# OKR Baseline Worksheet

> **Formats:** Markdown (canonical) | DOCX | PDF | Miro/FigJam
> **Updated:** 2026-03-22
> **License:** CC BY 4.0 -- Kate Makrigiannis / k8mak.com

A team worksheet for establishing baselines for your Key Results before you set targets. You can't measure progress if you don't know where you started.

## When to use this

You've written your OKRs for the quarter and now need to ground each Key Result in actual data. This worksheet helps your team brainstorm data sources, vote on the most reliable ones, and document the baseline values that make your targets meaningful. Run it in 30-45 minutes with the team that owns the OKRs.

---

## Process

### Step 1: List your Key Results

Write each Key Result on the board. For each one, note what you'd need to measure.

| Key Result | What we need to measure | Current baseline (if known) |
|---|---|---|
| (KR text) | (Metric description) | (Value or "unknown") |

**Example:**

| Key Result | What we need to measure | Current baseline |
|---|---|---|
| Reduce support call volume by 30% | Monthly support calls related to billing and claims | Unknown |
| Increase online payment adoption to 40% | % of patients paying bills online vs. phone/mail | ~22% (Q4 estimate) |
| Achieve 90% task completion rate for self-service flows | Completion rate for top 5 self-service tasks | Unknown |

**Common problem:** Teams set targets without knowing the baseline. "Reduce support calls by 30%" sounds great until you realize nobody tracks support calls at the level of granularity you need. This step surfaces that gap early.

---

### Step 2: Brainstorm data sources

For each Key Result where the baseline is unknown or uncertain, brainstorm where the data might live.

**Silent brainstorm:** 5 minutes. One sticky note per data source idea. Don't filter -- capture every possibility.

**Common data source categories:**

| Category | Examples |
|---|---|
| **Analytics platforms** | Google Analytics, Mixpanel, Amplitude, Adobe Analytics, Heap |
| **Support systems** | Zendesk, Intercom, Freshdesk, call center reports, ticket categorization |
| **CRM / sales data** | Salesforce, HubSpot, pipeline reports |
| **Product databases** | Internal dashboards, data warehouse queries, Looker/Tableau |
| **Financial systems** | Revenue reports, billing system exports, payment processor dashboards |
| **User research** | Survey results, NPS scores, usability test completion rates |
| **Manual tracking** | Spreadsheets someone maintains, meeting notes, Slack channel archives |

**Facilitation tip:** Ask "who on the team has looked at this number before?" Often someone has a rough answer from a previous project, a quarterly review, or a support team report. Rough is fine -- you're establishing a starting point, not publishing a research paper.

---

### Step 3: Dot-vote on best data sources

Not all data sources are equally reliable or accessible. Vote on which ones to pursue.

**For each Key Result, evaluate data sources on two criteria:**

| Criteria | What to ask |
|---|---|
| **Reliability** | Is this data accurate? Is it collected consistently? Can we trust it? |
| **Accessibility** | Can we get this data within 1-2 weeks? Do we need engineering help, a data request, or vendor access? |

**Dot-vote:** Each person gets 2 votes per Key Result. Vote for the data source that's most reliable AND accessible.

**Select the top 2-3 data sources per Key Result.** More than that means you're over-investing in measurement infrastructure. Pick the sources you can actually use and move on.

---

### Step 4: Document baselines and assign owners

For each Key Result, capture the baseline plan.

```
Key Result: (KR text)

Data source(s): (Top-voted source)
Current baseline: (Value, "estimated," or "to be determined by [date]")
Measurement method: (How will we pull this number? Query, dashboard, manual export?)
Owner: (Who is responsible for establishing and tracking this baseline?)
Baseline due date: (When will we have the confirmed number?)
Target: (Only set after baseline is confirmed)

Notes: (Any caveats, limitations, or known data quality issues)
```

**Rules for good baselines:**
- A baseline should be a specific number, not a range. "~22%" is acceptable as a starting estimate. "Somewhere between 15-30%" is not a baseline.
- If you truly cannot establish a baseline within 2 weeks, the Key Result might need rewriting. A KR you can't measure is a KR you can't manage.
- Baselines from different time periods need normalization. "Last quarter" and "last month" tell different stories. Pick a consistent window.
- Document the measurement method so anyone can reproduce the number. "Priya ran a SQL query" is not documentation. The query itself is.

---

## Worked example

**Team:** PatientConnect product team
**Quarter:** Q2 2026
**Objective:** Make it effortless for patients to manage their health between appointments

**Key Results and baseline establishment:**

### KR1: Reduce support call volume by 30%

| Field | Value |
|---|---|
| Data source | Zendesk ticket categorization + call center monthly report |
| Current baseline | 4,200 calls/month tagged "billing" or "claims" (Feb 2026 average) |
| Measurement method | Zendesk export filtered by category tags, cross-referenced with call center report |
| Owner | Support lead (Jordan) |
| Baseline confirmed | Yes -- Feb data pulled Mar 5 |
| Target | 2,940 calls/month by end of Q2 |
| Notes | Call categorization was inconsistent before Jan 2026. Using Feb as baseline because it's the first month with clean tagging. |

### KR2: Increase online payment adoption to 40%

| Field | Value |
|---|---|
| Data source | Billing system dashboard (payment method breakdown) |
| Current baseline | 22% of payments made online (Q4 2025 average) |
| Measurement method | Monthly billing system export, % of total payments by channel |
| Owner | Product analyst (Mei) |
| Baseline confirmed | Yes -- Q4 data available in billing dashboard |
| Target | 40% online payments by end of Q2 |
| Notes | Baseline excludes patients on auto-pay (already "online"). Including them would inflate the starting number. |

### KR3: Achieve 90% task completion rate for top 5 self-service flows

| Field | Value |
|---|---|
| Data source | Mixpanel funnel analysis |
| Current baseline | To be determined by Mar 28 |
| Measurement method | Define the 5 self-service flows, create Mixpanel funnels, measure completion rate over 2-week window |
| Owner | Product analyst (Mei) + UX researcher (Anika) |
| Baseline confirmed | No -- funnels need to be set up |
| Target | TBD pending baseline (will set target once we know current completion rate) |
| Notes | Need to agree on which 5 flows to measure. Candidates: password reset, appointment scheduling, bill payment, prescription refill, message clinician. |

---

## Facilitator tips

- **Baselines before targets.** Teams love setting ambitious targets. But "improve by 30%" is meaningless if nobody knows the starting point. Resist the urge to set targets until baselines are confirmed.
- **"Unknown" is a valid starting state.** The worksheet's job is to make unknowns visible and assign someone to resolve them. Don't panic if half your KRs have unknown baselines at the start of the quarter. That's normal -- and that's why you're doing this exercise.
- **Beware the vanity baseline.** Some data sources tell a flattering story. If the team picks the data source that makes the baseline look worst (so the target looks easier to hit), call it out. Baselines should be honest, not strategic.
- **Measurement ownership matters.** Every baseline needs one person who will pull the number, track it over the quarter, and report on it. No owner means no tracking. No tracking means the KR is decorative.
- **Timebox the data source brainstorm.** Teams can spend 30 minutes debating the perfect data source. You need a good-enough data source you can access this week. Perfect measurement is the enemy of useful measurement.
- **Revisit baselines at mid-quarter.** If a baseline turns out to be wrong (data quality issue, methodology change), adjust it. A corrected baseline is more useful than a wrong one you're committed to.

---

## How did it go?

After completing the worksheet, check:

- [ ] Every Key Result has at least one identified data source
- [ ] Baselines are specific numbers, not ranges or qualitative descriptions
- [ ] Each baseline has an owner and a due date (if not yet confirmed)
- [ ] The measurement method is documented clearly enough for someone else to reproduce
- [ ] Targets were set after baselines were established, not before
- [ ] Data source limitations and caveats are documented (not hidden)
- [ ] Unknown baselines have a plan and timeline to resolve

---

*Part of the [k8 Agent Toolkit](https://k8mak.com/agent-toolkit). Download other formats at k8mak.com/resources.*