
  Board: quarterly business efficiency
  Analysts: comparable to peer benchmarks (if disclosed)

CAUTION:
  External metrics need careful definition + consistency
  Once published externally, the definition is hard to change
  without explanation

MISTAKE                                       FIX
──────────────────────────────────────────────────────────────────
Too many metrics on one dashboard              Pick one primary;
(visual overload)                              demote others to drill

No annotations (trends unexplained)            Annotate events as
                                              they happen

Bad axis scale (small variations look           Use absolute axis
huge or vice versa)                            range matched to value

No context (just current number)               Add comparison +
                                              trend

Updated weekly but reviewed never              Adoption ritual missing;
                                              build the review cadence

DASHBOARD: "Acme: Unit Economics Q2"

BIG NUMBER:
  Cost-per-paying-user: $11.83
  Q-o-Q change: -9% (improving)
  Y-o-Y change: -14% (improving)

LINE CHART (12 months):
  Trend visible: descending from $13.50 (Jun) to $11.83 (May)
  Two visible bumps:
    Feb (data pipeline migration)
    Apr (brief spike during EU launch)

COMPARISON BLOCK:
  vs Q1 2026:    -9% (improving)
  vs Q2 2025:    -14% (improving)
  vs forecast:   -3% (better than forecast: wins)

PER-TEAM BREAKDOWN (chart):
  Platform team:    $4.20 / paying user
  Product team:      $5.10
  Data team:         $1.80 (heavy ETL but small surface)
  ML team:           $0.73 (small inference cost; large user base)

ANNOTATIONS (5 visible):
  Feb 8:   Data pipeline migration (one-time +$0.30)
  Mar 15:  Auto-tagger campaign (+5% coverage, no cost change)
  Apr 1:   Q1 optimization completed (-$0.80)
  Apr 18:  EU expansion launch (+$0.40, expected)
  May 5:   Auto-rem rule activated (-$0.20)

DISCUSSION at monthly review:
  "Trend is positive; Q1 optimization sustained."
  "Investigate why ml-team's cost-per-user is so low:
    can we apply the pattern elsewhere?"
  "Q2 target: cost-per-paying-user < $11."

OUTCOME: dashboard is used weekly; annotations are kept current;
team is making decisions from the data.

PRIMARY METRIC (from L1-L2):    $___ per __________

LAYOUT elements you want:
  □ Big number with delta
  □ 12-month trend line
  □ Quarter-over-quarter comparison
  □ Year-over-year comparison
  □ Per-team breakdown (if relevant)
  □ Annotations

REVIEW CADENCE:
  Daily / weekly / monthly: __________
  Owner of the review: __________
  Distribution (who sees it): __________

ANNOTATIONS to seed (events you remember from past 3 months):
  __________________________________________________________
  __________________________________________________________
  __________________________________________________________

FIRST REVIEW MEETING (calendar date):    __________