Why this matters
As a BI Analyst, you turn raw data into decisions. KPI hierarchy and information architecture ensure leaders see the right numbers first, understand what drives them, and can drill into details without getting lost. In real work, this looks like:
- Prioritizing KPIs for an executive dashboard so it fits on one screen and answers: Are we on track?
- Designing a drill path from a top KPI (e.g., Revenue) down to drivers (Traffic, Conversion, AOV) and diagnostics (by segment, by channel, by time).
- Organizing pages, filters, and interactions so users don't need training to use your dashboard.
Progress note: The quick test is available to everyone. If you log in, your progress will be saved.
Concept explained simply
KPI hierarchy is a tree from outcomes to drivers:
- Goal/North Star: The business outcome (e.g., Net Revenue).
- Outcome KPIs: What success looks like (e.g., Monthly Revenue, Retention).
- Driver metrics: What directly moves the outcomes (e.g., Conversion Rate, Average Order Value).
- Diagnostic metrics: Where to look when drivers move (e.g., by device, region, product, campaign).
Information architecture (IA) is how you structure this in your dashboard:
- Pages and sections (Overview, Drivers, Diagnostics).
- Navigation and drill paths (click to drill-through to detail).
- Layout and hierarchy on each page (what sits top-left, what follows).
Mental model
Think "pyramid and tree":
- Pyramid = page layout: most critical KPI top-left, context around it, details below.
- Tree = drill path: Outcome → Drivers → Diagnostics → Records.
Visualization checklist to keep nearby
- Top KPI visible without scrolling
- Drivers grouped next to their outcome
- Consistent time grain (with clear labels)
- Filters are obvious and do not hide key values
- Clear units and definitions for every KPI
Worked examples
1) E-commerce revenue
KPI hierarchy:
- Outcome: Monthly Revenue
- Drivers: Sessions Ă— Conversion Rate Ă— Average Order Value
- Diagnostics: by Channel, Device, Region, Product Category; Funnel steps (Product View → Add to Cart → Checkout → Purchase)
Information architecture:
- Overview: Revenue (current vs target), trend, variance; mini-cards for Sessions, CVR, AOV
- Drivers page: Decomposition waterfall; trends for Sessions, CVR, AOV
- Diagnostics: Breakdown by channel/device; funnel with drop-off; Top categories table
Why this works
Leaders see if revenue is on target first. If not, the waterfall shows which driver changed. Diagnostics show where to act (e.g., mobile CVR drop).
2) SaaS retention
KPI hierarchy:
- Outcome: Net Revenue Retention (NRR)
- Drivers: Expansion, Contraction, Churn; New MRR (for growth views)
- Diagnostics: by Plan, Cohort, Segment; Leading indicators: Activation, Feature adoption
Information architecture:
- Overview: NRR gauge with trend; expansion/churn stacked bar; Active users
- Drivers: Cohort retention heatmap; churn reasons table
- Diagnostics: Feature adoption by segment; drill to account list
3) Operations/call center
KPI hierarchy:
- Outcome: Customer Satisfaction (CSAT) and Service Level
- Drivers: Average Speed of Answer, First Contact Resolution, Agent Occupancy
- Diagnostics: Queue by hour, Agent-level metrics, Topic types
Information architecture:
- Overview: CSAT and Service Level scorecards; trend
- Drivers: ASA and FCR trends with thresholds
- Diagnostics: Heatmap by hour, topic distribution, agent table
Design steps you can reuse
- Clarify outcomes: Write one sentence: "Success means X by Y date." Extract 1–3 outcome KPIs.
- Map drivers: Break the outcome into mathematical components (e.g., Revenue = Sessions Ă— CVR Ă— AOV).
- List diagnostics: Dimensions you’ll slice by when drivers move (channel, region, device, product, cohort).
- Choose time grain: Keep it consistent per page (daily/weekly/monthly). Label it.
- Lay out the overview: Top-left: primary KPI and target. Top-right: time filter. Middle: driver cards. Bottom: breakdowns.
- Define drill paths: From KPI → driver page → diagnostic page → record table.
- Standardize labels: Definitions, units, and thresholds near the charts.
Helpful layout patterns
- F-pattern reading: 1) Top-left main KPI, 2) Across the top drivers, 3) Down the left details
- Compare vs target and vs previous period side by side
- Use small multiples for consistent comparisons
Common mistakes and self-check
- Too many top KPIs: More than 3 at the top creates noise. Self-check: Can a new user state the #1 KPI in 5 seconds?
- Mixed grains: Daily and monthly trends on the same chart confuse. Self-check: Does every chart on a page share a grain label?
- Unclear definitions: KPI names without definitions cause disputes. Self-check: Hover/legend/footnote defines each KPI.
- Drill dead-ends: Clicks go nowhere. Self-check: Every summary has a linked detail view.
- Filter overload: Too many filters or hidden defaults. Self-check: Keep 3–5 high-impact filters visible; explain defaults.
Exercises
Do these to cement the skill. You can compare with the solutions below each exercise. In the quick test, anyone can attempt; logging in will save your progress.
Exercise 1 — Build a KPI tree
Pick a business: subscription video app. Define the KPI hierarchy and highlight which drivers you’d monitor weekly.
- 1–2 outcome KPIs
- 3–5 drivers (mathematical where possible)
- Diagnostic dimensions
- Weekly monitoring focus
Show solution
Outcome KPIs: Net Revenue, Monthly Active Subscribers
Drivers: New Subs, Reactivations, Upgrades, Downgrades, Churn. Net Revenue = ARPU Ă— Active Subs
Diagnostics: Plan tier, Device, Region, Acquisition channel, Content genre
Weekly focus: Sign-ups, Trials → Paid conversion, Early churn (week 1–4), Content engagement
Exercise 2 — Design the information architecture
For the same app, sketch the dashboard IA: pages, top-of-page layout, and drill paths from the main KPI.
- Overview layout (top KPI, drivers, key breakdown)
- Drivers page content
- Diagnostics page content
- Drill path to records
Show solution
Pages: Overview, Drivers, Diagnostics, Records
Overview: Top-left: Net Revenue vs target; top-right: time filter; center cards: Active Subs, ARPU, Churn; bottom: revenue by plan and by region
Drivers: Decomposition of Net Revenue to ARPU and Active Subs; trends for sign-ups, upgrades, churn
Diagnostics: Heatmap by content genre Ă— region; device breakdown; cohort retention
Drill path: Net Revenue → Drivers page → Diagnostics by plan → Record table of subscriber accounts
Mini tasks
- Rename a vague KPI name like "Users" to a precise one like "Monthly Active Subscribers (MAS)." Add its definition.
- Choose a single time grain for your overview and label each chart with it.
- Limit top KPIs to three: primary outcome, secondary outcome, leading indicator.
Learning path
- Define business outcomes and success targets.
- Map KPI hierarchy (outcomes → drivers → diagnostics).
- Design dashboard IA (pages, layout, drill paths).
- Prototype with dummy data to test flows.
- Validate with a stakeholder and refine.
Who this is for & prerequisites
Who: BI Analysts, data-savvy PMs, analytics engineers supporting dashboards.
Prerequisites: Basic KPI concepts, time series literacy, ability to read simple funnel/decomposition charts.
Practical projects
- Take an existing busy dashboard and reduce it to 3 top KPIs, 3 drivers, and 3 diagnostics. Document before/after.
- Create a decomposition waterfall for your primary KPI and connect it to a driver page.
- Ship a self-serve diagnostics page with 3 filter dimensions and a drill-to-records table.
Next steps
- Apply the checklist to your current dashboard this week.
- Present the KPI tree to a stakeholder and confirm the top KPI and targets.
- Take the quick test below to check your understanding.
Mini challenge
Your exec says: "Revenue is flat, traffic is up." In one sentence, say what you’ll check next and which page you’ll open first in your IA. Then outline the drill path you’ll take.