I am in Love with a Man, Mr Genichi Taguchi

January 10, 2007 - Get free updates of new posts here

I have to come clean and tell you about my love affair. I can’t bare to keep it inside any longer. Let me tell you about the love of my life.

Mr. Genichi Taguchi is a great man. He showed me the way of building successful businesses. I am indebted to him forever. He taught me one critical thing:

Measure Everything!

There can be no progess without knowing what you are improving from.

This is something I think a lot of just assume. How many times have you built a website or done marketind and heard:

  • “Oh yea, I know people are going to like this better”
  • “This worked in the past and should work again”
  • “Come on Noah, trust me. I am always right”
  • “Of course a HUGE red button makes sense there”

How can you prove what you did was right? Hint: You have to measure it.

Mr. Taguchi created a method that helps you test multiple sections of a web page and analyze which things are really making a difference. It is not so much his test that has made me love him like no other but more the idea that marketers, product managers and creators in general need to consider using metrics and quantitative data when making decisions.

Read this article from Offermatica on how Taguchi works. Also, the CEO’s site Landing Page Optimization has some good articles on page optimization.


If you enjoyed this post or not, get free updates by email or RSS.

Leave a Reply

Your email address will not be published. Required fields are marked *


16 responses to “I am in Love with a Man, Mr Genichi Taguchi

  1. Sean Tierney Reply

    noah, completely agree: you can’t improve anything until you can get a baseline measure on it, a reliable metric in place and systematically make changes controlling for other variables and attributing results to those things which you change.

    Check out “Google Web Optimizer” if you haven’t already seen it- it allows you to serve A/B marketing tests via javascript on your sate. It automatically tracks conversion ratios and intelligently serves the better-performing layouts / graphics / design. Very cool stuff.

    sean

  2. Dave Morgan @ SiteSpect Multivariate Testing Reply

    Hey Noah, sounds like you have seen the light!

    Let me add that not only is measurement key, but knowing who to test & measure is also critical.

    Think about it. Are you trying to optimize for “everyone”, or only for those visitors who are likely to become a customer? If you only sell chocolate ice cream, make sure you optimize for “chocolate ice cream eaters”, not “all ice cream eaters”.

    Another multivariate testing tool to check out — which lets you test and optimize by segment — is http://www.SiteSpect.com

    Best of luck!
    Dave

  3. Dennis Reply

    Let’s not overstate the truth. It’s actually possible to improve things without measuring them — it’s just not as predictable or efficient. Have your tennis skills improved? did you measure them as Taguchi would have — or did you trust your informal (though perhaps somewhat data-based) assessment of your progression?

    Did Apple measure something in order to improve MP3 player design? I’m sure you can think of other examples as well.

    That being said, there are many things that ARE measurable that typically aren’t measured — and measuring them opens entirely new possibilities to improve and optimize them.

  4. Noah Kagan Reply

    Sean, I am dying to try out the Google optimizer and applied a few times. I am holding my breath while I wait.

    Thanks Jason and Ralph!

    Dave, Your product seems interesting. I will email you because I am working a lot on a/b and multivariate testing and would like to evaluate your service.

    Dennis, My mentor and old Intel friend, you are right again. I do think there is a TON of qualitative data that needs to be considered. Taking everything mathematical would not always produce the best results. I think this discussion is unfair with you considering you have an Masters degree on the subject;P But I do appreciate you stop by. I hope the kids are great=)

  5. john Reply

    Hmmm. Interesting point, and one certainly containing a grain of truth. But as another smart guy named Albert Einstein once said,

    “Everything counts, but everything doesn’t need to be counted.”

  6. Dave Morgan @ SiteSpect Multivariate Testing Reply

    Size matters if that’s what’s important to you. 🙂

    If you want sales, measure “completed transactions” (plus all the funnel steps that lead up to the transaction) …

    If you want recurring/loyal visitors, measure the frequency with which your target audience visits your site …

    There are innumerable things you could measure about your site (or anything). Some matter, others are just peripheral.

    Articulate your goals — measure what counts towards achieving those goals — experiment — learn — optimize. Then, repeat!

    DM

  7. Noah Kagan Reply

    Dave you really hit the nail on the head.

    What are you aiming for? So what your click-through-rate goes up if your sales stay flat. The line I think most marketers need to go by is

    “Ready-Fire-Aim-Fire-Aim-Fire-Aim….”

  8. Dave Morgan @ SiteSpect Multivariate Testing Reply

    Love it!

    But that’s a great example — what matters more? Clicks? Or sales?

    I can think of a number of SiteSpect test campaigns where our customers set out to optimize landing pages, and found that certain page adjustments sent CTR through the roof. BUT, of those incremental clicks, far fewer visitors actually converted into buyers. Optimize what matters!

    On a similar note, not all customers are created equal, so not all sales are the same. There is such a thing as a “bad customer” — the one that you *don’t* want to buy, or at least for whom you don’t want to optimize your site.

    A bad customer is one who is more likely to return a product, or requires abnormally high amount of customer service. By tracking post-sale data like returns and call center volume, SiteSpect users have found some surprises … that some “winning” promotions or site changes actually lose them money in the long run!

    DM

  9. conversion rate optimization Reply

    Many companies that have applied basic SEO principles to standard web site content see substantial benefits.