Input vs Output Metrics

Focus on controllable inputs that drive customer value; outputs will follow

Bill Carr
Unpacking Amazon's unique ways of working | Bill Carr (author of Working Backwards)

Input vs Output Metrics

"We took it as an article of faith that if we can just improve these inputs, the outputs will take care of themselves. The inputs are the things that drive the outputs, which are revenue, customer activity, free cashflow." - Bill Carr

What It Is

Input vs Output Metrics is Amazon's framework for focusing teams on controllable, causal factors (inputs) rather than lagging financial indicators (outputs). The core insight: revenue and profit are outputs that you can't directly control—they result from improving customer experience inputs that you can control.

This approach was crystallized when Amazon read "Good to Great" and codified their growth flywheel. They identified the specific inputs that, if improved, would drive the outputs they cared about. Then they shifted goal-setting and weekly business reviews to focus obsessively on those inputs.

In one telling data point: Amazon's S-Team goals list (about 500 items) contained only 10 financial metrics. The rest were all input metrics related to selection, price, speed, and customer experience.

How It Works

Output Metrics (What you want but can't directly control):

  • Revenue
  • Active customers
  • Free cash flow
  • Share price

Input Metrics (What you can control and measure):

  • Selection (how many products available)
  • Price (relative to competition)
  • Customer experience (page load speed, click rates, delivery time)
  • Cost structure (enables lower prices)

The Amazon Flywheel Example: Amazon's inputs formed a reinforcing loop:

  • Broad selection → Better customer experience
  • Low prices → More customers
  • More customers → More third-party sellers
  • More sellers → More selection
  • Lower cost structure → Lower prices

How to Apply It

Step 1: Map your end-to-end customer experience Walk through every step of your customer journey. Bill Carr illustrated this with an Airbnb example:

  1. Customer clicks an ad, arrives on website/app
  2. Browsing or searching (how easy? how fast?)
  3. Looking at a listing detail page (what actions can they take?)
  4. Reserving and messaging the host (how many messages? good or bad?)
  5. The actual experience
  6. Post-stay

Step 2: Identify what you can measure at each step For each stage, ask:

  • What can we measure about speed, quality, and ease?
  • Is the metric something we can actually improve with resources?
  • Does it touch the customer experience?

Step 3: Experiment to find causal inputs Not all input metrics are created equal. You need to discover which ones actually drive outputs:

  • Throw many metrics at the wall
  • Measure, observe, improve
  • Look at effect on outputs
  • Iterate using DMAIC (Define, Measure, Analyze, Improve, Control)

Step 4: Refine your measurements Your first measurement approach might be wrong. Bill Carr tells the story of Amazon measuring selection incorrectly for years before refining the metric.

Step 5: Stop the fire drills The old pattern: "We're not going to hit our revenue number this quarter. Quick, run promotions!"

The new pattern: If you improve inputs consistently, outputs follow over time. Short-term fire drills don't actually work and distract from what matters.

When to Use It

  • Weekly and quarterly business reviews
  • Setting team goals and OKRs
  • Diagnosing why growth is slowing
  • Prioritizing roadmap investments
  • When you're tempted to chase short-term revenue tactics

Signs You Have the Right Input Metric

  1. Controllable - You can apply resources to make it better or worse
  2. Touches customers - It affects the customer experience directly or indirectly
  3. Causal - Changes in the input correlate with changes in outputs over time
  4. Measurable - You can instrument and track it reliably

Common Mistakes

  1. Compound metrics - Amazon tried "fitness functions" that weighted multiple metrics into one index. It obfuscated what actions were driving results. They abandoned it.

  2. Optimizing the output directly - Chasing revenue with promotions and discounts doesn't improve the underlying flywheel. It's like treating symptoms, not causes.

  3. Measuring the wrong thing - Amazon's selection metric was wrong for years. Be willing to refine how you measure.

  4. Short-term thinking - Input metrics require faith that improving customer experience will drive outputs over time. It's not a quick fix.

Source

  • Guest: Bill Carr
  • Episode: "Unpacking Amazon's unique ways of working | Bill Carr (author of Working Backwards)"
  • Key Discussion: (00:55:03) - Full explanation of input vs output metrics and the flywheel
  • YouTube: Watch on YouTube

Related Frameworks

Back to Decision Making