A facilitation guide for building shared understanding
How to run a metric tree workshop
A step-by-step facilitation guide for running a metric tree workshop. Covers preparation, agenda, techniques, and pitfalls so your team builds a shared metric tree in one session.
11 min read
Why a workshop, not a spreadsheet
The most common way organisations attempt to build a metric tree is also the least effective: one person opens a spreadsheet, maps out the metrics they think matter, and circulates it for comments. The result is a document that reflects a single perspective, collects a handful of superficial edits, and gathers dust within weeks. The tree might be technically correct, but nobody feels ownership over it because nobody helped build it.
Behavioural science explains why. Shared mental models are not built through documentation. They are built through dialogue. When a Head of Product and a Head of Sales sit in the same room and debate whether activation rate or pipeline velocity is the stronger driver of revenue growth, both leave with a deeper understanding of how the business actually works. That understanding cannot be transmitted through a slide deck or a Notion page. It has to be constructed through conversation, disagreement, and eventual alignment.
This is why the workshop format matters. The process of building the tree is as valuable as the tree itself. A workshop forces cross-functional teams to surface assumptions, resolve conflicts, and commit to a shared model of cause and effect. The metric tree you produce at the end is the artifact. The real output is the alignment that happens along the way.
Key insight
The process of building the tree is as valuable as the tree itself. A metric tree built by one person in isolation reflects one perspective. A metric tree built collaboratively reflects shared understanding, and shared understanding is the prerequisite for coordinated action.
Before the workshop
Good workshops are won or lost before they start. The facilitation techniques matter, but they cannot compensate for poor preparation. Invest an hour or two upfront to set the conditions for a productive session. The following steps will prevent the most common failure modes: wrong people in the room, no agreement on the starting point, and unrealistic expectations about what can be achieved.
- 1
Identify the right participants
You need cross-functional representation. Include someone from each major branch of the business: product, engineering, marketing, sales, finance, and customer success. The goal is to have every perspective that shapes how the business creates and captures value. Aim for 6 to 10 people. Fewer than 6 and you will have blind spots. More than 10 and the session becomes unwieldy. If your organisation is large, run separate workshops per business unit and align the trees afterwards.
- 2
Agree on the North Star metric in advance
Do not spend workshop time debating which metric sits at the top of the tree. This decision should be made by leadership before the session. Circulate the North Star metric, its definition, and its current value at least a week in advance. If there is genuine disagreement about the North Star, resolve it in a separate, smaller meeting. The workshop assumes alignment on where you are heading.
- 3
Gather baseline data for top-level metrics
Bring current values and recent trends for the North Star and its likely first-level drivers. Having real numbers in the room grounds the conversation in reality rather than opinion. It also helps participants spot gaps: if nobody can produce a number for a metric you think matters, that tells you something important about your measurement maturity.
- 4
Prepare a blank tree template
Whether you use a whiteboard, a digital tool, or sticky notes on a wall, have the structure ready before the session starts. Pre-populate only the North Star at the root. Everything else should be empty. The blank space is intentional: it signals that the group will build this together, not react to a pre-built draft.
- 5
Set expectations with participants
Send a brief pre-read explaining what a metric tree is, why the team is building one, and what the session will involve. Be explicit: this is a working session, not a presentation. Participants should come prepared to contribute, debate, and commit. Share the agenda and time blocks so people know what to expect and can plan their energy accordingly.
- 6
Schedule 2 to 3 hours
A one-hour session is not enough. You will spend the first 15 minutes aligning, the middle hour decomposing, and you need at least another hour for debate, relationship definition, and ownership assignment. Book 2.5 hours with a 10-minute break in the middle. If your organisation resists long meetings, frame this as a quarterly investment that saves hundreds of hours of misalignment downstream.
The workshop agenda
The following agenda has been refined through dozens of metric tree workshops across SaaS, e-commerce, and marketplace businesses. The total time is approximately 2 hours and 20 minutes. Adjust the time allocations to fit your context, but preserve the sequence. Each phase builds on the previous one, and skipping ahead creates gaps that are expensive to fill later.
- 1
Phase 1: Align on the North Star and first-level drivers (15 minutes)
Start by presenting the North Star metric, its current value, and its trend. Then facilitate a brief discussion to agree on the two to four first-level drivers. This should not take long if you did the pre-work. The facilitator proposes a starting decomposition, and the group validates or adjusts it. Keep this tight. The goal is a shared starting point, not a perfect answer. You will revisit these relationships later.
- 2
Phase 2: Decompose each branch in breakout groups (45 minutes)
Divide participants into small groups of 2 to 3 people, one group per first-level branch. Each group decomposes their branch as deep as they can go. Use the "what drives this?" question recursively: for every metric, ask what the two to three factors are that determine it. Stop when you reach metrics that a single team or person can directly influence. Give each group a whiteboard section or a digital canvas to work on. The facilitator floats between groups to check progress and prevent rabbit holes.
- 3
Phase 3: Reconnect and debate the full tree (30 minutes)
Bring the groups back together. Each group presents their branch to the full room. This is where the real value happens. Cross-functional challenge surfaces assumptions that a single-function group would miss. Sales might point out that the product group forgot to include pricing as a driver. Finance might challenge the relationship between marketing spend and pipeline. Capture disagreements. If something cannot be resolved in the room, mark it for follow-up. Do not let perfect be the enemy of good.
- 4
Phase 4: Define relationships (20 minutes)
Walk through the tree and label each connection. Is it multiplicative (Revenue = Users x ARPU), additive (Total Users = Segment A + Segment B), or influencing (NPS affects retention but is not a formula)? Also note the direction: does the child metric increase or decrease the parent? This step transforms a diagram into a model. It is the difference between "these metrics are related" and "here is specifically how they are related." Skip this step and the tree loses most of its predictive value.
- 5
Phase 5: Assign ownership to every metric (20 minutes)
Go node by node and assign a named owner to each metric. Not a team. A person. Ownership means this person is responsible for monitoring the metric, investigating when it moves, and taking or coordinating action. This step is uncomfortable. People resist putting their name next to a number they cannot fully control. That discomfort is productive. It forces honest conversation about who actually influences what. If nobody will own a metric, question whether it belongs in the tree at all.
- 6
Phase 6: Agree on next steps and review cadence (10 minutes)
Close the session by agreeing on three things: who will digitise and circulate the tree within 48 hours, when the first review meeting will happen (recommend two weeks out), and what cadence the tree will be reviewed on going forward (recommend monthly for the first quarter, then quarterly). Assign these as explicit actions with owners and deadlines. A workshop without follow-through is just a team-building exercise.
Facilitation techniques that work
Running a metric tree workshop is not the same as running a typical meeting. You are asking people to build a model of how their business works, which requires structured thinking, honest debate, and disciplined time management. These techniques address the most common group dynamics challenges that derail workshops.
The "five whys" for decomposition
For every metric, keep asking "what drives this?" until you reach something a team directly controls. This recursive questioning prevents the tree from staying too abstract. If a group gets stuck, prompt them: "If this metric dropped 20% tomorrow, what would you investigate first?" The answer is usually a child metric.
Silent brainstorming before discussion
Before any group discussion, give participants 3 minutes to write down their ideas individually. This prevents anchoring bias, where the first idea spoken aloud dominates the conversation. Research on group decision-making consistently shows that silent ideation followed by structured sharing produces more diverse and higher-quality outputs than open discussion alone.
Dot voting for prioritising branches
When the group has more potential branches than time allows, use dot voting. Give each participant 3 votes to place on the branches they believe are most important to decompose further. This surfaces collective priorities without lengthy debate and prevents a single loud voice from setting the agenda.
The "red team" challenge
Assign one or two people to play devil's advocate for each branch. Their job is to find gaps, challenge assumptions, and ask "what are we missing?" This role should rotate so the same people are not always in the critic seat. Red teaming catches blind spots that consensus-driven groups naturally overlook.
Time-boxing each branch
Allocate a fixed number of minutes per branch and stick to it. Use a visible timer. Without time-boxing, groups inevitably spend 80% of their time on the first branch and rush through the rest. If a branch needs more depth, schedule a follow-up session for that branch specifically rather than stealing time from others.
Parking lot for tangential discussions
Keep a visible "parking lot" list on a whiteboard or shared document. When conversations drift into tangential territory, such as data quality concerns, tooling decisions, or organisational restructuring, capture the topic in the parking lot and return to the tree. Review parked items in the last 5 minutes and assign owners for follow-up.
Common workshop pitfalls
Even well-prepared workshops can go wrong. These are the pitfalls that occur most frequently, along with how to recognise and address them in the moment. Knowing these in advance helps the facilitator intervene early rather than discovering the problem after the session ends.
The HiPPO problem
The Highest-Paid Person's Opinion dominates the room. Senior leaders propose a decomposition and nobody pushes back. Counter this by using silent brainstorming, asking juniors to share first, and explicitly inviting dissent: "Who sees this differently?" If the CEO is in the room, brief them beforehand to hold back and let others lead the decomposition.
Going too deep too fast
One branch gets decomposed to five levels while others remain at one level. This happens when subject matter experts get excited about their domain. The facilitator must enforce breadth-first decomposition: get every branch to two levels before any branch goes to three. Depth can always be added in follow-up sessions.
Confusing metrics with initiatives
Participants add "launch new pricing page" or "hire two SDRs" to the tree instead of measurable metrics. Initiatives are actions. Metrics are the numbers those actions are designed to move. When this happens, ask: "What metric would improve if that initiative succeeded?" Put the metric in the tree and capture the initiative separately.
Too many or too few participants
With fewer than 6 people, critical perspectives are missing and the tree reflects a narrow view of the business. With more than 12, coordination costs exceed the value of additional perspectives. If you must include more stakeholders, use a two-phase approach: a core group of 8 builds the tree, then a wider group of 20 reviews and challenges it asynchronously.
Skipping the ownership step
Time runs short and the group decides to "assign owners later." This almost never happens. The ownership conversation is uncomfortable, which is exactly why it must happen in the room while people are engaged. A tree without owners is a diagram, not a management tool. Protect the last 20 minutes of the agenda for this step, even if it means cutting the decomposition short.
Treating it as a one-off event
The workshop produces a beautiful tree that is never revisited. Within a month, the business has changed and the tree is outdated. The workshop must end with a commitment to a review cadence. Schedule the first review meeting before people leave the room. A metric tree is a living document. If it is not maintained, it becomes decoration.
After the workshop
The 48 hours after the workshop are critical. Momentum decays exponentially. If the tree is not digitised and circulated within two days, participants start to forget the nuances of what was agreed. Within a week, the shared understanding that was so carefully constructed begins to fragment as people revert to their pre-workshop mental models.
The first step is to document the tree digitally. Transfer the sticky notes, whiteboard drawings, or photos into a structured format that everyone can access. Include the metric names, the relationships between them, the owners, and any notes or open questions from the session. Circulate this to all participants and ask for asynchronous feedback within 48 hours. This is not an invitation to redesign the tree. It is a chance to catch errors, add context, and confirm that the digital version matches what was agreed in the room.
Within two weeks, connect the tree to live data sources. Start with the metrics you can already measure. For metrics that do not yet have a data source, create a manual tracking process as a temporary bridge. The goal is to make every node in the tree show a real number as quickly as possible. A tree with numbers is a tool. A tree without numbers is a poster.
Schedule the first review meeting for two weeks after the workshop. In this meeting, walk through the tree with the owners. Has anything changed? Are the relationships holding up? Are there branches that need more depth or metrics that turned out to be unmeasurable? Use this meeting to refine the model based on the first round of real data. Then continue with monthly reviews for the first quarter before moving to a quarterly cadence.
“A metric tree is a living document. The version you build in the workshop is version one, not the final version. The businesses that get the most value from metric trees are the ones that treat them as an evolving model of how their business works, updated regularly as they learn more about cause and effect.”
A worked example
To make this concrete, here is what a completed workshop output might look like for a B2B SaaS company. The North Star metric is Annual Recurring Revenue (ARR). The workshop participants included the CEO, VP Product, VP Sales, VP Marketing, VP Customer Success, and Head of Finance.
In Phase 1, the group agreed on two first-level drivers: ARR = Number of Paying Customers x Average Revenue per Customer. In Phase 2, the breakout groups decomposed each branch. The Sales and Marketing group broke down Number of Paying Customers into New Customers, Expansion (existing customers upgrading), and Churned Customers. They then decomposed New Customers further into the pipeline stages: Leads, Qualified Opportunities, and Won Deals. The Product and Customer Success group decomposed Average Revenue per Customer into plan mix and usage-based add-ons, then decomposed churn into voluntary churn and involuntary churn (failed payments).
In Phase 3, the reconnection surfaced an important insight: the Product group had not included onboarding completion as a driver of retention, and the Sales group pointed out that deal size at close strongly predicted expansion revenue. Both metrics were added. Phase 4 labelled the top-level split as multiplicative, the customer segments as additive, and relationships like onboarding-to-retention as influencing. Phase 5 assigned owners: VP Sales owned New Customers, VP Customer Success owned Churn, and VP Product owned Activation Rate.
The tree below shows the result. Notice how each bottom-level metric maps to a specific team and person. This is what makes a metric tree actionable.
Run your first metric tree workshop with KPI Tree.
Build your metric tree collaboratively, connect it to live data, and give every team line of sight from their work to company outcomes. Start with a guided proof of concept.
