Vendor
Genesis
NetGravity
was founded in 1995 by Chairman John Danner as the first commercial provider
of online advertising solutions. NetGravity had $11.6 million in revenues in
1998, with an 81.2% increase over 1997's $6.2 million. Revenues for the six
months ending June 1999 were $10.5 million, a 136% increase over the same two
quarters in 1998. The company lost $11.3 million in 1998, compared with a $6.9
million loss in 1997. The Q2 1999 loss was $2.1 million ($0.12 per share), an
improvement over the Q1 loss of $3.7 million ($0.28 per share). Revenue growth
from 1997 to 1998 was heavier in Service and Maintenance ($5M, representing
growth of $2.8M or 220%) than in software licenses ($6.5M, representing growth
of $2.5M or 163%). Both cost of sales and R&D have decreased as a percentage
of revenues, the former from 97% in 1997 to 90% in 1998, and the latter from
47% to 40%. The company went public (NETG; NASDAQ) in June of 1998, and had
a second public offering in April 1999. Recent prices have been in the high
twenties to low thirties, with earnings per share of -$1.28. The number of NetGravity's
customers increased from approximately 250 at June 30, 1998 to approximately
380 at June 30, 1999.
The
company's initial and still core product (AdServer 3.5) is software that serves
ads to website pages. The AdServer product is licensed to a company and runs
on the licensing company's own servers. Ad serving to a web page is triggered
by special tags that the web page designer inserts into the page. The tags describe
characteristics of the ad to be served to that position. These characteristics
correspond to the model against which ads are sold. Characteristics can take
into account the page on which the ad appears, the time of day and other environmental
factors, demographic information such as the user's zip code or company, search
terms entered by a user, or information about the user that might have been
previously captured by the website. (See TEC Technology Research Note: "How
to Serve an Ad" October 23rd, 1999) Advertisers can view reports detailing
the numbers of impressions and clickthroughs. Ad reports are accessed over the
Internet.
The
entire suite of NetGravity products includes:
-
AdServer 3.5: this is the core product. It is licensed
in two varieties: The Enterprise model supports a single webserver; the
Network model can support many servers. The two varieties have slightly
different features and a site can run both simultaneously to take advantage
of all the features, although few sites need to do this. (See Last Page:
Two Models) AdServer runs on machines owned by the website publisher.
-
AdCenter for Agencies: A version of AdCenter for
Publishers designed for advertising agencies, which sometimes wish to host
their clients' ads themselves to get better control. The ad agency will
buy space on various web sites and provide each with a pointer to the agency's
AdCenter server. (The websites on which ads are purchased need not be using
NetGravity ad servers.)
-
Professional Services: This group assists customers
with installation and upgrades, with evaluating their successes using NetGravity
products, and with developing custom solutions to particular ad serving
problems.
On
July 13, 1999 NetGravity announced a planned merger with DoubleClick Inc. (NASDAQ:
DCLK). DoubleClick is an outsourcing service which differs from NetGravity's
AdCenter in that DoubleClick accumulates information about ads served to particular
users, regardless of which site the user had visited, and uses that information
to target ads to users based on their past history. Under the terms of the merger,
DoubleClick will issue 0.28 shares of DoubleClick common stock for each share
of NetGravity common stock. DoubleClick stock hovered around 100 during the
end of the summer and has been generally much more volatile that NetGravity's.
The merger is expected to close during Q4 1999.
Vendor
Strategy and Trajectory
Although
estimates of the amount that advertisers will spend on the Internet, and the
value that they get from advertising, fluctuate weekly there is no doubt that
advertising will continue on the Internet, and that it will continue to become
more complex. We believe that there are two areas in which companies involved
with ad serving must evolve technologically. First, the technologies that are
used to capture surfer attention (so called "rich media ads") will continue
to evolve. Whereas most ads today are static images or contain simple animation,
we expect to see banner positions becoming mini-entertainment centers, serving
up sound, full-scale animation, and video. NetGravity expects that its customers
will soon be able to offer their advertisers "transaction-enabled" banners,
which will make it possible for users to complete transactions through the banner.
Nothing that happens here should be a key differentiator between different products.
Whatever capability one company announces another company should - and must
- be able to announce soon after.
The
second and more critical area of technology evolution is to serve the growing
requirements from advertisers for micro-targeting of ads. Advertisers are coming
to expect ever more sophisticated abilities for targeting those surfers who
are most likely to convert from ad viewers to buyers. NetGravity sees targeting
as a major piece of their future, and is preparing for it in two ways. The next
version of the Ad Server (AdServer 4.0), announced for release in October 1999,
allows the system to collect information about target audience segments to be
used as the basis for daily forecasts into the future. Meanwhile, NetGravity
maintains flexibility by allowing customers to choose any provider of targeting
software and integrate it with the ad server. NetGravity has partnerships with
vendors like Net.Genesis and Net Perceptions, both of which have capabilities
for collecting, reporting on, and using information about the behavior of users.
There
are two big questions about NetGravity's future. One is the effect of the merger
into DoubleClick. We do not believe that DoubleClick will interfere with NetGravity's
core software licensing business. The merger was made, at least in part, to
give DoubleClick access to NetGravity's technology, so interference with the
AdServer product would be self-defeating. The AdCenter product is more vulnerable.
AdCenter could be seen as competing with DoubleClick; both serve customers who
want ad serving capabilities but do not want to host their own servers. However,
there are other ways in which the products differ. DoubleClick positions its
DoubleClick Dart service as a network. For the purposes of comparison with AdCenter
the salient difference is that DoubleClick merges information about user behavior
across all the sites it serves. Thus if website A wants to serve car ads only
to surfers who have previously shown interest by clicking through on car ads,
website A will have access to information about whether a surfer has clicked
through on car ads on any of the sites served by DoubleClick. This is a powerful
capability for a website to be able to offer to an advertiser, but there are
many website publishers who are unwilling to share their user data with other
sites. Although DoubleClick Dart does offer its customers the opportunity to
opt out of data sharing, the physical ownership of user behavior data resides
with DoubleClick, and this had been unappealing to some publishers.
Logically,
therefore, the merger should not affect the AdCenter product. However, the business
story has some other considerations. Revenues from AdCenter were $1.1 million
for the six months ending in June of 1999. The cost of providing transactional
services (including GPS) was $2.2 million for the same period. This is hardly
out of line for a new product line. However, given that additional expenses
are expected for equipment and personnel, and that NetGravity will have to duplicate
DoubleClick's expertise in managing outsourced services, we think it is quite
likely (probability 85%) that AdCenter will be merged into DoubleClick Dart.
We suspect that DoubleClick is prepared to make this move fairly swiftly, but
think that reports from the field about customer concerns with both data privacy
and performance may delay any announcement until DoubleClick has introduced
NetGravity's technology into its Dart service.
Vendor
Strengths
NetGravity
is above average for both its features (including the ability to target ads,
the administrative capabilities provided to the website managers, reporting
and support for adding custom features) and, especially, its scalability. The
software is fast and generally reliable: One of their major customers serves
one billion ad impressions each month. It is easy and relatively inexpensive
to scale up as the number of ad impressions grows, so a NetGravity customer
can handle almost any imaginable volume of ad impressions. The software runs
on both Unix and NT servers, and with the Netscape and Apache webservers. Version
3.5 does not work with the Microsoft Internet Information Server, a weakness
that is expected to be repaired in AdServer 4.0. Users provide their own databases,
which can be Oracle, SQL Server, or any ODBC compliant database. Operationally
the software is relatively easy to use, although the basic Ad Master training
is strongly recommended for new users. Bringing up a new ad is relatively easy,
as is generating an ad hoc report. However, the standard reports are not presented
in real time, and a site with many hundreds of advertisers (a rarity) can expect
report generation to take up to a day The reporting package offers a number
of different options and a customer can develop custom reports if needed. The
courses presented in their education centers are useful, and on-site consultants
are knowledgeable and helpful. All consulting engagements are documented carefully,
and the company builds the cost of post-installation reviews into the license
cost, ensuring customers that they will be set off on a strong footing. Post-installation
customer service gets generally acceptable marks.
Vendor
Challenges
Out-of-the
box installation is painless, especially since a NetGravity consultant will
be on-site to help. However, straying from the simplest case is not as easy
as it should be. Customizations tend to be difficult, in part because detailed
technical information is not always easily available to the customer or to the
support staff. Also, uses that stray from the basic model, that of a site serving
consumer ads, have caused difficult integration problems for at least one business-to-business
customer.
The
control software used by customers is powerful, but does not support the situation
where different units within the customer want to handle managing their own
ads; this is a promised feature in the next version, however. While reporting
facilities are good, generating reports can be slow, and collecting additional
data to report on targeting information that might be collected by the customer
can strain the database and reports. The product does not offer real-time reporting,
although ad hoc reports are fairly easy to generate. As the competition for
ad dollars increases and advertisers look to maximize the effectiveness of their
online campaigns, the lack of real-time reporting may become a significant issue.
It will not be easy to add real-time reporting to the architecture used in AdServer
3.5, and the situation is as yet unknown for the 4.0 release.
NetGravity
allows websites to take advantage of information they collect about individual
customers, but does not integrate this information into its own database. This
makes the process harder than it should be: The customer has to maintain its
own database of information and, before a page is served to a known individual,
derive from that database a code that will direct NetGravity to serve a particular
kind of ad. There will be an explosion of precision targeting of this kind in
the next few years, and NetGravity will be pressured to respond by making such
targeting part of its basic offering. This will be a challenge to the current
architecture.
The
merger with DoubleClick should not prove a negative for NetGravity's customers.
But NetGravity must be concerned with the response by its main rival, CMGI.
(See TEC Technology Research Note: "'Ads
Are Us', Boasts CMGI" October 7th, 1999). NetGravity's major weakness
against CMGI's Engage subsidiary has been Engage's targeting capabilities. Engage's
actual ad server, Accipiter, has not been a serious threat to NetGravity, but
the integrated tracking and targeting capabilities of Engage, which were added
to Accipiter when the latter was purchased by Engage, are very attractive. NetGravity
does not have equivalent features, although it has partnerships with Net.Genesis
and Net Perceptions, which fill in the gaps. Still, the capability would look
a lot stronger if it were integral to AdServer.
We
are concerned about the lack of attention that NetGravity has paid thus far
to security issues. An advertiser's ad traffic reports can be important competive
information, and in AdServer 3.5 these data, once an advertiser supplies a password,
are sent of the Internet with no advanced for of security protection. Security
is relatively easy to implement, and should be high on the feature list.
Vendor
Predictions
NetGravity
is well positioned to survive any shakeouts in this area due to its new relationship
with DoubleClick, and will be one of the top two choices for licensed ad serving
software for the foreseeable future. With the backing of DoubleClick and the
aggressive moves by CMGI, enhancing the capabilities for intelligence gathering,
and targeting are imperative. The probability is near 100% that NetGravity will
make a significant announcement along these lines by early 2000.
As
noted above, we expect (probability 85%) that DoubleClick will eventually phase
out the AdCenter product line, hoping to convert most of its customers to DoubleClick's
own service or to NetGravity's AdServer product.
Vendor
Recommendations
A
logical step first step would be acquisition of enhanced user tracking and targeting
capabilities, perhaps from one of its current partners. However, NetGravity
should also be looking at some of the leading edge work university work on extracting
information from unstructured data, which has the potential of providing its
customers with data mining capabilities to capture unexpected relationships
between surfers, the pages they visit, and the ads that interest them. Any improvements
in analysis and targeting would benefit DoubleClick as well as NetGravity.
Although
the software's usability is good given the complexity of the task, NetGravity
must show continuous improvements to avoid becoming known as a "high-end" company
that can be too complex for the "little guy." On the other hand, those high-end
customers will increasingly be pushing the edge of the technical envelope. NetGravity
will have to improve the tools it provides to its customers to modify the out-of-the-box
functionality if those customers are to be able to react swiftly to demands
from their advertisers (or their marketing departments).
NetGravity's
support staff and consultants are successful because of their intimate knowledge
of the product: The difficult problems frequently require understanding of what
the customer wants to do, how others have done similar things, and what technical
capabilities of the product are being stretched. It is important for the company
to preserve its strengths in this area. It would be a natural bottom-line oriented
result of the merger for the support functions to be considered as commodities,
and combined into a single organization serving both products. We believe that
it is in the best interest of NetGravity 's customers that any such action be
delayed, or at least driven according to NetGravity's technical and service
standards.
User
Recommendations
The
first two questions a prospective user of ad serving software must ask are:
- Whether to license
the software or to use a transactional model
- Whether to be
totally responsible for its own ad sales and data collection or to be part
of an ad network
If
a website wants to be part of an ad network, the details -- such as whether
they want to do some of their own ad sales, all of their sales (but use a network
like CMGI's newly acquired Flycast subsidiary to handle unsold inventory) or
none of their sales - will be a strong driver in a buying decision. The decision
to license the software and host servers or to pay by the ad is also one that
conditions the decision. However, there are very few combinations of these factors
that rule out a NetGravity solution.
Larger
websites, any site choosing to license and host the software itself, and those
who have requirements that no vendor addresses directly, will definitely have
NetGravity on their short list. Unless NetGravity abruptly discontinues sales
of version 3.5 when it announces 4.0, users should consider both products, trading
the proven reliability of the older product against the new features of 4.0,
at least until 4.0 has established its own reputation. Small sites with simple
requirements that want to license software will want to consider NetGravity
as well, but are safer making their decision based on cost factors alone.
Note
that anyone interested in serving their own ads must consider the total cost
of ownership, which must include assuring reliability and redundancy for both
servers and storage systems. Once a website begins serving ads any loss of uptime
has a direct impact on the bottom line - stronger even than the effect of losing
revenue from consumer sales because advertisers will expect refunds if their
ads are not served as promised. This is not an issue for or against NetGravitry
as opposed to any other vendor, but might well tip the decision to an outsourced
solution.
Sites
that want to use the transactional model and outsource ad serving will weigh
the advantages of NetGravity's AdCenter against that of other services where
aggregate user data from different sites is pooled in one way or another. If
the NetGravity model is the one that fits best, the considerations are cumulative
cost of transactions and the performance and uptime of the servers. The quality
of NetGravity's technology is likely to make AdCenter an attractive choice for
both performance and uptime, but the service is not sufficiently mature for
a good prediction. Calling current users for referencs and testing response
on sites that use AdCenter will be necessary components of the decision.
Two
Models
There
are two different models under which a user can license NetGravity's AdServer
3.5 software. In the Enterprise model, the ad server resides on the same machine
as the one that serves the web pages. When a surfer's browser loads a page,
the HTML code is first sent to the ad server, which replaces ad tags with images
and clickthrough addresses. The resulting HTML code is then passed to the web
server and sent to the browser.
In
the Network model the AdServer 3.5 software resides on a different machine.
(Actually, in this model you may have separate machines for serving ads, running
the user console, serving advertiser reports, and running your AdServer database.
All of these functions can reside on one machine, but performance considerations
may make the multi-machine option more attractive). When a surfer's browser
loads a page it sees an image reference to a URL on the AdServer machine. This
reference contains parameters that specify the characteristics of the ad. The
browser sends a request to the AdServer machine for the image; the AdServer
chooses an ad that meets the parameters and returns it as an image.
In
the Network model the AdServer 3.5 software resides on a different machine.
(Actually, in this model you may have separate machines for serving ads, running
the user console, serving advertiser reports, and running your AdServer database.
All of these functions can reside on one machine, but performance considerations
may make the multi-machine option more attractive). When a surfer's browser
loads a page it sees an image reference to a URL on the AdServer machine. This
reference contains parameters that specify the characteristics of the ad. The
browser sends a request to the AdServer machine for the image; the AdServer
chooses an ad that meets the parameters and returns it as an image.
The
Enterprise model is generally used by organizations that have only one website.
Since ad serving is an extra load on the existing web server these tend to be
sites with a small number of ad impressions, although a site that serves content
from many servers can mount AdServer software on each. The Network model is
used by larger organizations and by those that want to serve ads for many different
websites from one server.
One
difficulty with the Network model is that when there are two ads on one page,
each is served independently of the other. This means, in particular, that if
there are two positions on a page that can contain ads from the same pool of
eligible ads, it may happen that the same ads shows up in both positions. Since
the Enterprise AdServer sees the entire page at one time, it can prevent this
from happening. Should this be an issue a single website can use both models:
in this some ad tags will be interpreted by the Enterprise AdServer while others
would be sent to the browser, which would call on the Network AdServer to resolve
them.