The Data-Informed Product Loop
Strategy to models to measurement to bets to impact to learning — find the broken link.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 7
- Confidence
- 95%
Cutler taught this loop at Amplitude as a diagnostic rather than a process. Seven linked stages run from strategy through qualitative models, measurement, prioritisation, bet design, impact measurement and back into strategy. Its value is not in following it forwards but in locating which stage you are weak at — because every stage is a hard gate: doing six of them perfectly and failing the seventh means the whole loop produces nothing. It pairs with Cutler's definition of skill as knowledge times practice, mediated by environment, habits and motivation, which explains why consuming more product content never substitutes for going around the loop.
Origin
Developed and taught at Amplitude as the 'data-informed product loop', alongside the North Star Playbook Cutler co-wrote with Jason. Cutler is explicit that Amplitude did not invent the North Star framework itself and credits Sean Ellis and other earlier practitioners.
Core principles
- 01Every stage is a gate — the loop is only as strong as its weakest link.
- 02Skill = knowledge x practice, mediated by environment, habits and motivation. Knowledge alone is inert.
- 03The number of loops you complete matters more than the number of podcasts you consume.
- 04A qualitative model (like a North Star) precedes measurement; measurement without a model measures nothing meaningful.
How to run it
- 1
Have a strategy
Without a strategy, nothing downstream is possible. This is the hard precondition of the loop.
Watch out An amazing strategy that never gets modelled is still a dead end — do not stop here.
- 2
Develop qualitative models
Turn the strategy into a shared qualitative model of how value is created — the North Star framework with its inputs is one example of such a model.
Pro tip Models are what make prioritisation legible to people other than the strategist.
- 3
Add measurement to the models
Instrument the model so you can see how you're actually doing. Measurement attaches to the model, not to a list of competitor metrics.
Pro tip Start absurdly small — twenty events on a free plan, three hours with your developers. You can throw them away later.
Watch out The classic anti-pattern is a months-long implementation project trying to document every question and every metric before shipping a single event.
- 4
Prioritise where to focus
Use the measured model to decide where attention goes. If nobody can prioritise from your model, the model is not doing its job.
- 5
Design bets
Convert priorities into explicit bets that can actually ship. A team that measures beautifully and designs no bets ships nothing.
- 6
Measure the impact of the bets
Close the evidence gap between what you shipped and what changed. Shipping without knowing the impact of anything you shipped is a broken loop.
- 7
Circulate the learning back into strategy, models and prioritisation
Feed what you learned back into the strategy, the qualitative models and the way you prioritise. Companies that do everything else right and skip this stage still fail.
Pro tip Ask which stage of the loop you are currently weakest at — that answer is the actual next investment, not the newest framework.
In the wild
Highly qualified companies would spend months in back-and-forth trying to document every question, every metric and every event before implementing analytics — treating it as one big project. Cutler wanted to tell them to just use the free plan, get twenty events going, and accept they might wipe them out later.
→ Companies delayed getting any value for months when three hours with their developers would have started the loop immediately.
Cutler met leaders who felt beaten up by the advice industry — loaded with knowledge, convinced their company could never be good enough, unable to follow anyone's advice.
→ His prescription was to stop accumulating knowledge and start getting the loop going for their team, career or company.
Common mistakes
Mistaking knowledge intake for skill
Skill is knowledge times practice, mediated by environment, habits and motivation. The number of podcasts and posts you consume contributes only to the knowledge term; without reps around the loop it multiplies by zero.
Copying competitors' metrics instead of measuring your own model
Asking 'what are all our competitors measuring, which exact metric should we track' skips stages 1 and 2. Measurement exists to reduce uncertainty about a decision you actually need to make.
Is it for you?
Best for
Product teams and product leaders who feel busy and well-informed but cannot show compounding learning or impact.
Not ideal for
Pre-strategy exploration where the goal is still to discover what business you are in.
From the transcript
“at an amplitude we have this data informed product Loop uh that we would teach and it's basically like you need a strategy you need…”
“you need a strategy without that nothing is possible but you could have an amazing strategy and you don't deploy it with the right models…”
“this is a skill and skill is knowledge times practice mediated by your environment the habits you form and the motivation that you have”
From the episode
What differentiates the highest-performing product teams
John Cutler (Amplitude, The Beautiful Mess)