June 18, 2002
    Case Study
    
    
    
      
        
          | SUMMARY:
            We have already had a bunch of reprint requests for this article since it came out on Tuesday.  Seems like everyone wants to share it with their own private newsletter. Here is the scoop: late last year Kelley Blue Book, which is the most popular site for millions of US car buyers to use for research prior to their purchase, conducted a unique visitor metrics measurement test to learn where people were clicking and scrolling. The results were shocking.  Big surprises. Hear all about them and how Kelley Blue Book was able to grow profits from data: |  | 
      
     
    
      CASE STUDY
CHALLENGE
 Ok, on the face of things Kelley Blue Book (KBB) would 
not appear to have any challenges at all.  The site, just named 
(yet again) to Yahoo! Internet Life's annual list of "50 Most 
Incredibly Useful Sites" gets loads of traffic from the very 
people that advertisers with money really want to reach:
millions of car buyers intent on researching an upcoming 
purchase.  
In fact, VP Marketing & Business Development Stephen Henson 
admits their biggest problem is one that most online publishers 
would love to fret about: "We sell out.  We don't have enough 
inventory."  
However, the issue was less about getting more traffic than 
making the most of the traffic the site was already getting.  How 
do you get visitors to follow the path through your site that 
makes you, and your partners, the most money?  
In KBB's case, the most profitable path is their New Car Pricing 
section.  Carole Booth, Director of Web Operations, explains, 
"People are about to spend $20,000-$30,000 or even more and they 
want to get the proper research.  We found ways to monetize this 
path over time.  We keep it free for consumers, but what we 
ultimately hope they'll do is go off to one of our partners.  So, 
it's important to make sure visitors are tooling through, not 
stopping, going back, or abandoning the site."
CAMPAIGN
 KBB's home page is deceptively simple.  It looks clean, 
lean and obvious, but every element has been tested and tweaked 
relentlessly.  
For example, the Web team has tested as many as two-dozen 
different ways to word each of the various text links on that 
page.  
Henson says, "The link that sends consumers to the free price 
quote on new cars started out as 'Buy a New Car' because we 
thought that's the action that's going to be taken.  We found 
when we softened it a bit, it got a better click through.  People 
felt like clicking and seeing what's behind the curtain with 
wording that wasn't so immediate.  We tested things like 'Online 
Price Quote.'  Then we added the word 'Free' to it and it tested 
best of all the 20 different combinations."
This test, test, test culture meant that when a site conversion 
improvement firm NetConversions approached KBB last November with 
a few suggestions, they were eagerly received.  
Booth explains, "They could track some things we couldn't, such 
as if people click on white space, if they are scrolling or not 
scrolling, and at what points did they stall or abandon the 
site."  Also, instead of having KBB's busy in-house staff 
redesign the site to test things, NetConversions could create and 
host test pages that only one out of every 1,000 visitors would 
see.  That way KBB could compare live control to test cells 
easily.
During December 2001 and January 2002, KBB tested a variety of 
tactics and studied page usage patterns to improve traffic to and 
through its New Car Pricing section, including:
  - Adding a drop down box with main navigational links to the 
home page to supplement the navigation bar and text-description 
links already there.  
  - Noting where visitors clicked on each page and whether it was 
a clickable link or not. (Yes, surfers throughout the Web often 
try to click through on things are not hotlinked.)
  - Noting whether or not visitors scrolled down on critical 
pages, and if they scrolled, how far.
  - Testing verbiage, color and positioning of text on the 
navigation bar and in the text links.
RESULTS
 "The results that just made our jaws drop open," says  Booth, "were about scrolling.  When visitors get to their final 
new car pricing report, we would display a nice detailed report 
with everything you could want to know (i.e. destination charges, 
colors, options, what's standard, etc.), and nobody was scrolling to 
see the contents of this report!"
She continues,"25% of people didn’t scroll at all.  Of the people 
who did scroll, 90% only scrolled a few pixels and that was to 
see the base price number, when below was all this crucial 
information."
KBB's Web team redesigned this critical section completely 
from one long scrollable page to a single starter page which 
requires no scrolling, with a series of tabs for each additional 
section.  Booth says happily, "We launched it June 1st, and so 
far pageviews have increased 62%.  Those tabs are being clicked 
on."
KBB's business development team also shared these results with 
advertisers during their big quarterly presentations.  Henson 
says, "In general everybody's interested in how they can improve 
the click through rate and branding.  Whenever you've got hard 
research as opposed to 'this is what we think' it's always viewed 
in a positive light."  
Skyscrapers (very long, vertical banners) have been the sexy 
banner format of choice for advertisers on KBB and elsewhere for 
the past year.  KBB's scrolling test results helped many 
advertisers rethink their creative to make these more effective.
Booth explains, "One of the mistakes they were making was putting 
more important information at the bottom.  Generally there was a 
lot of white space, a picture of a car, and then the car brand 
name, and then people click here was in the lower end.  That 
important information was below the fold!  Advertisers didn't 
realize how much people do not scroll, and people really may not 
have seen a lower portion of your ad." 
More Results:
  - Click analysis revealed that about 50% of visitors were 
clicking on small images of cars to view a pop-up of a larger 
image.  Booth says, "It was a you hit yourself on the forehead 'I 
could have had a V8' kind of thing.  Why aren't we monetizing 
that?"  KBB added text links to ecommerce partners within the 
larger image pop-ups. 
  - Click analysis also revealed that visitors frequently clicked 
on KBB's logo throughout the site, despite the fact that it 
was not hotlinked to anything at the time.  Henson says, "It's 
becoming fairly common on the Web that when you click on the logo 
you go back to the home page."  Now their logo's hotlinked.
  - Adding the drop-down navigation device to the site was 
another definite success.  Booth notes that even though your home 
page already may have three redundant ways to navigate to the 
same places, it does not hurt to add another.  People prefer to 
find their ways through Web sites by different means and you need 
to please each visitor navigation-type.
  - Tests on textlink wording and changing the color of certain 
navigation items, showed minimal improvements, sometimes as 
little as .7%.  However, generally the changes were so easy to 
make that these little incremental improvements were worth 
working for.
http://www.kbb.com
http://www.netconversions.com