Frontier of Understanding (Goal by Risk Type)
Before committing to an outcome goal, set the goal at the true edge of what your team knows — understanding, dependency, execution, or strategic risk.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 90%
A goal-setting method that fixes broken OKRs by matching the goal to the type of risk actually blocking progress. There's a frontier between what a team knows and doesn't know; the right goal pushes that frontier forward rather than blindly committing to move a metric you don't yet know how to move. The four risk types — understanding, dependency, execution, strategic — form a sequence you advance through.
Origin
Developed by Ravi Mehta after watching multiple companies struggle with OKRs, and published on the Reforge blog as an argument for NCTs (Narratives, Commitments, Tasks) over traditional OKRs. It's a reaction to the dogma of always setting outcome-over-output goals.
Core principles
- 01Outcomes are the ultimate aim, but committing to a specific outcome you don't know how to move is a mistake.
- 02If you don't understand how to move a metric, the right goal is to increase understanding — not to move the metric.
- 03Every quarter should push the frontier of understanding forward, even when the metric doesn't move.
- 04Judge yourself on a two-by-two: did we hit our goals, and do we know why.
How to run it
- 1
Locate your frontier of understanding
Identify the junction between what the team knows and doesn't know about the metric you care about. If you can only brainstorm ten experiments but can't say why users behave as they do, you're at the understanding frontier.
Watch out Committing to a metric goal while sitting at the understanding frontier means throwing spaghetti at the wall — you may move the number without knowing why, or move it in a way disconnected from strategy.
- 2
Address understanding risk first
If you don't know the levers, set a goal to increase understanding: spend a couple of weeks talking to customers, running analysis, and forming strong hypotheses about what moves the metric.
- 3
Clear dependency risk
Once you think you know the levers, check whether you have the tools and resources to act on them. If not, the goal is to acquire what you're missing.
- 4
Set an execution goal
When you have the resources and a strong hypothesis, commit to an execution target — e.g. 'hit 20 experiments this quarter.' Even experiments that fail move the frontier forward and prove the team can execute.
Pro tip An execution goal keeps a team productive and learning even before you're sure a lever works.
- 5
Test strategic risk last
The final frontier: you understand how to move the metric and execute the plan, then learn whether the hypothesis was right. If right, pull the lever harder; if wrong, return to understanding and re-goal.
Pro tip When you miss an outcome goal, come back to leadership with where the frontier was, what you did, and where it went off the rails — that clarity earns more trust than blindly hoping.
In the wild
Asked to move retention, a team can brainstorm ten experiments but genuinely doesn't know why people keep using the product. Rather than commit to a retention number and hope, they spend the first weeks understanding — talking to customers and analyzing — to form real hypotheses, then convert those into execution goals, and only then test whether the hypotheses hold.
→ The team either validates its understanding and doubles down, or misses the metric but can explain exactly why and propose a higher-confidence commitment for next quarter.
Common mistakes
Committing to an outcome you don't know how to move
Setting a metric goal at the understanding frontier forces the team to throw many experiments at the wall; some stick, but you never learn why, or you move the metric in a way untied to your strategy — fragile progress that catches up with you.
Optimizing for goals over learning
Teams that fixate on hitting the number get less focused on the learning. Hitting goals without knowing why is the dangerous quadrant — fine for now, but it eventually collapses because you can't repeat or defend the result.
Is it for you?
Best for
PMs and product leaders wrestling with OKRs on a metric they've been told to move but don't yet understand.
Not ideal for
Mature teams with well-understood levers and abundant traffic, where straightforward experimentation reliably moves the metric.
From the transcript
“there's a point at which what the team knows and what the team doesn't know there's a junction point there which is this Frontier”
“if you don't understand how to move a particular metric then the right goal is to set a goal to increase your understanding not to…”
“the four buckets are it starts with understanding risk which is we have something that we want to do but we don't really understand what…”
“on one axis of the Matrix you have did we hit our goals and on the other axis we have do we know why”
From the episode
Building your product strategy stack
Ravi Mehta (Tinder, Facebook, Tripadvisor, Outpace)