Ten Questions with Kenneth Norton
Introducing
Kenneth Norton, Director of
Marketing and Product
Strategy for Inktomi Web Search.
In the following interview, Kenneth
talks about Inktomi's new Web
Search 9 offering, index refresh
rates and tell's us why "we're
on the cusp of the third dynamic
shift in Internet search".
Thanks for
talking to us Ken. Can you tell
us a little about your background?
How did you end up at Inktomi?
I've
been working in the Web search
space for more
than six years. Although I've
crossed over into the "dark
side" of marketing, my background
is in engineering. I was the
founding architect of Snap.com
at CNET in 1996, and went on
to become the CTO of Snap and
ultimately NBC Internet, which
was formed after NBC invested
in the company. I left in 2000
to co-found a company called
Grand Central, one of the first
Web services players. I missed
the thrill of the fast-paced
consumer world, specifically
Web search, and decided to get
back into it.
I came to Inktomi
because of the company's reputation,
technology and experience. But
what impressed me most was the
people here and their relentless
commitment to Web search - it's
amazing to be surrounded by such
extraordinary people who are
pioneers in this industry.
You've recently
released Web Search 9. Here's
your chance to plug ;) What is
so great about this new offering
and why should we all be clamoring
to use it? What's in it for online
marketers and searchers?
Inktomi Web Search
9 represents a year's worth of
effort to deliver better relevance
and the freshest content on the
Web. It also combines algorithmic
and editorial relevance techniques
to better interpret user intent.
With
Web Search 9, we now lead the
industry in
relevance, freshness and size.
That said, we believe the next
order of magnitude improvement
in Web search technology is going
to come from better understanding
the user. By that I mean a deeper
understanding of what the user
actually means, not just what
they typed. The first of several
innovations to come in this area
from Inktomi is an industry-first
capability we introduced in Web
Search 9 called Smart Summaries.
With Smart Summaries, Inktomi
has moved away from the "one
size fits all" approach
to result abstracts, and is now
dynamically selecting and combining
contextual, editorial and custom
summaries based on the query,
pages and user intent.
For online marketers,
Inktomi Web Search 9 allows paid
inclusion subscribers to more
easily target regional audiences
with a unique Index Connect Geotargeting
option. Also, online marketers
can better customize how summaries
appear in the search results
- for example, retailers can
ensure that search summaries
include special promotions such
as free shipping. Lastly, a new
Web services version of our interface
allows online marketers to more
easily feed content directly
into the Inktomi index.
The press release
claims Inktomi will refresh the
entire index every 10-14 days.
If so, that's pretty impressive.
Is this happening now? If not,
when will we see this occurring?
Yes, we are refreshing
our entire index every 10-14
days now. In fact, a third-party
test conducted by Search Engine
Showdown proved that Inktomi
delivers fresher content than
any other Web search technology
provider. The combination of
Inktomi's fast content discovery
system that crawls the Web and
our 48-hour paid inclusion content
refresh, allows us to deliver
the most up-to-date Web pages
in the industry.
Your
press release states: "Web Search
9, delivers significantly improved
relevance and a more intuitive
Internet search experience to
users". How is this measured?
I think a lot of
people in the search space suppose
that relevance is too subjective
to be measured quantitatively,
but we vigorously disagree. We
take relevance measurement and
testing very seriously, and have
a large team dedicated to testing
relevance on Inktomi and other
search engines on an ongoing
basis. The majority of our relevance
team is comprised of PhDs, and
their backgrounds vary from statistics
to mathematics to linguistics.
The multiple tests
we conduct on relevance involve
both automated and human analyses.
By gathering feedback from the
human we can measure both relevance
and perceived relevance, which
determines the user experience.
The human tests involve blind
judges who determine whether
results are relevant for a random
selection of actual queries.
According to the
data we've gathered, it's pretty
much a two-horse race when it
comes to relevance - Inktomi
and Google. We're currently looking
for independent third parties
who might be able to regularly
test all of the search engines
and then publish the data, much
as performance benchmarks have
been published for years in other
technology markets. We encourage
our competitors to join us in
promoting this idea and being
publicly evaluated on a regular
basis.
Inktomi has
relied on portal partnerships,
rather than being a stand alone
search service in its own right.
Is this the approach that Inktomi
will be continuing with?
Inktomi is the
only search engine that has consistently
remained a pure OEM provider
of search technology. We think
the OEM approach has and will
continue to serve us well over
time. Our OEM Web search technology
can be customized to create a
unique search experience, helping
our portal partners drive and
retain traffic on their sites,
not steal it away. Additionally,
because we don't compete with
our partners, we are able to
work more closely with them to
deliver a better user-centric
search experience.
Inktomi has
recently sold off its enterprise
search, in order to refocus on
the Internet search offering.
Can you talk a little about this
move?
Web search is Inktomi's
core competency and the business
the company was founded upon
in 1996. We made a strategic
decision to focus the company
on the growing Internet search
market and build on our leadership
position in OEM Web search and
paid inclusion services. Upon
the completion of the enterprise
search transaction, Inktomi will
be in a much stronger financial
position and able to direct all
of our resources toward delivering
innovative Web search technology
to our portal partners, paid
inclusion subscribers and, ultimately,
the end-user.
It has been
reported that Google shakes their
heads at the need for paid inclusion/indexing,
yet this has been a key part
of Inktomi's strategy. Can you
talk a little about this?
The
rise of paid listings has really
changed the
economics of Web search and created
huge opportunities for marketers.
The paid listings category is
comprised of two very complementary
approaches - paid placement (what
Google does) and paid inclusion.
Paid placement is great for general
searches that advertisers can
predict in advance, such as "books" or "DVD
players." However, we estimate
that this approach only addresses
about 30 percent of searches
conducted everyday.
In order for marketers
to tap into the remaining 70
percent of Internet searches
for more specific items, such
as particular book titles or
product models, it is more efficient
and cost-effective to turn to
paid inclusion. With paid inclusion,
subscribers do not bid on keywords
for preferential placement within
a results set, but instead feed
their most current, deep Web
site content directly to the
search engine for consideration
in a search. Since paid inclusion
ranking is driven by relevance,
the marketer's pages returned
to users truly meet their needs
and translate into higher conversion
rates than paid placement. Not
surprisingly, someone searching
for a very specific product is
more likely to complete a transaction
than someone searching for a
general phrase.
Inktomi's first
priority will always be relevance
and we see the user as the most
important constituent in the
paid inclusion model. Paid inclusion
programs enhance the user experience
by making some of the most valuable,
deep Web content directly accessible
through an Internet portal search.
A large amount of the content
we get through paid inclusion
is not crawled by search engines
and locked away in databases
or on dynamic sites. Also, content
acquired through paid inclusion
is updated every 48-hours - a
key factor in our ability to
provide fresher content than
other search engines and a critical
element of relevance.
The Search
Engine Marketing world exists
in a strange place where the
relationship between the search
engine and the Search Engine
Marketer is not clearly defined.
How do feel about those who practice
Search Engine Marketing/Optimization?
Is there a way for both sides
to see eye to eye?
We work with a
lot of great companies in the
search engine marketing business
though our paid inclusion programs,
including firms like Position
Tech and Inceptor. The paid inclusion
model has opened the lines of
communication between content
providers (most of whom work
through search engine marketing
firms) and search engines, relieving
some of the tension that existed
previously. Ultimately, we are
all working toward the same goal
- ensuring that end-users find
the right content on the Internet
- and we need to work together
to provide the best experience
for the user. I think search
engine marketers who are willing
to work with, and not deceive,
search engines will find a great
deal of success going forward.
The search
engine landscape evolves constantly.
What is your take on where things
are at and where they are heading?
I think we're on
the cusp of the third dynamic
shift in Internet search. The
first, in the early nineties,
was focused primarily on discovery.
The challenge was finding all
of the new Web sites that were
published. At that time, much
of our research investment was
concentrated on crawlers and
indexing, and scalability was
the focus. Inktomi was started
with a government research grant
at the University of California
at Berkeley to address this very
problem.
By the late nineties,
however, the Web had grown exponentially,
and just discovering and crawling
sites wasn't sufficient. Now
that there were thousands if
not millions of relevant pages
for each query, the challenge
became ranking these pages to
ensure the best, the cream of
the crop, were returned first.
In the past, it might have been
enough to treat pages as discrete
buckets of words, using conventional
text retrieval methods, but that
was no longer sufficient. Much
of our research emphasis at that
time was on optimizing our ranking
algorithms, and search engines
began to see the Web as a rich
network of interconnected content
instead of a collection of independent
pages.
With all of this
focus on crawlers, clusters and
algorithms, search engines became
very introspective. They were
so focused on understanding the
Web that they neglected to consider
the other constituent in every
search - the user. Although great
advancements were made to the
interface, most user-focused
features (such as personalization,
collaborative filtering and highly
complicated search forms) came
across as arrogant, and often
mandated that users change their
behavior. But search engines
are the ones that need to adapt
their behavior, as it is unrealistic
to expect this of our users.
We sense that the
next few years in search are
going to be driven by a better
understanding of users, their
intent, their objectives, their
context and the tasks they are
hoping to accomplish. So while
crawlers and algorithms continue
to be essential, understanding
the user will involve a human
element to search. We're still
in the very early days of this,
but you can already see some
features in Web Search 9 that
are a step in this direction,
such as Smart Summaries.
Any other initiatives
are in the pipeline?
On the business
side, we're aggressively pursuing
new distribution partners as
we continue to change the economics
of search with our paid inclusion
programs. On the product side,
Inktomi Web Search 9 was the
culmination of a year's effort
to regain our leadership position
in algorithmic search - the
next initiatives underway are
focused on changing the game
through user intent and context.
</end>
Thanks a lot, Ken. We're looking
forward to seeing further developments
from Inktomi.
Next week, we'll (hopefully)
be talking to everyone's favourite
search commentator, Danny Sullivan.
|
