54 Comments

Summary:

Overnight successes make for great stories, but that’s largely because they’re so rare. Success, the kind that leads to great products and businesses, is built on the foundation of a huge amount of hard work over many years and is achieved by continuous improvement.

baby_stepsMathematicians will tell you that the only way to learn math is to do math. Lots of it. The same is true in music and sports. While with math you quickly find out whether you’re right or wrong at a very atomic level with each problem you try to solve, with music a student listens to a song many times before she tries to emulate it — and then gets feedback on a note-by-note basis. And the same goes for sports — the stroke, the kick, the catch, the swing, the run and so on. Practice makes perfect, right?

Yet in business you often find people who have been doing something for a long time and just aren’t very good at it. Why? Lack of feedback. After all, imagine trying to solve math problems and waiting an entire year to get the answers, or hitting 1,000 serves and getting a summary of your performance at your “annual review” rather than after each serve or at the end of a game. Practice only makes perfect when there is frequent, high-quality feedback so that the right adjustments can be made, be it in math, sports, music — or business.

The Corporation and Feedback Systems

In certain disciplines, like engineering and sales, there is somewhat objective and frequent feedback. Your program compiles without an error and does what it was meant to do. Or you close the deal and make your quota. If, however, you’re in one of the many disciplines in which immediate and objective feedback is not available, practice may not lead to perfection so much as enforce bad habits.

Let’s say you’re a mid-level executive — a GM or product manager of some sort. More than likely, you’re measured by how well you interact with and present to your manager and senior executives. Consequently, you optimize to managing the bureaucracy (your boss in particular) rather than delivering the right product or service to customers. And so does your boss, and her boss, and so on and so on. Here the only thing that you’re practicing and perfecting are your brown-nosing skills. How can you expect to learn in an organization with that type of feedback and incentive system? How can such an organization, by extension, possibly produce excellence?

Product Development and Continuous Improvement

The organizations that produce excellence are those that continuously improve. The more granular and frequent the improvements, the better. For illustrative purposes, let’s compare two hypothetical companies that are going after the same opportunity with a similar product concept.

Company 1:  Large Web Company

It’s August 2009 and the annual plan has just been completed, a few months late. Regardless, the road map is set through August 2010, so it’s time to get to work. But new features continue to roll in from various executives, and since none of the older features are being dropped, the schedule is continuously moved out. Finally, after a death march, you launch your product — in October 2010. The press doesn’t like it. Millions of expected customers don’t show up. Somebody screwed up. Time to go back to the drawing board and start over again.

Company 2:  Web Startup

It’s August 2009 and you and a few friends are fired up to change the world. You decide that you’re going to launch your product in 90 days. You’re resource-constrained, so you add nothing to the requirements list and launch on time. The press doesn’t like it. Millions of expected customers don’t show up. Investors aren’t interested in what you have to offer. Every week you ship a new feature or two and learn what works and what doesn’t work. After several months of failure, some feature you cranked out late on a Sunday night leads to an uptick in traffic. Over the coming weeks and months you pull that thread. It’s now October 2010 and millions of people are using your product. You raise a hefty Series A and the press declares your company an overnight success.

This comparison in how organizations react to “failure” reveals the key reason why innovation comes from startups more often than large organizations (and why startups are often a very good training ground for product development and business talent). Large organizations generally aim to “get it right” from the get-go — an unreasonable requirement that leads to fear, posturing and endless delays. Startups, on the other hand, are just trying to get it right before they run out of money.

Human Resource Development and Continuous Improvement

Consider as well the feedback cycles in these two organizations. In the large organization, you likely get feedback when you do periodic releases — perhaps every six or 12 months — with the really important feedback saved for your annual review. In the startup, you get feedback after 90 days and then every single week. And there exists both carrot and stick in terms of providing honest and meaningful feedback — the carrot is equity and the rewards of creative ownership while the stick is running out of money.

Overnight successes make for great stories, but that’s largely because they’re so rare (and more often the stuff of fiction). Success, the kind that leads to great products and businesses, is built on the foundation of a huge amount of hard work over many years. Baby steps are the sure path to excellence in everything from product development to managing your own career — continuous adjustments based on frequent feedback lead to cumulative gains.

Mike Speiser is a Managing Director at Sutter Hill Ventures. His thoughts on technology, economics and entrepreneurship will appear at this time every week.

Related research

Subscriber Content
?
Subscriber content comes from Gigaom Research, bridging the gap between breaking news and long-tail research. Visit any of our reports to learn more and subscribe.
By Mike Speiser
  1. I like your math analogy. That’s why we have instant feedback on our online maths revision site. If only I could convince my kids on the value of continuous practice :)

    Share
    1. Thanks Wilf. Will check out your site.

      I believe that the vast majority of people have the capacity to be very, very good at mathematics. But to do so you must start early, because once a child falls behind it gets harder and harder on a relative (to peers and expectations) basis. Compounding learning is as powerful (or daunting) in mathematics as compounding interest is in investing.

      On your kids… why not buy a few workbooks from Amazon and get up every morning with the kids for a little dad / mom 1:1 time? Just 30-60 minutes each day before or after school will do it… That’s my plan ;–)

      Share
  2. Awesome post. Growth mindset/continuous improvement is indeed a great tool!

    Share
    1. Thank you Edwin.

      Share
  3. Good analogy! Successful companies follow this approach.

    Share
    1. Thanks, and I agree that great companies are learning machines. Even Apple, despite the assumption among many that Jobs knows everything before he ships a product. For example, the iPhone App Store was a product of learning…

      Share
  4. Pretty relevant post in any company. I think startups do have feedback, but to retain people the need for frequent review is very important.

    Share
    1. Thanks. Good point that feedback is about retention in addition to products and so on.

      Share
  5. I worked in process management for about 10 years of my corporate career. Plenty of people didn’t like it while others begrudgingly accepted it. If we keep doing things the way it has always been done, we’d never grow or innovate.

    Sometimes you discover a faster and more efficient way of doing things. Everyone — including freelancers — can always be on the lookout to create and improve their working processes. Thanks for showing how important it is and doing it well.

    Share
    1. Thanks for the comment Meryl. I agree with your insight that we can all get better at just about everything we do.

      As I was writing this I was thinking about how various signals help me tune my writing. Number and quality of comments, Tweets, Delicious bookmarks, number of emails I get from friends… These indirect signals help me get better and better (hopefully).

      Share
  6. Great article!

    Hard and continuous work is highly underrated nowadays as it doesn’t make a good (short) news story. As a CEO of several SUs’ I can sya, this is one of the main lessons young companies and PM’s should learn.

    Having looked at the authors resume, it seems this article is a results of many years of hard work- Keep on the good job your doing Mike..

    Share
    1. Thank you Alon.

      The two things I have found in common amongst the people I respect most is humility and focus. You cannot learn if you’re not open to finding out where you are wrong. If you change functions and/or industries you may carry over some of your learnings, but there is some amount of starting over.

      On a personal level, hopefully a focus on these two things in addition to hard work will help me do something meaningful in the coming decades…

      Share
      1. Thanks Mike,

        The fact that you have so dligently replied to most of the comments here says ait all..

        Anyway we can follow you on twitter?

        Alon Atsmon

        Share
  7. This is interesting reading and an important subject, however it touches only on the very surface of continuous improvement and neglects any standards and methodologies to create a continuous improvement system. Anything from ISO9000 to Six Sigma, Lean or CMM in their various reincarnations.

    Share
    1. Agreed on CMM, Six Sigma and all that. But these are way more than some small businesses’ need. To include all of this would take a series.

      Share
      1. I don’t disagree and not necessarily every small business should implement a full fledged CMM, Six Sigma, et al. However, at the very least being familiar with the body of knowledge and understanding the principles and practices supporting them will make any continuous improvement program better and will help avoid reinventing the wheel.

        Share
    2. Thanks Haim and Meryl.

      I agree that this only touches on the surface and appreciate your specific comments Haim. As Meryl noted I’m somewhat limited on how much I can fit in <1000 words. However, I'm also limited by the depth of my knowledge on this topic. There are so many different methods and angles that I personally couldn't write anything close to a complete synopsis of them all.

      In addition to the process approaches you note, I would also point out that agile programming is at its heart about continuous improvement. And just about every startup these days uses agile (my implicit point in the comparison of the two companies).

      On the philosophy side, I would encourage people to read Zen and the Art of Motorcycle Maintenance and The Goal. I read both over a decade ago and this conversation has inspired me to go back and re-read them.

      Share
  8. Mike, another inisghtful article. I have enjoyed your many missives over the past few months. IT is funny you mention “The Goal”. You recommended that book to me when we were in college. I read it then and it is still on my shelf. I think I will pull it down and give it another read. dm

    Share
    1. Thanks Dan! The Goal is great, but I like Zen and the Art of Motorcycle Maintenance even more. Worth a read. Thanks for the comment.

      Share
  9. Cool observation, and definitely something I felt when joining salesforce.com in 2003 where we did quarterly releases, versus my previous company that tended to plan out a year or more.

    Also interestingly it’s the same in AI as it is with human/organizational learning.

    In AI and Machine Learning there’s a branch of research on Reinforcement Learning, or learning from delayed rewards. It deals with how an intelligent agent can learn to modify its behavior to maximize total reward, in an environment where rewards are potentially quite delayed from the action(s) that ultimately caused the reward.

    Reinforcement learning in complex regimes (where experimentation is expensive) becomes intractable without factoring the problem into subproblems, and designing “pseudo-rewards” that you hope will cause the right behavior in the end.

    In your organizational learning example, the mid-manger’s pseudorewards (pleasing the boss) are all screwed up, and that sends the whole product and company in the wrong direction.

    As computers get faster and industries evolve, some endeavors that historically have been solved via long-range plans with pseudorewards and organizational structures can now be solved via direct rewards and math.

    For example, take online advertising — there’s a huge array of potential marketing actions that a company can invest in. Ultimately the goal of marketing is to drive sales, but it’s perceived as too hard to mathematically solve for the right long-term sales-maximizing marketing policy, so pseudorewards for things like good brand campaigns are designed.

    But also interestingly, online marketing is actually a regime where fast experimentation is very cheap, so it’s actually possible to test interactions among brand and direct response campaigns in a way that solves for true reward versus pseudoreward.

    How much does it help to “tee up” a direct response ad by showing a drip campaign of brand ads for weeks or months prior? What’s the optimal frequency and recency or mixture of brand and direct response ads? How do multiple publishers and ad networks interact? Who deserves credit for driving a conversion?

    My bet is that success in online advertising will fall along these lines — the fast experimenters vs the slow planners.

    Share
    1. Awesome comment, thanks George.

      I’ve been thinking a great deal about machine learning. I wonder if the putting *everything* effectively in memory — the cost of DRAM is dropping like crazy and the emergence of SSDs means that spinning media will be a thing of the past for high IOPS environments — allows for far more rapid learning?

      Sure, some firms have already done so, but at a massive cost or for relatively small capacities. With the cost of putting petabyte level data stores in memory or on SSD dropping at 50% every 9-12 months, won’t most machine learning become a real-time process?

      Share
  10. Google (Chrome/mail, etc) and SAAS deliver a good job of incremental updates. MSFT on other hand makes products in 4 year cycle ?

    Share
    1. Totally. Google is one of the few big companies that get continuous improvement. Google Labs was a brilliant device to allow a big brand the opportunity to ship imperfect products. But it seems that even the mighty Google is getting more conservative these days, no?

      Share

Comments have been disabled for this post