Forum Moderators: buckworks

Message Too Old, No Replies

My new favorite e-commerce book and usage based segmentation

Some ideas on how to boost your conversion rate

         

tolachi

8:35 am on Mar 8, 2004 (gmt 0)

10+ Year Member



I read a new old book over the weekend called "Submit Now: Designing Persuasive Websites" by Andrew Chak. I think it is the best primer on e-commerce that I have encountered thus far. The writer is an obvious admirer of Steve Krug and the book has a similarly simple layout and straighforward delivery. I thought I would post about it since I couldn't find mention of it or some of the topics it covers in site search.

Much of the information is just a coherent distillation of best practices and common sense. However, early on he introduces the concept of occasionaliztion, an idea that originated in a study by Booz-Allen & Hamilton and Nielsen/NetRatings, which he defines as:

segmenting users not by their demographics, but by the tasks they want to accomplish during their visit. In other words, we shouldn't be defining our users strictly according to their age, occupation, and shoe size. Instead, we should define users by whether they are at the site to browse or they know what they want and want to transact right away.

Chak distills these 7 distinct groups into the more easily digestible, "Browsers", "Evaluators", "Transactors", and "Customers". He proceeds to talk about how a site should be able to address the needs of all 4 of these segments in order to maximize it's potential revenue. For example:

Browsers aren't sure what they want yet, they're looking for a product or company to work with. You may need to recognize that they are unfamiliar with your site, introduce your company, or provide easy access to site search.

Evaluators know what they want and are trying to find the best place to get it. Anything that can help differentiate you and your product. Great content, pictures, and easily digestable stats.

Transactors know what they want and where they are going to buy it. Make it as easy and sweet experience as possible for these people to buy your product. Short simple checkouts.

Customers repeat business, order tracking, referring site to friends. Order tracking, incentives to refer you to friends, easy reorder lists for items that are purchased periodically.

Looking around at some of the most succesful sites on the internet, like Dell and Lands End, I see that they have made an effort to address the different usage states he mentions. I wish that Chak had more case studies to cite and that he could have given in a more depth review of the material.

More on the Occasionalization study

The original study on occasionalization, which can be found here [extfile.bah.com], defined users by 4 characteristics, Session Length, Category Concentration, and Site familiarity. We have three of those available, we can't really determine Category Concentration since we probably won't be sure which other sites the user has been visiting. But three should be enough to start to define what type category a person belongs in. For example if someone is on their first visit you may want to have a link to "About Us" or some similar document that talks about how great your company/products are in your sidebar. However, on the 3rd or 4th visit when the user already has something in their cart you may want to start using that same space to cross sell them on proucts related to what is in their cart.

The study offers some suggestions on how to tailor the a user's session based on which segment they fall into. The advice is quite generalized, and I don't necessarily agree with all of it, however I am sure that if I had that info about one of my customers I could put it to good use.

As the study concludes:

The challenge for the e-tailer is to use available technology to detect which occasion a user coming into the site may be in, and to use that information to trigger an interface geared to that occasion.

Since it gives a thorough outline of it's methodology and the exact numbers that correlate with each usage group, we are left with a good starting point if we wanted to start identifying which state of the usage a visitor is in. I'm wondering what other people think of this and whether they have had a chance to experiment with these ideas.

Herath

6:05 pm on Mar 8, 2004 (gmt 0)

10+ Year Member



Thanks Tolachi. I read this book today. Pretty darn good book.

your_store

6:36 pm on Mar 8, 2004 (gmt 0)

10+ Year Member



A rep from Overture discussed this issue during one of the sessions at PubCon. He was more concerned w/ determing the a user's place in the buying cycle by examining the search phrase. You should def. check it out once the presentations are posted to WW.

BTW, thanks for the book tip. Looks like an interesting read.

cfx211

9:04 pm on Mar 8, 2004 (gmt 0)

10+ Year Member



The book looks interesting and I am sure its worth the read.

After reading the article though, my first thought was geez this is great for doubleclick, but doesn't help the individual webmaster all that much.

This is because occasionalization relies very heavily on the string and behavior of sites visited when someone sits down at their computer. Outside of the referring url of the first request, you know nothing about what this person has been up to.

Yes you may be able to deduce their behavior type based on how quickly they are requesting pages on your site, but that requires them to have requested a few pages already. If they are already into and using your site, you don't want to change your nav structure in mid-session. It would just confuse the user.

I am a big believer in usage based segmentation, and I think you can only influence your overall site performance if you view it as the sum of several different segments, I just think occasionalization is probably something for consultants to make big clients feel smarter when making network media buys.

For individual sites, I would love to hear what you guys use for segments.

We use a combination of account data (when available), and previous visit history. Our site is very focused on an event and all of our site behavior is more or less a countdown to that event so we do a lot of our usage segmentation based on where in the preparation timeline someone is to that event. We also find the number of visits that person has made to the site already is another good determinant of what they are up to.