Averages aren’t great for action. Just ask my friend Avinash Kaushik. As he likes to say, the average income between himself and Bill Gates is an astronomical number.

The point being, in web analytics, there’s very little we can act on when we review averages. Even really good analytics—click-throughs, bounce rate, conversion rate, customer satisfaction—become little more than vanity metrics in the hands of analysts with a penchant for averages. What, for instance, will knowing that you have an average bounce rate of 46% tell you? What will you need to change? What functions and pages will you have to prioritize?

Data on the web is about action. It’s richer than what companies advertising on TV are getting. It’s richer than print advertisements by a mile. But that richness also means that averages won’t cut it—you need to segment, and then keep on segmenting until the numbers tie directly into actions.

But where do you start segmenting?

A good place to start is the online optimization model: acquisition, conversion, and retention.

Acquisition

In acquisition, you can focus on the different ways people get to your site—organic search, paid search, social referrals, referrals from industry sites, and direct visits. Then you dig deeper. For instance, you can segment to look at organic search, and at people who are getting to your website using branded terms and non-branded terms.

Then you can segment deeper again. Review the non-branded terms, and which areas of the site visitors head to most. These visitors are your potential new clients—people who do not know you and your brand, but still end up on your website. The pages that they visit a lot are your top of the funnel entry points, and you do not want these visits to bounce.

Then you can segment deeper still. Of those pages, which ones have the most referrals and the highest bounce rates? At that point, you’re no longer working with a “status” metric. Now, you’re looking at actionable data. You can find the top five pages you need to fix, the top five search terms they cater to, change elements on the page, and test away. The end point isn’t an Excel file or a presentation. Instead, it’s a proposal for an action plan, and there’s a dollar value tied to it.

Conversion and Retention

Just as you should track details of how people get to your website, you should do the same once they are already there. Let’s say you know what the top tasks are on your website. The end point of the task is a visitor getting what he or she wants from your website, and the ratio is your task accomplishment rate.

Then segment deeper. Check the commonly done tasks on your website and find the one with the worst task accomplishment rate. Cross-reference that statistic with how users navigate your website. Deep segmentation like this can uncover critical usability failures within a website.

It could be your engine for on-site search. It could be your left navigation. It could be the top navigation. Maybe your mega-menu is the issue. Whatever you yield then is something you can launch a proposal to fix—not a general trend line for your site.

Again, these are just examples. But the point is, do not use averages when analyzing the web. They will get you vanity metrics, and nothing more.

You should segment until you have an action plan. Ultimately, it’s the only way to use the science to move the needle.

Tim Ash is the CEO of SiteTuners, a leading conversion rate optimization agency that has helped over 850 clients worldwide. He is chairperson of Conversion Conference, the first conference focused on improving online conversions, and is a highly-regarded presenter at SES, eMetrics, PPC Summit, Affiliate Summit, PubCon, Affiliate Conference, and LeadsCon. Tim is author of the bestselling book Landing Page Optimization and is a contributing columnist to many publications. You can connect with him on Twitter and LinkedIn.

Lie Button image courtesy of Shutterstock.