Your job as a marketer or site operator, is to optimize the efficiency at which your machine is working.  You want it to produce more value for less cost.  To manage the efficiency of this machine, you need to understand two things very clearly: how does your site contribute value to your business; and what are the visitor behaviors that lead to that value.

If you don’t have clear understanding of those two things, or buy-in from all of your stakeholders on those two things, no web measurement tool in the world can help you.   When you do gain a clear understanding of those two things, and when you become comfortable with them — when they are second nature — it will be easy to figure out what you should be measuring to help you manage and tweak your machine.  When you know what you need to measure to manage and tweak your machine, web analytics tools will cease to seem overwhelming and daunting.   You’ll know just what you need them to do, and, conversely, you’ll know what features of your tools you can ignore as simply more noise.

In my experience, friction is often introduced by the very features that were intended to reduce friction and drive conversion and revenue performance improvements.  The result?  As friction goes up, return on investment tanks, and so, too, does revenue.  That’s why, to maximize success, it’s critical to isolate and understand the bottom-line impact of any new site feature and to eliminate (or modify and retest) any features that drag down performance of the site.

Here are 5 things you can do to reduce friction on your site.

Allow Guest Checkout

It’s easy to allow people to check out as guest, without requiring them to create an account. Sometimes people are positive that they’ll never come back to the site, and don’t want to bother.  Wouldn’t you rather have their money than not have it because creating an account adds too much friction?  You can always offer to “save” their “info” for future orders at the end of the process.  If they take you up on it, you both win.  The customer got their friction-free checkout process, and you still got a registered customer.

Allow authentication with 3rd Party Authentication Providers

With the increasing popularity of 3rd party authentication systems such as oAuth and OpenID (which allow you to log-in to a site with your Facebook, Twitter, Google, or any number of other site IDs)  inability to use these credentials to authenticate on your site will increasingly be perceived as friction — you standing in the way of how consumers want to be able to interact with you.  There are other good reasons to adopt 3rd party login, too.  Principally, the data that becomes available to you from most 3rd party credentialing services about registrants is more robust than you’d get from consumers by having them register directly with you.  While it can be complicated to implement more than one 3rd party authentication protocol, companies like JanRain make it easy to authenticate with and consume data from multiple authentication providers with a single API that abstracts the functions from multiple providers and normalizes the data that comes back into a unified format.

Offer only highly relevant up- and cross-sells

It’s important to have a considered strategy around cross-sells and up-sells.  There are many product recommendation companies who can provide the technology for you, but only you can drive the strategy and decisions around where on the site products should be recommended, and what the goal is for each location on the site.  Be explicit with your goals, because without explicit goals you can’t measure the performance of the recommendations against goals.  On a product page, your goal for up-sells should be something like “move shoppers to higher margin items without negatively impacting overall revenue or conversion rate.”  Since your strategy is to drive up margin and revenue, don’t suggest products at a lower margin.  It’ll be bad for your business, and will add unnecessary cognitive weight to the buying process.

Even if you only offer products that make sense from a business standpoint, the potential for friction is still there.  If you offer too many choices, you may end up simply killing the sale by introducing cognitive drag.  If the customer can’t make up their mind because there are too many choices, you both lose. The added cognitive weight of additional suggested items, which take attention away from completing a purchase, is likely to cause some people to bail on the sale.  The product you’re suggesting has to make up for that either by converting at a higher than average rate, by having a significantly higher margin, or both.

Sometimes it may be better to suggest nothing at all.  Testing well will tell.

Eliminate or mask complex cart editing and other unnecessary features during checkout

It’s tempting to offer complete item editing functionality in the cart, and through the checkout process.  It is often wise to check this temptation, however.  How often am I really likely going to change size or color in the cart?  Does the need to completely change an order happen often enough to overcome the cognitive drag — the friction — introduced by the presence of the capability?  Changing quantity makes sense in most contexts, but simple testing will reveal the true impact of more complex functionality on conversion performance.

If you must retain the ability to edit all properties of an item in the cart, mask them behind a well placed “edit” link or button so most people don’t have to deal with them.

Deploy no new features that add drag

As noted at the beginning of the post, friction produces drag on conversion rate and revenue.  As such, there’s an easy test for friction, and you should be employing it rigorously.  Every major change or new feature on the site should be tested for added drag.  Features or changes that add drag should be yanked from the site, and either tossed in the trash bin or re-thought out and re-tested.  Under no circumstances should you leave a feature on the site that adds even the slightest drag.  There isn’t an argument in the world that should persuade you to leave it.

Your friction test is a simple A/B split test, with equal percentages of your visitors getting A (the site as it is now) and B (the site with your new feature).  If B (the new feature) drives higher conversion and revenue than A, the deploy the feature globally.  If not, kill the feature and move on.

You don’t need any fancy multivariate testing tool to run this test.  You simply need a web analytics package with the ability to pass either a custom parameter for segment reporting, or the ability create a segment based on a non-custom parameter.  Most of the work is done with JavaScript and cookies (identifying who is A, who is B, who is not participating in either group, and setting the appropriate parameter).  The segment parameter allows you do do the reporting in your web analytics tool.  I’ve done this in both Webtrends and Coremetrics, and I know it’s possible in the other major tools.

Of course, there are many other things you could be doing to reduce eCommerce friction, but this should keep you busy for a while.  And if you start to think that you’ve reduced all the friction possible, you can always go back and start retesting some of your site features for friction.  Friction is a fluid concept related to consumer expectations, wants and paradigms which themselves are dynamic and change over time, sometimes very rapidly.