Navigation Functionalism and Web Analytics

    September 11, 2006

Every shopping experience is framed by a specific set of navigational elements. This is as true in traditional retail as it is on the web.

You’d probably be surprised at how carefully traditional retail has studied the problems of navigation at almost every level. Store planners have made considerable study of product groupings and aisle placements. It isn’t by accident that the two things you want are often at opposite ends of the store. Or that the ultimate impulse buy – candy – is sitting right next to the cash register along with reading material for your bored eyes. Retailers try to understand the various shopping basket mixes that drive shoppers – and how to place goods both next to – and away from each other – to promote maximum cross-sell. Impulse buys need to be next to common necessities. Common mixed baskets need to be widely separated to force shoppers to traverse aisles. Aisle endpoints are loaded with impulse buys. It all makes perfect sense, and it doesn’t happen by accident.

One level up from this, you can think about how mall escalators are often arranged. They aren’t positioned for maximum convenience – but to insure significant store traversal. Indeed, even the arrangement of a mall is a matter of considerable study. Anchor stores (the core shopping stores that draw visitors) are placed at the ends. Boutiques line the connecting ways – drawing in pass-by traffic and supplementing overall traffic by adding to the experience.

Every website faces a similar set of challenges about how to move people from one area to another, where to place add-sells, when the shopper needs to rest and when the shopper needs to move, and how to get the shopper to the area(s) of the store they want. And the Functional page type that does most of the heavy lifting in this regard is the Router page.

Here’s the basic idea as stated in the whitepaper: Router Pages are those whose primary purpose is to move visitors into particular sections of the site. The presumption is that there is fairly substantive information about what the visitor might be interested in and these alternatives are presented as navigational elements in the body of the page.

Probably the most common Router pages exist as “mini-homes” underneath the top level navigation from the Home Page. Go to most web sites and pick an area like Products or Services or Support and you get a page with lots of drives to various sub-sections of the site.

One of the things that defines this page-type is that the visitor – by arriving – has indicated some interest in the topic. That’s why the home page is not a pure router – and why Search is a special case of router. So if a visitor arrives on your Main Products Page, you expect that visitor to drill down to content about Products. Another aspect of the Router page is that the page itself is generally light in content to “Convince” or “Close” the visitor to buy. It is trying to get the visitor to the right “Convincer” pages – not sell them itself.

So how do you measure Routers? By how well they move visitors to the Content they are supposed to. That means that for Routers (unlike most other pages) the type of action you’re most immediately interested in is what happens next. Routers shouldn’t be measured against conversion – because they aren’t responsible for selling. They shouldn’t be measured for return visitors – because they aren’t trying to get visitors to come back. In other words, you don’t measure the performance of your escalators by sales – you measure how often they deliver traffic to each door.

For Router Pages, that means that the basic KPI you’re focused on is the percentage of visitors who followed appropriate routes versus the percentage who didn’t. We call appropriate routes Body Routes, since they are typically the routes linked in the main body of the page. In addition, we like to measure various groupings of “bad” routes including “sideways,” “back-outs,” and “exits.”

Sideways routes are typically other top-level navigation routes. These aren’t horrible, in other conceptual schemes they wouldn’t raise an eyebrow. But if a router page shows lots of top-navigation next steps then it isn’t doing its job properly. Back-ups are another type of sub-optimal route – cases where a user goes from home-page to router and then back to home page. Again, this isn’t what you want of these pages. The last “bad” route is an exit – and, of course, an exit is almost always bad.

This grouping of routes is one of the cases where you have to do some extra work in the measurement. When you classify a page as a Router, you need to decide what the “intended” routes are. Grouping routes into the Intended, Sideways, Back, and Exit buckets makes for a great reporting template – one that meaningfully captures the comparative performance of each page as well as providing real information about how it may be sub-optimal.

The Functionalist KPIs also include a set of measurements for Re-Surface behaviors – cases where a visitor drills down on the Router Page as intended, then comes back up to the page.

We like to break-out these cases specially, because we’ve found that the performance for the page (especially for exits) is strongly effected by this behavior. Where sub-pages don’t do a lot of cross-linking, visitors may re-surface then exit. This can make a Router page look much worse than it is. In addition, re-surface behaviors may reveal useful information about where a visitor goes next and whether personalization of the page at this point might yield dividends. It’s quite possible that on re-surface views, a Router page should include a “Closer” element to try and drive the sale.

By far the trickiest aspect of router pages is measuring (and separating) re-surface from initial land behavior. This isn’t always an issue, and before bothering with a more complex analysis, the analyst should check and see the percentage of visits that contain multiple pages views of a Router.

One method for studying re-surface behaviors is to create segments based on visits with a single page view for the target page and those with multiple page views. By comparing the next steps for the single page view segment with the total, you can see the choices visitors made differently when re-surfacing.

Web sites, of course, aren’t like traditional retail in many respects. Visitors can jump will-nilly from here to there. They are more likely to be single product shoppers. Your competitors are always just a single step away. So you can’t expect to make the same choices as you might in a bricks-and-mortar world. But the lessons learned in traditional retail can help clarify your thinking about what you are trying to measure and how you think about the elements of your web site. You don’t measure your escalators by conversion – and you shouldn’t measure your Router Pages that way either!

Add to | Digg | Yahoo! My Web | Furl

Bookmark WebProNews:

Gary Angel is the author of the “SEMAngel blog – Web Analytics and Search Engine Marketing practices and perspectives from a 10-year experienced guru.