Eric Peterson recently revised Avinash’s 10/90 rule of web analytics to the 10/20/70 Rule for Achievable Web Analytics Success. I agree with both that the key to value and success in web analytics is actually taking action on the data. You don’t get ROI for web analytics by just distributing reports; you actually have to do something with the data

Most people will agree that you need three things to extract value from web analytics:

1. You must have clean, not accurate (that will always be a challenge), data.
2. You must have a talented person who can convert that data and into insight.
3. You must have a talented person who can absorb that insight and act on it..

And you’ll need to follow this process over and over again to get continuous improvement. This is what everybody agrees to be the formula. The truth is that this ONLY works when you have all 3 ingredients.

It’s not easy to find all three, right? There are only a few handfuls of companies that have managed to find all the talent. I wonder how long they can manage to retain it. As Anil points out, there are plenty of people looking to hire these people away from you.

Anyone who’s been in business long enough knows it’s virtually impossible to scale talent significantly. The secret in every industry category is process, people, then tools. Why do so many companies do this in reverse? They buy the tools or tactic du jour (a shiny object problem), then they try to hire someone (that person will “know”) and they never get far enough to actually develop process or get value (the tool didn’t work out).

Think of web analytics as accounting for the Web. Most companies would go out of business (or be unhappily popular with the IRS) if they looked for talent and disregarded standard accounting processes. When everyone is trained in the right process, you can always try and find people who are more talented in accounting. If you can’t find your own accounting genius, then at least you can substitute with someone familiar with the process to get the job done–even if it’s not very creative. Ideally, all your processes should be designed to be utilized by those with less talent, so they can scale.

The reason we at Future Now developed Persuasion Architecture™ (pdf)–a Six Sigma-like process that provides blueprints to plan, measure, and improve your online sales and marketing–is because after years of working with clients optimizing their sites with web analytics and A/B & multivariate testing tools, we realized clients would hit a plateau where they could not break through the optimization brick wall. It didn’t matter if they were retail, B2B, B2C, media or service organizations. They all suffered the same fate.

The main reason for this fate is not because they didn’t have the tools or the talent, but because they didn’t have processes in place to make sure that every bit of marketing they wanted to measure was planned and implemented with measurement in mind. They needed to sort out “fine” signal from noise. When the signal was loud and clear, the low hanging fruit of optimization, it was easier to detect. But without a process to define what signal actually means up-front, it becomes increasingly difficult to listen for it.

Eric Peterson was the first to write about Persuasion Architecture in a book called Web Analytics Demystified. And I’m continuously grateful to Jim Sterne for asking us to keep trying to explain it to his audiences at the Emetrics Summits. Although we haven’t always done the best job explaining it, Persuasion Architecture solves this critical “process, people, tools” dilemma.

I’ve written volumes on Persuasion Architecture; but because it’s made up of so many disciplines, people have a hard time grasping it without actually seeing it. I’m going to summarize the process again:

• People do things for their own reasons. So, first clarify what makes different kinds of people click. This is segmentation based on psychographics and linguistic preferences, since that is how people actually navigate and make decisions on the web. (Uncover your Personas)
• Model a blueprint of those clicks to plan the experience that helps your Personas buy the way they want to. This is a click-based experience model, unlike the wireframes people use, but more similar to the way developers plan interaction states in software.
• Build your website from that model which defines the responsibilities of every word, pixel and click. This is where it’s valuable to have talented marketing and creative staff or resources. Could you go to your site’s homepage now, click on any link and then tell me why any phrase, link or image exists? Which market segment should it appeal to? What action do you hope people in that segment will take after reading those words, seeing that image or clicking that hyperlink?
• Once every business decision and click is documented in the model, development times can decrease by over 30%. You’ll recover more than the time spent by reducing the number of iterations copywriters, designers and developers go through in more traditional website design.
• Learn from every click you’ve modeled to see if your Personas click as you expected (the virtually neglected but real value of web analytics).
• If the clicks aren’t what you expected, continuously improve what to click and what the clicks say. You do this by first testing your execution (try variations of the same headlines, copy or images to accomplish the same objectives you specified in the model in different styles or formats); if you try several variations and you don’t move the needle, you should re-examine the assumptions that went into defining that particular piece of the overall objective.

Like accounting, it’s deliberate, and less exciting creatively, but anyone can do it. And anyone can then use web analytics to continuously improve their web marketing. Isn’t that what companies are looking for?