How Many Metrics Do You Need to Run Your Startup?

Answer: Only one.

One of the concepts we’re pushing hard in the Lean Analytics book is the One Metric That Matters (OMTM).

The idea is this: at any given time in your business there’s one key metric you should be focusing on. Not two, or three, or four. Just one. And that one metric becomes the driving force for what you do, experiments you run, and decision-making.

Picking the metric is hard. Drawing a line in the sand for your goals with the metric is even harder. But these are both important for achieving the levels of focus and speed required to succeed.

At GoInstant we’ve been very analytical. We measure lots of different things from overall usage to specific feature success, to performance and scalability. But none of these are the One Metric That Matters. We’re still working on how to define our OMTM but ultimately GoInstant will be a huge success when most of the Web is co-browsable “out-of-the-box.” That will mean more customers can get up and running with GoInstant more quickly. Everything we do has to make more of the Web and more of the elements of web sites and web applications co-browsable. It’s a huge challenge.

So GoInstant’s OMTM, which is still a work in progress, is essentially a percentage of the Web that is co-browsable. Measuring that is tricky. There are lots of variables. We clearly can’t test the entire Web. Nor is that really necessary. So we’re building systems and developing processes for testing a smaller but significant list of websites, and a large array of web elements (think: all different types of forms, form fields, types of code, popups, logins, interactive elements, and so on; basically the “building blocks” of the Web) as a proxy for “the entire Web.” We test on a continuous basis and will invest more than most companies on testing because of the challenges in building GoInstant’s technology.

What’s your OMTM?

Well that depends on a number of things. Check out a previous blog post (from Lean Analytics) for more info: While you’re there, please sign up for updates too.

In the meantime, here are some elements of the One Metric That Matters to keep in mind:

  1. It will change over time. The OMTM isn’t static. When you’re focused on acquiring users (and converting them into customers) your OMTM may be around which acquisition channels are working best or conversion rate from sign-up to active user. When you’re focused on retention, you may be looking at churn, and experimenting with pricing, features, improving customer support, and so forth.
  2. It answers the most important question you have. At any given time you’ll be trying to answer a hundred different questions and trying to juggle a million things. You need to identify the riskiest areas of your business as quickly as possible; that’s where the most important question lies. And the OMTM should be there to measure and answer that question.
  3. It has a clear goal. A metric without a goal is largely useless. You need to draw a line in the sand that helps you understand whether you’re achieving success or not. The One Metric That Matters needs a goal.
  4. It focuses the entire company. The OMTM should be front and center, physically visible for everyone to see all the time. Think about putting the OMTM on a giant TV screen in your office, or somewhere that people will look multiple times per day. The One Metric That Matters drives focus by being ever-present, and reminds everyone of its importance and the work that has to go into improving it.
  5. It inspires a culture of experimentation. Experimentation is key. Moving through the build->measure->learn cycle as quickly (and as often) as possible is key to generating and amassing enough learning that you can start executing effectively in the right directions. You want to inspire and instill a culture of experimentation throughout your organization — the One Metric That Matters can help.

Collecting lots of data in and of itself isn’t the problem. Most startups do that. Unfortunately that’s what they focus on –collecting lots of data– instead of figuring out what matters. When you fall too far into the numbers you actually lose focus on what’s important. Instead, pick a single number that matters above all others. Maybe you pick the wrong one, but at least you’re making a decision, which is better than dying from indecision.

The One Metric That Matters is both a concept and a tool for building an experimentation-first startup, focusing on measurable goals, improving clarity, and using analytics in the most effective way possible to achieve startup success.

Want more content like this?

Signup for free and you'll get new content as soon as it's available. Thanks!

  • Thomas Rankin

    Great piece Ben. Any comments on metric goal setting? Lots of tension in goal setting in general between achievable and ‘stretch’ goals.

  • Benjamin Yoskovitz

    I think there are a number of factors.

    * Historical data is one. If you have historical data you can use that as a benchmark or the beginnings of a benchmark.

    * Gut is another. Use your gut and answer the question, “What does success here look like?”

    * Find parallels. Through advisors, investors, other entrepreneurs, etc. can you find parallels or benchmarks that might give you a sense of targets you should set as well. Quora has a lot of data hidden away. The guys at are also working on producing lots of benchmarks.

    * I would set big goals — why not?? — and then adjust from there. Better to overreach and miss the mark, but still push hard, then to under reach and not know if you’re really succeeding or not.

  • How to Write a Demand Letter

    Nice Post! I a very happy to see this! 😛

  • Diana Schneidman

    Ben, One Metric That Matters is very well put. My company is not tech and that’s why I need this concept more than most. Tracking too many numbers becomes a job in itself, and if you haven’t thought about which one really matters at this time, you’re off in lots of directions chasing numbers that don’t mean much at all.

  • Benjamin Yoskovitz

    Thanks Diana. Analytics is definitely not owned exclusively by tech startups. All companies can and should use analytics to improve their businesses.

  • Pingback: A Holistic View of Products and Product Management()

  • Pingback: Doing a Lean Startup Conference Workshop on December 4th in San Francisco()

  • Denice Lorraine

    Companies don’t care about environmental impact, specially in African countries; they will comply only if they are forced and many African countries don’t have a stable government. Companies like Field and Technical Services LLC are helping with environment friendly practices and the same time making big money; but they are having hard time growing outside North America.

  • Pingback: I Look Good In Design | Doing a Lean Startup Conference Workshop on December 4th in San Francisco - I Look Good In Design()

  • alanya


  • alanyam
  • Pingback: Managing Startups: Best Posts of 2012 |()

  • Matthew Kirk

    Ben, while I agree that one metric is easier to track against your goal I feel that it can be abused as well. For instance many e-commerce sites pick conversion rate as the be all end all metric.

    Really their goal is to maximize total profit per customer. But using conversion rates as the only metric would yield some interesting gaming of the system like faster checkout screens, when maybe the user needs more interaction.

    I really appreciate your insistence on setting goals first. It’s very frustrating when people dont :). Cheers!

  • Benjamin Yoskovitz

    Totally agree re: the risk of picking the WRONG metric. If you focus on conversion rate as an e-commerce company when you shouldn’t, you’ll have problems. But focusing on more than one metric doesn’t solve that problem necessarily; maybe it helps you avoid a dependency on a single metric, but it’d be better if you picked the right metric for your business and the stage you’re at and focus on that. Hopefully with Lean Analytics we can help people find the right metric.

  • Pingback: Managing Startups: Best Posts of 2012 (guest post)()

  • Pingback: Startup Strategies – Best of 2012 – Digital Cowboys()

  • Phanio

    Shouldn’t that one metric be what your customer wants – what your customer is willing to pay for?

  • Benjamin Yoskovitz

    Revenue per customer, for example, could be your One Metric That Matters, but it might not be. You might not be at the Revenue Stage yet. You might still be working on getting engagement and retention, increasing virality.

    The one metric could be churn (tracking how many / what percentage of customers are abandoning), which is an indication you’re not doing something right. You want the One Metric That Matters, especially early on, to be narrower than “what a customer wants” and it’s highly dependent on the business you’re in (your business model) and the stage you’re at.

  • Pingback: Weekend Reading: A Product / UX Rant, Choosing Metrics, and Setting Goals | Product Talk()

  • Pingback: Managing Startups: Best Posts of 2012 | andrewbruce01()

  • Pingback: Find Big Wins by Setting Exponential Goals | Product Talk()

  • Pingback: | Interesting Links for February 1, 2013()

  • Pingback: Lean Startup Zappos: How Zappo's Validated Their Business Model With LeanBullet HQ()

  • Pingback: Fun with dating and network matching models | Musings on strategy, technology, and design()

  • friv

    I am definitely bookmarking this page and sharing it with my friends.

  • friv3


  • yepi1

    In life we always need to keep the customers belong to their importance and me too. thank you

  • Pingback: Read Lean Analytics | Product Talk()

  • Pingback: Lean Startup and Lean Analytics in the Enterprise()

  • friv 2 friv 3 friv 4

    Thanks Diana. Analytics is definitely not owned exclusively by tech
    startups. All companies can and should use analytics to improve their

  • Pingback: How to choose the one metric that matters | A Founder's Notebook()

  • Pingback: Managing Startups: Best Posts of 2012 |

  • Pingback: The Importance of Understanding Your Best Users()

  • Pingback: pboxco()