Overview

You're in a budget meeting. CFO asks: "How many people do we have? How much are we spending? Where are the gaps? Will we hit headcount targets?" You don't know off the top of your head. You go back to HR, spend 2 hours scrambling to pull together numbers, realize they're inconsistent (someone counted people three different ways), and by the time you have answers, the meeting is rescheduled.

This happens because most organizations have people data scattered across multiple systems, HRIS, payroll, hiring platform, performance system, with no single source of truth. Finance thinks headcount is 186. Operations sees 189. HR is still updating the spreadsheet from last month. Leadership makes decisions on incomplete information.

People analytics is where AI creates significant value for the entire organization, not just HR. Finance needs to understand compensation spend and forecasts. Operations needs to forecast headcount for capacity planning. Leadership needs to understand retention trends and hiring progress. Sales wants to know team velocity based on headcount. Everyone needs reliable data at the moment they need it.

Without good analytics, decisions are made with incomplete information. Budget is allocated wrong. Hiring targets are set without understanding attrition. Growth is planned without visibility into team capacity. You end up hiring aggressively while losing people at rates that make growth meaningless. Or you freeze headcount without knowing you're about to lose key engineers.

This lesson covers building dashboards and reporting systems that answer the questions leaders actually ask, give them answers in minutes not hours, and provide the visibility needed to make informed decisions about hiring, compensation, retention, and growth.

Why People Analytics Matters

Most organizations don't have real people analytics. They have a spreadsheet someone updates monthly, or quarterly if you're lucky. Inconsistent definitions plague every report. Is someone in headcount if they're part-time? If they're on unpaid leave? If they're a contractor? If they were hired on the 31st? Different systems answer differently.

Reporting is slow. A simple question like "How many engineers do we have in the Bay Area?" requires pulling data from HRIS, filtering for department, filtering for location, cross-checking with payroll to exclude people on leave, and realizing someone updated their location wrong three months ago. By the time you have the answer, the decision has moved on. So leaders stop asking HR for data and start making guesses.

Forecasting doesn't exist. Nobody knows what headcount will be next quarter. You can't say "At this attrition rate, we'll lose 12 people. At this hiring velocity, we'll replace 8. Net: we'll have 4 fewer people unless we speed hiring." So you plan for growth without understanding the headcount reality. You commit to a revenue target that requires 25 people, but you're only forecasting 18.

Good people analytics solves this: Answers questions in real-time (literally, a leader logs into a dashboard and sees current state). Consistent definitions enforced at the data layer (no more "is this person in headcount?"). Forecasting capability (scenario planning around hiring and attrition). Insights about trends (not just "attrition is 2%," but "attrition is rising in Engineering and we're losing people before 18 months").

AI's value is in: Organizing data from multiple sources (HRIS pulls, payroll data, performance system, hiring platform, exit data, all different systems). Calculating metrics consistently and auditing them for quality. Identifying trends and patterns automatically. Creating forecasts based on historical patterns and stated plans. Flagging anomalies (someone's salary jumped 40%; hire date is in the future; department is blank).

The business impact is huge. You can answer "Will we hit hiring targets?" in 30 seconds instead of 2 hours. You can forecast "Based on attrition trends, we'll have 15 open reqs in Q3" instead of guessing. You can show "Engineering is losing people at 4% per quarter, Sales is losing at 1%. Here's why and here's the cost." You can prove to Finance that compensation spend is aligned with plan. You can demonstrate to the board that promotion rates for underrepresented groups are improving.

Important: Without data, people decisions are made on emotion, politics, and gut feel. With data, they're made on facts. The HR leader who has data wins every meeting.

Why HR Owns This

Analytics is often owned by Finance or a data team. But HR owns this because:

HR has the most detailed people data. You manage the HRIS. You own compensation benchmarking. You track exit reasons. You administer benefits. You run performance reviews. Finance knows spend; HR knows the people behind the spend.

The insights drive HR strategy. If attrition in Engineering is rising, HR needs to respond with retention programs, compensation review, or management coaching. If diversity representation is low at senior levels, HR designs a pipeline. If new hires are leaving before 18 months, HR improves onboarding. The data informs your work.

Trust matters. If Engineering asks "Are we paying competitively?" they want to hear from HR, backed by data. If employees ask "What's promotion velocity around here?" that comes from HR. If the CEO asks "Are we as diverse as we say we are?" HR provides the audited answer.

Practically: Work with Finance and IT to set up the data infrastructure. HR teams maintain it, update definitions, audit quality, and communicate insights.

Core Metrics for Workforce Planning

You need metrics that answer the questions leaders actually ask in board meetings and budget reviews. Pick the core few, track them consistently, expand over time.

Headcount (The Foundation):

Total headcount is simple: count people on payroll as of a specific date (month-end works). But the devil is in the definitions. Does this include contractors? People on unpaid leave? Someone hired on the 31st? Someone who resigned but hasn't left yet? Pick a definition and stick with it. Example: "Active full-time and part-time employees on payroll as of the last day of the month, including people who have resigned but not yet left, excluding contractors and people on unpaid leave longer than 30 days."

Then slice it by what leaders care about: By department (Engineering, Sales, Operations, Marketing, HR). By level (Individual Contributor, Manager, Senior Manager, Director, VP). By location (if multi-office). By tenure (less than 1 year, 1-2 years, 2-5 years, 5+ years). By hire cohort (all people hired in 2023, 2024, etc.). This lets you answer "How many engineers do we have? How many are managers? How many in the Bay Area? How many are new?"

Track month-over-month change. "We had 186 people last month, 189 this month, change of +3." More useful: "We hired 8, lost 5 (attrition), net +3." Trend it over time so you see whether you're growing, stable, or shrinking.

Attrition (The Metric That Matters Most):

Attrition is people leaving. Calculate: (Number of separations in period) / (Average headcount during period). Monthly attrition of 1% means ~12% annualized, which is very low. Monthly attrition of 3% annualized to ~36%, which is crisis territory.

Separate voluntary (people chose to leave) from involuntary (you terminated them). A company with 2% voluntary attrition and 1% involuntary (3% total) is different from one with 1% voluntary and 2% involuntary. One suggests people leave on their own; one suggests you're managing out. This matters for narrative. Boards care about voluntary attrition as a proxy for culture and retention. Involuntary reflects your management and performance management.

Annualize the monthly number. If you lost 4 people this month out of 200, that's 2% for the month, ~24% annualized. Is 24% your target? Probably not. Typical healthy attrition ranges from 10-15% annually depending on industry. Tech is higher. Finance/accounting is lower. Know your target and whether you're above or below.

Slice by department, level, tenure, manager. "Attrition is 2% overall, but 4% in Engineering" is a data point that drives action. "New hires (0-6 months) have 8% attrition; experienced (2+ years) have 0.5%" tells you to improve onboarding. "Attrition under Manager A is 6%; under Manager B is 0.5%" tells you to coach or move Manager A.

Track by exit reason (gathered from exit interviews). "20% of people cite pay, 30% cite growth, 20% cite manager, 15% cite culture, 15% other." This is gold. Pay issues drive different solutions than manager issues.

Hiring (The Intake):

Track open requisitions in different states: Approved (budget is allocated but role not yet posted). Posted (recruitment has begun). Interviewing (candidates in process). Offers out (offer extended, waiting for acceptance). Hired (offer accepted, person starts).

Calculate time-to-fill: from the day a req is approved to the day the person starts. Most companies range 60-120 days depending on level and market. If your time-to-fill for engineers is 180 days, you have a recruiting problem. If it's 30 days, you're either underhiring or hiring quickly (rare).

Compare hiring plan to actual. "We planned to hire 10 engineers this quarter. We've hired 3. We have 2 offers out. At current velocity, we'll hire 6." This triggers questions: recruiting pipeline too small? Hiring bar too high? Market conditions? Compensation not competitive? The data shows the gap; you diagnose the cause.

Compensation (The Spend):

Average salary by level. "Individual Contributors average $85K, Managers average $110K, Directors average $150K." Understand what you're paying for each role.

Salary ranges: min/max/median for each level or role. If your IC range is $60K-$120K but your median is $65K, you're bunched at the bottom. This matters for retention (people don't see room to grow) and external competitiveness.

Compare to market. "Our engineers average $90K; market average is $100K." You're behind. If you're trying to hire from a hot market, this explains your slow time-to-fill. Cost-of-living matters too: $85K in Austin is different than $85K in San Francisco.

Track annual increases and promotion increases. "Average annual increase is 3%; promotion increase is 8%." This tells you how you're managing comp. Low increases with no promotion upside loses people.

Diversity (The Representation):

Gender representation: What % of your company is women, men, non-binary? By level? "Women are 40% of your company but 15% of your leadership." That's a pipeline problem. Track it to show progress. "Two years ago women were 10% of leadership. Now 15%. On pace to hit 25% in three more years."

Racial/ethnic representation. Legally, you only report EEOC data (US companies only report this). But if you want to track, similar metric: representation at each level and trend over time.

Promotion rates by group. "15% of men were promoted; 8% of women." This shows whether promotion is equitable. Healthy pattern: similar promotion rates across groups (within noise).

Pay equity: For the same level and role, are women/minorities paid similarly to majority group? "Female ICs average $88K, male ICs average $90K" might be noise (N=5 each). "Female ICs average $85K, male ICs average $95K" with larger Ns suggests inequity that needs investigation.

Development (The Pipeline):

Promotion rates: How many people get promoted each year? "10% of our 200 people get promoted annually = 20 promotions/year." If it's 2%, people see no upside and leave. If it's 25%, people get promoted too fast and burn out.

Internal mobility: How many people move to a different role (not necessarily a promotion)? "12% of our team moved to a different role this year." This suggests growth opportunity and flexibility. Low mobility (less than 5%) suggests people are stuck.

Tenure distribution: "5% of people have been here less than 1 year. 15% 1-2 years. 25% 2-5 years. 35% 5-10 years. 20% 10+ years." This shape tells you about stability. If it's all under 2 years, you're high-turnover. If it's all 10+, you might be stagnant.

Retention by cohort: "Of people we hired in 2023, how many are still here?" "We hired 50 in 2023. 45 are still here (90% retention). We hired 60 in 2022. 50 are still here (83% retention). We hired 40 in 2021. 35 are still here (88% retention)." This shows hiring quality and onboarding effectiveness.

Tip: Start with headcount, attrition, and hiring. Once you have those solid, add compensation and diversity. Full metrics stack (headcount + attrition + hiring + comp + diversity + development) takes 3-6 months to set up right. Partial dashboard is better than no dashboard.

Dashboard Structure: Design for Your Audience

A good HR dashboard is not one dashboard. It's a layered system where different people see different levels of detail. Your CEO doesn't need to see every line in the employee census; your compensation analyst does. Design for each audience.

Layer 1: Executive Summary (1 page, CEO/CFO/Leadership):

This is what gets shared in monthly all-hands or board prep. One page. 5 minutes to understand current state. Include:
- Total headcount (current, previous month, trend)
- Headcount change (hired, attrition, net)
- Attrition rate (monthly and annualized, vs. target)
- Hiring progress (plan vs. actual year-to-date)
- Open requisitions and average time-to-fill
- One key highlight and one key concern ("Engineering attrition up to 4%, we're investigating manager churn")
- Quarterly forecast (headcount at end of quarter given current hiring/attrition)

The CEO glances at it for 90 seconds. They see: growing? retaining? on hiring plan? healthy? You want it visual: charts, not tables. Red/yellow/green status indicators next to each metric.

Layer 2: By Department (1-2 pages per department leader):

Each VP or department head gets a page showing their slice of the organization. Include:
- Their headcount (current, change, by level)
- Their attrition (rate, trend, by level or tenure)
- Their hiring progress (plan vs. actual, open reqs, time-to-fill)
- Their diversity metrics if they have headcount
- Their average salary by level vs. company average
- Their promotion rates (if applicable)
- Comparison to company average for key metrics ("Attrition in your department is 3.2% vs. company average 2.3%")

This is practical data for the VP. They can answer their own questions: "Do I have enough engineers? Are people staying? Am I paying competitively? How diverse is my team?" They're not reliant on HR to pull reports.

Layer 3: Deep Dives (2+ pages per topic, for HR and analytics):

• 
  Attrition Analysis: By department, by tenure cohort, by manager, by exit reason, trends over time. Show which departments are losing people, when (after how long), and why. Flag high-risk groups. "Engineering is losing people after 18 months (2 people in past month, 3 in past quarter). Exit reason: growth/career development. Action: discuss career paths with Engineering leadership."
• 
  Compensation Analysis: Market comparison (are we ahead or behind for each role?), equity analysis (are similar people paid similarly?), by level and department, progression (how much do people make at each level?), increases over time.
• 
  Diversity Analysis: Representation by level, promotion rates by group, pay equity, trend over time, industry comparison if available. This is for board reporting and compliance.
• 
  Workforce Forecasting: Given hiring plan and expected attrition, what will headcount be at the end of Q2, Q3, Q4? Build multiple scenarios (base case, optimistic, pessimistic). Helps Finance plan budget and Operations plan capacity.
• 
  Manager-Level Attrition: If attrition is high in one department, drill into which managers have high attrition. This is sensitive data (don't share outside of HR and leadership), but critical for retention.

These are reports you update monthly or quarterly, deep-diving on trends, highlighting issues, and recommending actions.

Layer 4: Detailed Data (for troubleshooting, auditing, deep research):

Raw employee data available to HR (with appropriate access controls). Filters: by level, department, location, tenure, start date, salary, compensation changes, promotion history, exit reason. This is for HR analysts to pull cohorts and investigate. "Show me all people hired in 2022 who left in 2024 and their exit reasons." "Show me female ICs and male ICs at level 3 to audit pay equity." Not for casual browsing; for research.

Data Quality: The Foundation That Everything Rests On

All dashboards, all insights, all forecasts depend on clean data. A dashboard full of incorrect numbers is worse than no dashboard; it sends leadership in the wrong direction.

Data Sources (and the Problem):

Employee data lives in multiple systems:
- HRIS (Workday, SuccessFactors, ADP, etc.): Official employee records, name, role, start date, salary, department, location, job level, manager, hire status (active/inactive).
- Payroll (ADP, Gusto, etc.): Salary, tax info, benefits, hours worked (if hourly). This is the source of truth for who's actually on payroll.
- Performance System (Workday, Lattice, etc.): Performance ratings, review dates, promotion history.
- Hiring System (Greenhouse, Lever, etc.): Requisitions, status, time-to-fill, hiring plan.
- Exit Data (from exit interviews, separation documents): When people left, why they left, how they left (voluntary vs. involuntary).
- Benefits System (Guidepoint, ADP, etc.): Who's enrolled in benefits, which indicates who's active.

The Challenge: These systems don't talk to each other. Definitions of "active employee" differ wildly. Is someone in headcount if they're on unpaid leave? If they're a contractor? If they resigned but haven't left yet? If they're on sabbatical? Different systems answer differently.

Example: I worked with a company of 200 people. Finance's headcount: 186 (only on payroll). HR's headcount: 195 (includes contractors and people on unpaid leave). HRIS active employee count: 203 (includes people who haven't started yet). Same company, three different answers. Finance thought HR was sloppy. HR thought Finance was excluding people. Neither trusted the other's numbers.

Solution: Data Integration and Standardization

Create a single normalized dataset that pulls from all sources:
1. Extract employee records from HRIS (export weekly or daily)
2. Extract payroll records from payroll system (match to HRIS by ID)
3. Extract hiring/requisition data from recruiting system
4. Extract exit data from HR system or exit interview form
5. Merge all data into one master employee dataset
6. Apply standardized definitions (e.g., "headcount = active on payroll as of month-end, excluding unpaid leave > 30 days")
7. Run data quality checks (flag inconsistencies, gaps, unusual values)
8. Create audit trail (document all changes, source of each data point)
9. Update automatically (weekly or daily)

Data Quality Checks AI Can Do:

• Consistency checks: Someone marked as "active" in HRIS but last paycheck was 6 months ago? Flag it. Start date in future? Flag it. Hire date later than first paycheck date? Flag it.
  - Completeness checks: Required fields (name, role, department, salary) are blank? Flag it. Salary is $0? Flag it. Department is "TBD"? Flag it.
  - Range checks: Salary is $10K for a director? Flag it. Tenure distribution has someone with 50 years service? Flag it. Age calculation is negative?
  - Outlier detection: Salary jumped 50% month-over-month? Flag for promotion/role change or data error. Attrition for a cohort is 80% when average is 15%? Investigate.
  - Matching: Payroll shows person X at salary $120K; HRIS shows $130K. Which is right? Flag the gap, investigate in source system.

Definitions Document (Critical):

Create a data dictionary. Write it down. Update it annually. Share with all stakeholders.

Example:

HEADCOUNT DEFINITION:
Count: All employees with status = "Active" or "On Leave" in HRIS
Exclude: Contractors, temporary workers, people on unpaid leave > 30 days, people with end date <= today
As of: Last day of the month, 5:00pm Pacific time

ATTRITION DEFINITION:
Count: All employees who transitioned from "Active" to "Inactive" during the period
Include: Voluntary resignations and involuntary terminations
Exclude: People who moved to a different entity or subsidiary (internal transfers)
Calculation: (# of separations) / (average headcount during period)
Annualization: (monthly rate * 12)

LEVEL DEFINITION:
- Individual Contributor (IC): Individual contributor roles, no direct reports
- Manager (MGR): Management role, 1-3 direct reports or team lead
- Senior Manager (SM): 4+ direct reports, manages managers
- Director: Multiple teams, organization-level impact
- VP: P&L responsibility, board-level visibility

LOCATION DEFINITION:
Based on employee's work location in HRIS or primary office assignment.
If hybrid/remote and no office assignment, use headquarters location.

Document everything. Your future self will thank you when someone asks "Why does this number not match the board report from 2 months ago?"

Important: Bad data will destroy trust faster than no data. Invest in data quality upfront. Audit quarterly. When you find errors, fix them, document what was wrong and why.

Forecasting: Turning Data Into Decisions

One of the most valuable things a dashboard can do: forecast what your headcount will be in 3-6 months and what that means for operations, budget, and hiring.

Most companies don't forecast headcount. They set an annual hiring plan ("hire 50 people") and an annual budget ("headcount budget is 250"). Then they're surprised when attrition is higher than expected, they're at 235 headcount by Q3 instead of 250, and they have to cut hiring mid-year or trim budget.

The Forecast Model:

Starting headcount + Expected hires - Expected attrition = Future headcount

Example: Today is March 31. Current headcount is 200.

• Q2 hiring plan: 8 people (based on approved reqs)
  - Q3 hiring plan: 10 people
  - Q4 hiring plan: 5 people
  -
  Total planned hires: 23 people
• 
  Historical attrition: 2% per quarter (4 people per quarter at current size)
• Q2 expected attrition: 4 people (actually, 2% of 204 = ~4)
  - Q3 expected attrition: 4 people (2% of 214)
  - Q4 expected attrition: 4 people (2% of 220)
  - Total expected attrition: 12 people

Forecast: End of year headcount = 200 + 23 - 12 = 211

This forecast drives decisions:
- Facilities: "By EOY we'll have 211 people. Our current office fits 250. We're OK. We don't need to expand."
- Compensation budget: "We'll hire 23 people. Average salary is $85K. Fully loaded cost ~$120K per person. 23 * $120K = $2.76M additional compensation expense in the second half of the year. We need to budget for that."
- Recruiting: "We need to hire 23 people and expect 12 to leave. We need a recruiting pipeline of 35 candidates at various stages."
- Product roadmap: "Headcount grows from 200 to 211. That's 5.5% growth. Product roadmap should assume 5-6% velocity increase, not 10%."

Scenario Planning:

Create three scenarios:

Base Case (most likely):
- Hiring: On plan (23 people)
- Attrition: Historical rate (2% per quarter, 12 people)
- Result: 211 headcount

Optimistic (if things go well):
- Hiring: Above plan (30 people, recruiting pipeline is strong, offers accepted quickly)
- Attrition: Below historical (1.5% per quarter, 9 people, retention programs working)
- Result: 220 headcount

Pessimistic (if things go poorly):
- Hiring: Below plan (15 people, market is competitive, time-to-fill extends)
- Attrition: Above historical (3% per quarter, 15 people, people leaving for better opportunities)
- Result: 200 headcount (flat)

Leadership now has a range. "We'll be between 200 and 220, most likely 211." They can plan around that. If the pessimistic case is 200 and that breaks the business model, the hiring problem becomes urgent. If the optimistic case is 220 and that strains facilities/budget, you need to be ready.

Updating the Forecast:

Update monthly as actual hires and attrition unfold. "We planned 8 hires in Q2. It's mid-June and we've made 3 offers. Recruiting says 2 more by July. That's 5 by EOQ2, 3 short of plan. Adjust the forecast down." You spot problems early and can intervene (speed up recruiting, adjust hiring targets, reduce spending elsewhere).

Communicating Forecasts:

Charts work better than tables. Show headcount over time with a base case line, optimistic and pessimistic bands around it, and current actual headcount. Leadership sees the range and the trend instantly.

Workflow Diagram: Building a People Dashboard

STEP 1: Define Questions
- What do leaders want to know?
- Executive summary level? Details?
- Frequency? (Monthly? Quarterly?)

STEP 2: Identify Data Sources
- HRIS (employee records)
- Payroll (compensation)
- Performance system (ratings)
- Exit data
- Hiring system

STEP 3: Data Integration & Cleaning
- Ingest from all sources
- Standardize definitions
- Clean for quality issues
- Create normalized dataset

STEP 4: Metric Calculation
- Calculate headcount by slicing (level, dept, location)
- Calculate attrition (monthly, annualized, by cohort)
- Calculate time-to-fill, hiring plan vs. actual
- Calculate compensation metrics
- Calculate diversity metrics

STEP 5: Forecasting
- Plug assumptions into model
- Generate forecast scenarios
- Update quarterly

STEP 6: Dashboard Creation
- Executive summary (1 page)
- By-department breakdown
- Deep dives (attrition, comp, diversity)
- Detailed data access

STEP 7: Monthly Update & Reporting
- Auto-update from data sources
- Generate reports
- Highlight changes and alerts
- Share with leadership

Metrics to Include: Attrition Deep Dive Example

ATTRITION ANALYSIS

Company Attrition:
- Current month: 2.5% (turnover rate)
- Year-to-date: 2.1%
- Annualized forecast: 2.3%
- Target: 2.0%

Voluntary vs. Involuntary:
- Voluntary (chose to leave): 1.8%
- Involuntary (terminated): 0.5%

By Department:
- Engineering: 3.2% (above average)
- Sales: 2.1% (on average)
- Operations: 1.0% (low)

By Tenure:
- 0-6 months: 5% (new hires, some don't work out)
- 6-12 months: 2.5%
- 1-2 years: 2.0%
- 2-5 years: 1.5%
- 5+ years: 0.5%

By Manager:
- Manager A: 4.5% (high, flag)
- Manager B: 1.0% (low, good)
- Manager C: 2.1% (average)

By Exit Reason:
- Pay: 35%
- Growth: 25%
- Manager: 20%
- Culture: 10%
- Other: 10%

Trends:
- Engineering attrition is rising (was 2.5% 6 months ago, now 3.2%)
- New hire attrition is high (5%), suggest onboarding improvement
- Attrition under Manager A is high, suggest coaching or replacement

Actions:
- Engineering: compensation review (pay is exit reason)
- New hires: improve onboarding
- Manager A: performance coaching

This deep dive shows not just "attrition is 2.3%" but where the problems are and what to do about them.

Before AI vs With AI

OLD (Manual reporting):
- Monday morning: Finance asks "How many people do we have?"
- HR pulls out spreadsheet, updates it (2 hours)
- Some data is from last month, some from last week
- Numbers are approximate
- Finance is frustrated with inconsistency
- Forecasting: "We'll probably hire 10 people and lose 3, so let's plan for 187 headcount" (guessing)

NEW (AI dashboard):
- Finance logs into dashboard
- Real-time: 186 people as of 11am today
- Breakdown by department, level, location, all accurate
- Attrition trend visible: "Trending up from 2.0% last quarter to 2.3%"
- Forecast: Based on hiring plan and historical attrition, expect 212 people by end of Q4
- Historical data: See trends over time (attrition by month, headcount by quarter)
- Drills: Click to see details (which departments are hiring, which are losing people, why)

Finance has confidence in the numbers. Decisions are made based on reliable data.

What to Do Monday Morning

Step 1: Audit your current state (2 hours)

Create a simple spreadsheet. Document:
- Where employee data lives: HRIS system? Payroll system? Spreadsheets? Who owns each?
- Are systems integrated or manual hand-offs? (If manual, that's your first problem.)
- What reports currently exist? (Headcount? Attrition? Hiring? Who uses them? Are they trusted?)
- What questions does leadership ask that you can't answer quickly? (List them.)
- Who in IT/Finance/Analytics could partner on this?

Step 2: Define what leadership actually needs (1 hour)

Don't assume. Ask. Schedule quick calls with:
- CFO: "For monthly financial planning, what people metrics would help?" (Usually: headcount trend, hiring plan vs. actual, compensation spend forecast)
- CEO: "For board meetings, what would you want to show?" (Usually: growth rate, attrition, diversity, hiring progress)
- Your CHRO: "What are your top concerns?" (Usually: retention, pipeline, pay equity)
- 1-2 department heads: "What do you want to know about your team?" (Usually: headcount, attrition, hiring progress, pay competitiveness)

Write down the top 10 questions. Your dashboard needs to answer these.

Step 3: Start with one metric (2-4 hours)

Pick headcount. It's founditive. Create a simple view:
- Employee list: Name, Department, Level, Start Date, Salary, Status (Active/Inactive)
- Auto-calculate: Total headcount, headcount by department, headcount by level
- Manual or formula: Month-over-month change (this month vs. last month)

If you have HRIS export, pull from there. If not, ask Finance for the payroll headcount list and use that. You want one source of truth.

Test it against what you know. "Do we have 200 people? Does that match what Finance says?" If not, investigate. Don't publish until the base number is right.

Step 4: Add attrition (2 hours)

Pull exit data. You need: Who left? When did they leave? Did they leave voluntarily or were they terminated?

Calculate: (# who left this month) / (average headcount) = monthly attrition rate. Annualize it (* 12).

Compare to your target. If target is 15% annual and you're running 18%, you're slightly high. If you're running 25%, you have a problem.

Add context: By department, by tenure. "Overall attrition is 2%, but Sales is 3.5% and Engineering is 1.2%." Now you see where to focus.

Step 5: Create a simple one-page report (1 hour)

Design it as a one-pager:
- Top: Current headcount, month-over-month change, trend line (growing/stable/shrinking)
- Middle: Attrition rate (monthly, annualized), vs. target, by department
- Bottom: Open reqs, time-to-fill, hiring plan vs. actual YTD

Charts, not tables. Color-code status (green = on track, yellow = watch, red = problem).

Share with your CHRO and CFO. "Is this what you need? What's missing?" Iterate based on feedback.

Step 6: Automate the monthly refresh (varies)

Don't manually update this. It won't be current.
- If you have HRIS and payroll integrated: Export monthly, update dashboard from export. Set calendar reminder for first Friday of each month.
- If manual: Talk to IT about automating HRIS export. This is 2-4 weeks of work but saves 4 hours per month forever.
- If data tools (Tableau, Looker, etc.): Work with IT to build a connected dashboard. This is 4-6 weeks of work but gives you real-time updates and interactive filters.

Step 7: Build in feedback loops (ongoing)

Each month when you share: "Was this useful? What should we add or change?"

Track what people ask for. "Finance keeps asking 'How much will compensation spend be by EOY?' Add that to the next version."

Expand methodically. Attrition deep dives after 1 month. Compensation analysis after 2 months. Forecasting after 3 months. Diversity metrics after 4 months.

Step 8: Layer your reporting (month 2+)

Once the one-pager is solid, create:
- One-page executive summary for all-hands or board prep
- Department breakdowns for department heads
- Deep dive on attrition (why people leave, which departments, which roles)
- Compensation analysis (how we pay vs. market, pay equity)

Each document is targeted to its audience and answers their specific questions.

Tip: The difference between good HR and great HR is data. Every other HR leader is guessing. You'll have evidence. Invest in this early and it compounds.

Key Takeaways

Recognize that people data has value across the organization, not just HR. Finance needs headcount and compensation forecasts to plan budgets. Operations needs staffing projections for capacity planning. Sales needs hiring velocity to forecast revenue. Leadership needs retention trends and diversity metrics for board reporting and strategy. When Finance can log into a dashboard and see "If we hire 25 and lose 12, we'll be at 213 headcount," they don't need HR anymore to pull reports. You become a partner.

Invest in data quality upfront, not downstream. Systems don't talk to each other by default. Data definitions vary. You need to create a normalized dataset with standardized definitions, run quality checks, flag inconsistencies, and maintain it. This is boring work. No one gets excited about data cleaning. But garbage data produces garbage insights and erodes trust faster than anything. Spend 4 weeks on data quality before you build your first dashboard.

Define every metric and document it. "Headcount" means different things to different people. Write it down: "Active employees on payroll as of month-end, excluding people on unpaid leave > 30 days, including people on paid leave." Use that definition consistently. When someone says "But I counted 215 people," you can say "That includes contractors; headcount is 203." No argument. Data dictionary ends discussions.

Build dashboards in layers for different audiences. The CEO wants one page. The VP of Finance wants more detail. The HR analyst wants raw data. Design for each audience. One executive summary. Departmental breakdowns for department heads. Deep dives on specific topics (attrition, comp, diversity). Detailed data for research. More layers, more value to more people.

Create forecasts not just reports. "Headcount is 200" is a data point. "Headcount is 200, will be 211 by EOY if we hire on plan and attrition stays at current rate" is actionable. Forecasting gives leaders confidence in planning. They know the range (pessimistic to optimistic). They spot risks early (if pessimistic case breaks the model, hiring becomes urgent).

Automate the updates so data stays current. Manual dashboards die. Someone forgets to update, the data becomes stale, people stop trusting it. If possible, integrate with your HRIS and payroll system so data flows automatically. If not possible, set a calendar reminder for the same day each month (I use first Friday) and make it a 30-minute task. Current data builds trust. Stale data kills trust.

Validate data quality regularly and audit for anomalies. Even good systems have errors. Someone changes their title and it doesn't sync. Hire date is off by a month. Salary is missing. AI can flag these automatically (salary is $0, start date in future, no department assigned). You audit monthly. Fix issues in source systems. Document the fixes. Trust grows.

Failure Scenarios: What Goes Wrong

Scenario 1: Data Quality Crisis

You launch a dashboard. Everything looks great. Then the CEO asks "Why does your headcount number not match Finance?" You check HRIS, it says 200. Finance payroll, it says 195. Someone in IT hasn't updated the sync in 2 months. You lose credibility immediately. Now you're explaining why you weren't monitoring data quality instead of answering business questions.

How to avoid: Monthly data validation. Audit HRIS headcount against payroll headcount. They should match (or you understand why they don't). Document any exceptions. If they drift, investigate and fix in the source system.

Scenario 2: No One Uses It

You build a beautiful dashboard with 15 metrics. You send it to leaders. Crickets. They still ask HR "Can you send me a headcount report?" because the dashboard doesn't answer their specific question.

How to avoid: Build for a specific audience first. Start with what your CFO actually asks for (not what you think Finance needs). Make that one metric so useful and accurate that they start relying on it. Then expand. Ask before building. "What would make your job easier?" Then build that.

Scenario 3: Forecasting Goes Wrong

You forecast 220 headcount by EOY. You budget for 220. Then attrition spikes to 4% per quarter and you're at 180. Or hiring slows and you only hit 200. Now you have budget for people you don't have and can't hire.

How to avoid: Build multiple scenarios (base/optimistic/pessimistic). Update the forecast monthly as reality unfolds. If by August you're tracking to pessimistic case, flag it and adjust. Don't wait for December to be surprised.

Scenario 4: Definitions Keep Changing

"Is this person in headcount?" "Did they leave voluntarily or involuntarily?" Different people answer differently. Your November attrition is 2.1%. December someone changes how they calculate it. Now December is 2.9% but it's not a real change, just a definition change. Trust dies.

How to avoid: Write definitions down. Share widely. Make them mandatory. Review annually. When something must change, document the change and note historical data was recalculated (if applicable). Be explicit: "This month we changed the definition of 'attrition' to exclude people on unpaid leave. We recalculated historical data accordingly."

Scenario 5: Data Becomes Stale

You launch a dashboard in January. It's great. By June, no one updates it. The data is 5 months old. Finance sees May headcount on the dashboard and asks about June. You don't have it updated. They stop checking.

How to avoid: Automation. Set up a monthly refresh on the first Friday. Make it automated if possible. If not automated, make it a ritual. Same day, same time every month. 30 minutes. Non-negotiable. Calendar reminder. The moment data goes stale, you've lost.

Before/After Comparison: Manual vs. Automated

BEFORE (Manual Reporting):

Monday: Finance emails HR: "Need headcount update for budget meeting tomorrow."
Monday afternoon: HR scrambles, exports from HRIS (takes an hour to figure out the filter), cross-checks with payroll (they don't match), picks one source, creates a one-off spreadsheet with headcount by department.
Tuesday: Finance gets the number, they have 2 hours to use it for a board prep.
Finance says "This doesn't match what IT said. Are we really 186?" HR says "Maybe 183, depends how you count part-time." Finance doesn't trust it and makes the assumption instead of using data.
Friday: Someone asks "What was our headcount again?" No one knows. Nothing was documented.

Next month: Same story. Hours lost, no consistency, no trust.

AFTER (Dashboard):

First of month: Automated data pull from HRIS/payroll at 4am. Dashboard updates. AI flags any inconsistencies (someone marked active but no paycheck; hire date in future). HR reviews, fixes, documents.

Finance logs in, sees: Headcount was 186 last month, 189 this month, up 3. Breakdown by department. Vs. hiring plan. Vs. last year. All consistent.

CEO logs in for monthly review, sees: Headcount trend, attrition, hiring progress, forecast to EOY.

Department heads log in, see: Their team breakdown, their attrition, their hiring progress.

Data is consistent. Current. Trustworthy. Decisions are made on facts, not guesses.

Year-end: Finance has 12 months of consistent data showing headcount growth, attrition trends, hiring effectiveness. CFO presents to board with confidence.

FAQ

Q: Should employees be able to see this data?

A: Transparency is good, but be selective. Employees benefit from seeing: company headcount and hiring status (builds confidence), promotion rates and tenure distribution (shows career trajectory), diversity representation by level (demonstrates commitment to inclusion). Don't share: individual salaries (unless you're doing transparent compensation), performance ratings (privacy), detailed demographic data (privacy risk), or manager-level attrition (sensitive). Some companies publish "Here's our promotion rate by level and gender. Here's our commitment to improvement." This builds trust. Others keep everything closed. No one approach is right, but be intentional.

Q: How often should we update the dashboard?

A: Monthly is standard for operational decisions. Set a rhythm (first Friday of each month) and stick to it. Quarterly is fine if you don't have the data infrastructure, but you lose velocity. Real-time (daily) is nice if you have integrated systems but often unnecessary. Monthly gives you close enough accuracy for decision-making while keeping the update effort manageable. If you're updating quarterly or less frequently, you're already out of date.

Q: What if our data is dirty right now?

A: You have two paths: (1) Clean first, then build dashboard. Audit HRIS for 2-4 weeks, identify errors, fix them in the source system, then launch the dashboard clean. Or (2) Build dashboard now, flag quality issues, and clean incrementally. Most companies can't wait 4 weeks, so do (2): Launch with caveat that you're still cleaning data. "Headcount data is current as of today but we're still validating some hire dates and department assignments." Document issues as you find them. This maintains momentum while building trust through transparency.

Q: Should we have headcount targets?

A: Yes, absolutely. Finance should set hiring targets quarterly or annually. HR should track actual vs. target monthly. "We planned 5 hires in March, made 3 offers, hired 1 person." This variance, offers not converting, recruiting pipeline too small, hiring manager indecision, tells you what to fix. Targets create accountability. Without them, there's no baseline for "are we on track?"

Q: How do we handle role classifications and levels?

A: Document them explicitly. Don't assume "manager" means the same thing across departments. Write it down: "Individual Contributor (IC): No direct reports. Manager (MGR): 1-3 direct reports. Senior Manager (SM): 4+ direct reports or manages managers. Director: Multiple teams, org-level impact." Use these consistently. When someone gets promoted from IC to Manager, update their level in HRIS. When you're comparing compensation across departments, levels matter. You need to be comparing apples to apples. If levels drift, your analysis is wrong.

What's Next

You now have the foundation: headcount dashboards, attrition tracking, hiring visibility, and forecasting. The next lesson covers engagement survey analysis, taking the quantitative data and combining it with employee feedback to understand the "why" behind your numbers. Why is attrition high? Engagement survey data often tells the story. Why is promotion velocity low? Surveys reveal barriers. Analytics + insights = action.