|
The human mind loves to categorize things: kingdom, phylum, class, order, family, genus,
species. It makes us feel better. It helps us understand the world around us, so we can respond
to things appropriately.
Is it larger than you? Able to move quickly? Has large, sharp teeth? Seems to be interested in
you as a lunchtime snack? It doesn't matter if you know what it is or not, you have enough
information to respond in an appropriate manner: Run!
So the first thing we do when it comes to Web site visitors is segment them. First by way of
immediate observation, and then by way of long-term memory or database.
By Technographic
This is the information you get for free from your server logs. Do people who use Macintosh
computers buy more when offered free shipping than those who use Netscape on a Windows
machine? Are you more likely to up-sell a visitor using a newer version of the browser? How
important is it that they are Java-enabled?
By Interest
What are they looking at?
All the qualification process metrics apply—duration, depth,
recency, and frequency—but mostly we're interested in their interests at this point. You can
segment site visitors by the level of attention they pay to specific sets of pages on your site.
Cliff Allen is president of Coravue, publisher of a line of Web personalization and content
management system software. He's also coauthor of One-to-One Web Marketing: Build a
Relationship Marketing Strategy One Customer at a Time, Second Edition (John Wiley &
Sons, 2001).
I ran into Allen at a DCI CRM conference in Los Angeles, and he described a couple of
illustrative Coravue reports for me. The example results are real, even if the Web site
(SureToFind.com) is mostly a test/development site for the tool.
"The Interest Profile pulls together data from the user profile database, the
content database, and the Web access database," Allen explained, "and shows which type of
users are visiting various pages. The report shows that several pages are visited by clusters of
users, indicating that people with those interests visit those pages more than they visit other
pages."
The ID is the content identification, and the Total is the number of people who looked at that
content, so right off the bat, you can tell which content is the most popular. Then, by
categorizing the content itself, you can see that people who are interested in Manufacturing
(#4) don't care about Becoming a Consultant, Books for Consultants, or anything else that has
to do with consultants. They have a great deal in common with those interested in distribution.
The Customer Status Profile report relates the same content pages to
Customer Status; indicating whether the visitor is registered on the site, a newsletter
subscriber, a qualified lead, a prospect, or an actual customer.
The Registered and the Newsletter Subscriber columns are automatically set, while the
Qualified Lead, Prospect, and Customer columns are set from the sales automation
application by the representative assigned to handle the contact. Somebody who looks at
About Us is very unlikely to become a customer and those looking at Office Efficiency Tools,
Tips & Techniques are the most likely.
In an article he wrote back in 1999 (www.allen.com/1999-07-20.html), Allen describes how
visitor segmentation leads to collaborative filtering:
So what can we actually learn by segmenting our audience?
Perhaps the easiest thing to learn is which products are purchased by the same people. Web
sites that use collaborative filtering are helping marketers and consumers answer this
question. You've probably seen sites that say, "People who bought this product also
purchased these other products."
For instance, if a group of customers buys what appear to be unrelated products, try crosspromoting
the two products and see if other customers buy that combination, too. By using
market segmentation tools and techniques, unique groups of people can be identified and
marketing programs created to take advantage of this opportunity.
By Demographic
Even without expensive collaborative filtering tools, you can learn a lot about your customers
just by grouping them into logical clusters, writes Allen:
For example, how does the geographic distribution of your customers compare to the country
as a whole? California has 12 percent of the U.S. population, so if less than 12 percent of
your U.S. customers are from California, that segment might not respond to the same
marketing messages as other regions. By targeting a different email message to that market
segment, you might find results are higher than sending the same email newsletter to
everyone.
You might also find results are higher by dynamically serving specialized content to them. At
the very least, you can keep track of who's coming to your site in order to have the right type
of content waiting for the majority of your visitors.
Financial institutions mostly segment their audiences based on income and investment
approach (risk takers or avoiders), but they also look at something as simple as age.
AGE INVESTMENT INTERESTS
0–17 Savings account
18–24 Checking account with ATM
25–34 Car insurance
Car loan
Credit cards
Life insurance
College savings plans
35–44 Home mortgage
Home insurance
Stocks
Bonds
Mutual funds
45–54 Home equity loan
Estate planning
Retirement accounts
55–64 Certificates of deposit
65 + Trust funds
At Oracle, the product managers and program managers review real-time dashboards where
they can view registration demographics. Knowing that 20 percent of your audience are IT
managers and 40 percent are software developers helps put the right message where the right
person is likely to find it. Then, an interest profile report will tell you if you've made the right
changes to your content.
By Customer Knowledge
Once you start using cookies, you can keep track of people on a repeat visit basis. That means
you can grow your profile of the individual and cater to him or her better.
Terry Lund at Kodak describes their approach like this: "We put you in a bucket of high-end
amateurs based on prior knowledge and next time you come you're going to get something
we've tailored to high-end amateurs. We won't be making decisions about Jim Sterne as an
individual. That's a privacy issue. How far can we actually go when we know it's Jim Sterne
who just arrived?
"Most sites seem to be able to get away with 'Hi Jim, welcome back,' and we're not getting too
much push back on that. We could say 'Gee, you haven't been here in 3 weeks and here are
some new sections. And the last time you were here, you stopped your visit looking at this
section.'"
This sort of personalization works if you adopt a strict opt-in approach. You tell visitors all of
the great stuff that you can do, and invite them to sign up for level A, B, or C. They're pleased
as punch because they show up at a Web site that knows who they are and can cater to their
needs.
By Industry and Job Function
Got a headache? Go to www.excedrin.com. Want to color your hair? Go to www.clairol.com.
Those are the brands you know and love. It probably didn't occur to you to head over to
www.bms.com.
That's where you'll find Bristol-Myers Squibb, a $20 billion company doing business
worldwide. But they have nowhere near the brand-name connection between need and
solution that their products have. Why is that? Branding. What to do? Create multiple Web
sites, of course.
I recently spoke to a company that writes software for the Department of Defense. Their client
list includes customers like the Ballistic Missile Defense Organization. On the other hand,
they sell agricultural accounting and information management software. Clearly, the guys
building missiles and the guys driving tractors not only have nothing in common, they don't
feel comfortable buying from somebody who consorts with the other. Solution? Two Web
sites.
But then they had a little problem. They have a project-estimating tool that's great for
anybody and everybody. This is a product Uncle Sam can make use of just as well as Farmer
John. What to do? Create a third Web site? Maybe, maybe not.
Sometimes different markets call for a different look and feel for the same product. That
software package, dressed up in olive drab, will appeal to the DoD a lot faster than if it's
clothed in a plain blue suit. The same goes for draping it in denim coveralls. Tractor drivers
will cotton to it faster in that style than they will if it's wearing wing tips.
Create multiple Web sites for each vertical? When the verticals are that far apart, yes. But
we're not done yet. In time, the right choice will be to subdivide the brand by job
responsibility within a given company, within a given market segment.
Different people within each company have different needs and respond to different offers,
different language, and different branding. For those of you who missed the now out of print
"Strategic Selling: The Unique Sales System Proven Successful by America's Best
Companies," it may be worth looking for it in the used bin (Robert Miller and Stephen
Heiman, Warner Books, 1986). Maybe "The New Conceptual Selling" by one of the same
authors (Stephen E. Heiman, Warner Books, 1999) will fill the bill.
The original book divided prospective customers into four areas of interest: the product user,
the economic buyer, the advisor, and the manager. Each of these individuals is motivated by
different things in order to achieve different goals. Does your Web site recognize these
divergent needs and offer up a different brand to each?
Is your product easy to use? Will it save money? Is it used successfully by others? Will it
make a company more productive? The answer should be yes to all of these questions and
presented to different Web visitors in these different lights.
So now you have a Web site for a given vertical and a brand for buyer types within each
vertical. While we're at it, let's add another layer: the size of the prospective customer's
business. That's how many verticals multiplied by four types of buyers across four or five
business size classifications. How much is all this going to cost? Wait! Don't answer yet.
Because we're not done slicing and dicing yet.
How is your brand presented to John G. Smith who lives at 123 Main Street, Anytown, USA?
Mr. Smith is an interesting mixture. He works for a manufacturing company in the textiles
industry. His responsibilities revolve around facilities management and shop floor control.
He's more interested in quality than cost savings. He doesn't eat meat. He has a flexible
budget, but no ability to hire more people.
Does your site brand your product to cater to Mr. Smith's hopes and fears and sway him with
persuasion that is geared to his psyche? It will.
The same sort of tools being used to match up banner ads with Web surfers based on their
clicks and their search terms—used to put the right message in front of the right person at the
right time—can be used on your site to match up the right benefit statement to the prospect
and deliver the right brand at the right time.
And then it gets interesting.
By Personality
If you can get your site visitors to reveal how much they like variety, theories, outrageous
people and things, and whether they like to make things and follow the latest trends and
fashions—a total of forty questions—you can pigeonhole them into VALS categories.
VALS was created in 1978 by SRI International and is currently run by SRI Consulting
Business Intelligence (SRIC-BI; http://www.sric-bi.com/VALS). It's a consumersegmentation
system that goes beyond demographics and geographics, and focuses on the
psychographic factors (a combination of psychology and demographics) that motivate
consumer buying behavior. The categories include such designations as Fulfilleds, Believers,
and Strivers .
A VALS result of Actualizer/Achiever is interesting, but only if you know how to make the
most use of that information. That's why VALS works closely with its clients and provides
them with in-depth descriptions and product and media purchase data for each of the VALS
types. Common uses of the VALS system include product creation, target identification, and
product positioning and advertising.
One of the marketing segments Web sites have been toying with is matching the customer
experience on the site with the mood/mode the customer is in at the moment.
By Surfing Mode
Why did those visitors come to your site and how do they feel about it? They might be there
to do the following:
• Browse
• Hunt
• Research
• Compare
• Locate
• Acquire
• Complain
In April 2001, Booz Allen Hamilton and Nielsen//NetRatings put out a press release
identifying seven types of Web users:
Study of Online Consumer Segmentation Uncovers "Occasionalization" as Next Step to
Reviving Marketing and Retailing on the Web
NEW YORK, April 2, 2001—To increase the effectiveness of their online marketing,
advertisers and retailers must study how users actually use the Web, according to Booz Allen
Hamilton, the elite management and technology consulting firm, and NetRatings, Inc.
(Nasdaq: NTRT), a provider of the Nielsen//NetRatings Internet audience measurement
service. Web usage patterns fall into seven categories of online behavior, according to Booz
Allen and Nielsen//NetRatings, and while in some categories consumers are more likely to
buy, in others they are nearly immune to traditional online marketing pleas.
These findings are part of Booz Allen and NetRatings' co-branded study, Seize the Occasion—
Usage-based Segmentation for Internet Marketers, which was released today. This marks the
second in a series of studies from Booz Allen and NetRatings' Digital Customer Project,
which examines facts about online behavior to improve the ways that business interacts with
its customers via the Internet and other new technologies.
According to the study, focusing on how people actually use the Internet—exploiting Internet
technology's ability to track behavior—is superior to marketers' current reliance on "bestguess"
demographic and other user-based segmentations. The findings are based on analysis
of proprietary click-stream data collected from nearly 2,500 users between July and
December 2000.
Market Segmentation Based on Behavior
Focusing on the wide behavioral variations exhibited by online consumers and the
opportunities that these variations can provide for marketers, the study introduces
"occasionalization," a new form of Internet market segmentation that identifies consumer
segments based on online usage occasions rather than on user-based characteristics, such as
demographics or attitudinal data.
By exploring users' session characteristics—how long a user stayed online, how much time
the user spent on each page, site familiarity and the category concentration of sites visited—
the study uncovered seven types of sessions, and found that three—Information, Please;
Loitering; and Surfing—are more likely to involve shopping than others. These sessions are
the lengthiest, ranging from 33 to 70 minutes, and page views are 1 to 2 minutes, so users are
likely to linger on a page and be exposed to different messages.
[The seven categories are as follows:]
• Quickies: Typically short (1-minute) sessions that center around visits to two
or fewer familiar sites. Users spend about 15 seconds per page extracting
specific bits of information or sending email. Users in Quickie sessions may
not notice any type of message as they scoop up the needed information and
log off.
• Just the Facts: Users seeking specific pieces of information from known sites.
At 9 minutes, these sessions are longer than Quickies but share the aspect of
rapid page views. These occasions are less likely to involve sites best enjoyed
at leisure, such as entertainment. Users in Just the Facts sessions have a low
propensity to buy.
• Single Mission: Users want to complete a certain task or gather specific
information, then leave the Internet. During these visits, generally lasting 10
minutes, users venture into unfamiliar sites to find what they need, while
concentrating on sites within a single category. Users in Single Mission
sessions are only open to messages related to the purpose of the session, but a
well-targeted banner ad may provide a good return.
• Do It Again: These sessions are 14 minutes in length and are notable for
lingering page views: 2 minutes, tied with Loitering for the longest of the seven
types of sessions. Ninety-five percent of the time is spent at sites the user has
visited at least four times in the past. Users in Do It Again sessions may be
willing to click through banner ads that are strategically placed on their
favorite sites or react to site sponsorships that bring real content directly to
the consumer.
• Loitering: At 33 minutes in length, with 2-minute page views, Loitering
sessions are similar to Do It Agains: leisurely visits to familiar "sticky" sites,
such as news, games, telecommunications/ISP, and entertainment sites. A
company undertaking a brand positioning campaign would focus on Loitering
sessions, where the consumer spends more time on each page and is more
likely to absorb the marketer's message and develop the necessary brand
associations.
• Information, Please: These sessions average 37 minutes in length and are
used to build in-depth knowledge of a topic, perhaps for a research report.
They differ from Single Missions because users gather broad information from
a range of sites. Users in Information, Please sessions are mostly going to
familiar sites, but are willing to cross categories and linger on a page that
piques their interest, giving marketers an opportunity to expose them to
different messages.
• Surfing: Surfing sessions are the longest, averaging 70 minutes, with few stops
at familiar sites, as users hit nearly 45 sites in a typical session. Time per page
is a minute or more, suggesting wide, but not deep, explorations. Surfers
usually spend time on sites with lots of content, giving marketers opportunities
to build branding awareness, since during these occasions users will be
exposed to messages for a relatively long time. Sponsorships of content are
another good approach, encouraging users to associate their favorite content
with a specific brand name.
Lessons for e-Tailers
The Booz Allen and Nielsen//NetRatings study indicates that a successful e-tailing site is
likely to evolve from a "one size fits all" approach into a series of parallel sites targeted to
appeal to multiple usage occasions. The challenge for e-tailers 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. For example, users engaged in Quickie or Single
Mission sessions can get a rapid, no-frills self-serve experience marked by text-only pages
and no pop-up ads, while users in Loitering and Information, Please sessions can be steered
toward the full-service option, with video pop-ups and personal shoppers.
As you might expect, there are many ways to categorize Web visitors. Another study, this one
performed by Dr. William R. Swinyard and fellow Brigham Young University professor Scott
M. Smith at the BYU Marriott School of Management in July 2001
(www.byu.edu/news/releases/archive01/Jul/internetshoppingstudy.htm), looked at 4,000 Web
users to determine which types of surfers would respond to marketing efforts.
The BYU study found the following eight categories:
• With 11.1 percent of the market share, Shopping Lovers enjoy buying online and do so
frequently. They are competent computer users and will likely continue their shopping
habits. They also spread the word to others about joys of online shopping whenever
they have the opportunity. They represent an ideal target for retailers.
• Adventurous Explorers (8.9 percent) are a small segment that presents a large
opportunity. They require little special attention by Internet vendors because they
believe online shopping is fun. They are likely the opinion leaders for all things online.
Retailers should nurture and cultivate them to be online community builders and
shopping advocates.
• Suspicious Learners (9.6 percent) comprise another small segment with growth
potential. Their reluctance to purchase online more often hinges on their lack of
computer training, but they are open to new ways of doing things. In contrast to more
fearful segments, they don't have a problem giving a computer their credit card
number. Further guidance and training would help coax them into online buying.
• Among the most computer literate, Business Users (12.4 percent) use the Internet
primarily for business purposes. They take a serious interest in what it can do for their
professional life. They don't view online shopping as novel and aren't usually
champions of the practice.
• Fearful Browsers (10.7 percent) are on the cusp of buying online. They are capable
Internet and computer users, spending a good deal of time "window shopping." They
could become a significant buying group if their fears about credit card security,
shipping charges, and buying products sight unseen were overcome.
• Shopping Avoiders (15.6 percent) have an appealing income level, but their values
make them a poor target for online retailers. They don't like to wait for products to be
shipped to them, and they like seeing merchandise in person before buying. They have
online shopping issues that retailers will not easily be able to overcome.
• Technology Muddlers (19.6 percent) face large computer literacy hurdles. They spend
less time than any other segment online and show little excitement about increasing
their online comfort level. They are not an attractive market for online retailers.
• Fun Seekers (12.1 percent) are the least wealthy and least educated market segment.
They see entertainment value in the Internet, but buying things online frightens them.
Although security and privacy issues might be overcome, the spending power of the
segment suggests that only a marginal long-term payback would be possible.
Then, of course, there are those people who simply want to buy something. Robin Edwards of
the UK Web consultancy and design shop Clockworx (www.clockworx.com) likes to segment
people who have and have not purchased.
We are counting unique session ID cookies, as counting IP addresses (like so many stats
packages do) is kind of pointless in these days of proxy servers and corporate/education
access. This gives us a daily total of the number of people who were on the site and thus the
overall conversion rate. The next step is to break the conversion rate down further by looking
at historical data for each person so we can work out the following:
• Conversion rate for first time visitors becoming customers
• Conversion rate for repeat visitors becoming customers
• Conversion rate for repeat customers placing an order rather than just
Browsing
Each stat will tell us something different about the way the site works and how it can be
improved further. For example, if we don't get a high conversion rate for the latter type of
visitor, then we need to seriously look at our CRM systems.
If your company is in the entertainment business or your brand depends on being entertaining,
then catering to those who show up just to play games, hear music, or watch videos might be
important. These people might be more willing to register and return, giving you the
opportunity to do a lot of cross-referencing between visits, game plays, duration, and
purchases.
Understand that individuals can and do operate in different moods/modes at different points in
time, so segmentation only goes so far. At some point, you have to get down to the individual,
and that calls for Web site customization. | 
Disclaimer
