What metrics should you consider when calculating the value of your Web site, or your return on investment? Let’s start with performance and availability. To say that slow pages are the kiss of death doesn’t take all the possibilities into account. When National Semiconductor asked its design engineer customers about the need for speed, site visitors said that anything more than eight seconds for a navigation page and 30 seconds for a data sheet was intolerable. But it’s not just about minimal graphics; it’s also about process. When Kinkos reduced the number of steps to upload a file for printing from 12 clicks to three, sales went up 25% overnight.
You should also consider the type and number of pages your site visitors look at during a single visit. Are more pages better? Maybe. When visitors first arrive and are investigating the depth and breadth of your offerings, more pages is good. But if they are trying to solve a problem by slogging through the quagmire of your"frequently asked questions" pages that are intertwined like the roots of swamp trees, more is definitely not merrier.
Looking for patterns
The researcher, the comparison shopper and the buyer all have their own traffic patterns, and it’s up to you to serve them according to their needs. If people are always clicking from the product page to the warranty page and back, it’s good stewardship to put the pertinent information on the product page and save them a click. Watch where they go and then do what you can to make their way easier next time.
The final consideration is how often people visit your site—a combination of recency and frequency. It’s great that somebody came to your site 10 times in one day, but if that day was six months ago, it won’t do you any good.
The most desirable recency-frequency tally is different for different types of sites. If your name is Yahoo! or AOL, then you want all the people, all the time. If your Web site is located at www.urnsandcaskets.com, you may feel a little uneasy about the people who come back day after day and stay a while.
Recency-frequency alone is not nearly as interesting as the changes in that category for a single visitor, over time. Let’s say you sell cars or computers or dishwashers to individuals on your site. Let’s say one of those individuals shows up on the first of the month, on the fifth, on the seventh, three times on the ninth, seems to hang out most of the day on the 10th and then never comes back. If you guessed you missed a sale, go to the head of the class.
Had you noticed this traffic pattern, you might have offered free floor mats, free delivery or an extended warranty gratis. You might have noticed that the customer was drooling over your wares and moved in for the close.
What pattern is the best indication of imminent procurement? We’re back to the consultant’s favorite phrase: It depends. The solution is to review the site path taken by those who completed the purchase and watch for it again. Your site can make that special offer when a potential customer visits a certain number of times or looks at particular pages.
And then you can go one step further. A company in Boston called Genalytics (www.genalytics.com) has a clever piece of software that will determine the fewest attributes common to those who performed the desired act (purchase, register, subscribe, etc.). It sifts through mountains of data about your customers so you can more quickly recognize the ones who are statistically most likely to buy, and it can tell you which offer is the one most likely to make that sale. Nifty.
You can’t manage what you don’t measure. And now, measuring is making more managers merrier.
Jim Sterne has been following Internet marketing and customer service for the past eight years and has written four books on the subject; his focus now is e-metrics. Sterne can be reached at www.targeting.com