The Build · 20 min

Learning Loop

Turn any decision into a falsifiable bet.

A one-page test plan with five lines: the bet, the smallest test, the leading signal this week, the lagging signal this month, and the threshold that flips the decision. The artefact you can pin above your desk — and a pre-committed kill criterion that means you'll actually act on the result.

Quick Facts
Duration~20 minutes
CategoryThe Build
OriginThe Lean Startup, 2011
LineageRies · Osterwalder · Popper
Used in pathways
The Build
Start a Learning Loop →

What it is

A Learning Loop is the discipline of designing a measurement plan before you ship anything. It descends directly from Eric Ries's build-measure-learn cycle — and the trick most teams miss is that the loop is designed backwards. You start by deciding what you need to learn, then what you'd measure to know, and only then what you'd build to enable the measurement. Most founders skip the design step and run an unfalsifiable test.

It also borrows from Strategyzer's Test Card (Alexander Osterwalder), which locks every experiment into the same shape: "We believe X. To verify, we'll do Y. We're right if Z." The Learning Loop in The Studio reflects bets back into that structure on purpose — it exposes any holes before you spend a cent. A measurement plan beats a launch plan, every time, because a launch tells you what happened. A loop tells you what to do about it.

The five stages

Every Learning Loop fits on a single page. WAiDE walks you through one stage at a time:

1
The Bet
What you believe will happen, framed as a falsifiable cause-and-effect claim. Not a goal, not a feature — a claim that could be wrong.
2
The Smallest Test
The cheapest, fastest experiment that would tell you the bet was right. A fake door, a concierge MVP, five customer interviews — never a launch dressed as a test.
3
Leading Signal · this week
The first behaviour or response within seven days that says it's working. A specific count of a specific behaviour in a defined sample — not a vibe.
4
Lagging Signal · this month
The downstream consequence within thirty days that confirms the leading signal was real. Revenue, retention, repeat behaviour — harder to fake.
5
The Threshold
The number that flips the decision — pivot, stop, or double down. Pre-committed before you run the test. The most valuable line on the page.
6
If-Then commitment
The whole loop sealed as an If-Then with dates: "If [signal] is below [number] by [date], then I will [pivot/stop/double down] by [date]."

When to use it

Before a commitment you can't easily reverse
When you're about to spend money, pick a market, or hire a team — and you want to know how you'd find out you were wrong before you commit.
When a team is making a bet without a way to measure it
Roadmaps and OKRs name the destination. The Learning Loop names the signal. If your team can't say what number would tell them the bet failed, the bet is just a wish.
When the next move feels expensive
A cheap test before the build is almost always cheaper than the build that turns out to be wrong. The Learning Loop is the fastest way to size a test against a decision.

What you'll walk away with

Three things — every time:

  • A one-page test plan you can pin above your desk and share with your team in 30 seconds.
  • Pre-committed kill criteria — the threshold — so you act on the result instead of rationalising it.
  • An If-Then commitment with dates: "If [signal] is below [number] by [date], then I will [pivot/stop/double down] by [date]."

Sample output

Here's what a Learning Loop looks like at the end of a session:

Example: B2B SaaS founder testing a pricing change
The Bet
"If we raise the team plan from $29 to $49 per seat, conversion from trial to paid will hold above 8%, because the buyers are budget-holders, not price-sensitive end users."
The Test
A/B price page for 14 days. New trials see $49; control sees $29. Sample target: 200 trials per arm.
Leading Signal · 7 days
Of the first 100 trials in the $49 arm, at least 8 reach the "add team member" step (the highest-intent action).
Lagging Signal · 30 days
Trial-to-paid conversion in the $49 arm is at or above 8%, with 30-day retention matching the control.
Threshold · what flips the decision
Below 6%, we revert and pivot to a value-based pricing test. At or above 10%, we double the spend on the new price across all channels by end of month.
If-Then commitment
"If trial-to-paid in the $49 arm is below 6% by Day 30, I'll revert pricing and start the value-based pricing loop by Day 35."

The methodology behind it

Build-Measure-Learn (Eric Ries, 2011). The most-quoted line from The Lean Startup is also the most misunderstood. The loop is meant to be designed backwards — first decide what you need to learn, then what to measure, then what to build to enable the measurement. Most teams build first and then look for a metric that flatters the build. That's not a loop; it's a launch with a justification.

Validated learning vs vanity metrics. Total users, page views, registered accounts, "engagement" — all of these can rise while the business dies. Ries's distinction is between vanity metrics (dashboard numbers that go up and to the right) and actionable metrics (cohort behaviour that ties cause to effect at the per-user level). The Learning Loop forces you to define a specific behaviour in a specific sample. "3 of 10 customers asked about price unprompted" is a signal. "Engagement is up" is a vibe.

Pre-commitment and kill criteria. Borrowed from decision science and modern PE/VC discipline: write the failure threshold before you run the test. Without a pre-committed threshold, founders rationalise the result either way it lands. The threshold is the rigour test for the rigour test — the moment the loop becomes real is when you commit, on the record, that you'll actually pivot if the number hits.

Falsifiability (Karl Popper). A hypothesis isn't a hypothesis if no possible result would change your mind. The threshold is the falsifiability test made concrete. If you can imagine a result you'd shrug off, you don't have a hypothesis — you have a preference dressed up as one.

Where it sits in the curriculum

The Learning Loop is the rigour layer underneath every other Build-pathway tool. Lean Canvas describes the model; the Learning Loop describes how you'd know the model was right. It's taught in the Master of Entrepreneurship as the discipline that separates a hypothesis from an opinion, in the Growth Engine program as the operating cadence for early-stage growth experiments, and in Scaling Ops as the If-Then format for committing teams to action on the result.

Master of EntrepreneurshipGrowth EngineScaling OpsWade Institute

Why it works

Most planning tools fail in a specific way: they describe the future the team wants, but not the way the team would find out the future wasn't going to happen. The Learning Loop is engineered to close that gap. It assumes — correctly — that the most expensive thing a founder can do is run a test that can't be wrong, because the result tells them nothing they didn't already believe.

The structure does specific work. Forcing the leading signal to be a behaviour in a defined sample (rather than a number on a chart) eliminates vanity metrics by construction. Forcing the lagging signal to be a downstream consequence (revenue, retention, repeat) prevents the leading signal from being gamed without showing up later. And forcing the threshold to be pre-committed prevents the most common pathology in early-stage companies — running a test that always seems to confirm the founder's prior, because the result was never going to land anywhere else.

The mechanism: the Learning Loop doesn't make the decision easier. It makes the decision honest. You leave the session knowing exactly what would change your mind — which is the only foundation on which a real test can be built.

Frequently asked questions

What's the difference between a Learning Loop and a Lean Canvas?

A Lean Canvas describes the bet structurally — what the model is. A Learning Loop describes how you'd actually know if the bet was right. The canvas is the hypothesis; the loop is the test plan. Most teams have one without the other and wonder why they keep building things that don't land.

Why pre-commit to a threshold before running the test?

Without a threshold written in advance, founders rationalise the result whichever way it lands. Pre-commitment is the difference between rigour and theatre. If you can't name the number now, it means you haven't decided yet what failure looks like — and that's the conversation worth having.

What's a vanity metric, and why does it matter?

Vanity metrics — total users, page views, "engagement" — can rise while the business dies. Actionable metrics tie cause to effect at a per-cohort level. The Learning Loop pushes you to define a behaviour in a defined sample, not a number on a dashboard.

Is the Learning Loop only for early-stage startups?

No. The same discipline applies to corporate innovation, internal product bets, and any decision that's expensive to be wrong about. If you're about to commit resources and you can't say how you'd know if it didn't work, you need a loop before you need a plan.

Run a Learning Loop?

WAiDE will guide you. About 20 minutes.

← Back to toolboxStart a Learning Loop →