|
Because of the digital nature of the marketplace,
behavioral analytics is a natural advertising model for ad, social and wireless networks, search engine portals, demographic
providers, and multiple types of websites – they are however also very important to today’s enterprises across
all its available consumer channels. All of them are very much dependent on the strategic use of web data
stream. In this section we will discuss how a
behavioral analytics strategy and a tracking framework can be constructed by enterprises and marketers for relevant and real-time
business intelligence using web data streams. These data streams are created by consumers in their search
for relevant information on their desires, needs, appetites and preferences, all of which starts with the humble little cookie.
Cookies are simple
text files first cooked up by Netscape over a decade ago they are sent to a web browser from a web server where web pages
are being queried.
The same process of real-time
segmentation by age, gender, lifestyle, shopping preferences and other assortments of demographics or other behavioral data
as performed by web, wireless and social networks in their positioning of products, services and content can be leveraged
by agile enterprises either thru a subscription model, or by constructing it in-house – or through a combination of
both. Most importantly all of these services, techniques, and networks can today be replicated at the enterprise
level for up and cross selling, customer lifetime valuation and sustained growth profitability by all types of companies large
and small. The browser is the consumer – as such an enterprise needs to have a strategy for using
their website as the core to their marketing efforts via behavioral analytics of their web data streams – cookies, JavaScript
and Deep Packet Inspection (DPI) are the keys.
Cookies are an important component of web data streams and behavioral analytics insofar as it allows marketers, such
as ad networks and ISP providers to track and model consumer behaviors. Cookies also allow enterprises
to monitor and model customers behaviors at their website, which coupled with other legacy systems can be used to provide
relevant content to consumer while ensuring their privacy. Cookies have multiple purposes; however the
first application was to maintain a browser’s session state, such as a shopping cart and was a necessity to allow electronic
commerce to evolve. Cookies also allow automatic authentication of a web browser entity to a web server
setting the cookie. The fact that cookies can identify a user by their web browser unique identification number is paramount
to behavioral analytics because they can be used to track what web pages have been viewed.
A web site can generate a unique identification
number for each visitor and store their unique identification number on each user's machine using a cookie file –
the pieces of information are stored as name-value pairs.
For example, a cookie file for jesusmena.com could just contain the following information stored on
a user’s machine as a single name-value pair. This could simply be the name of the pair such as their UserID, and a unique value such as A9A3BECE0563982D:
UserID A9A3BECE0563982D
www.jesusmena.com/The first time a user visits jesusmena.com the site could assign them a unique identification number
value and stored it on their machine. The values cookies can store is totally flexible not only can a unique
number be stored and use to recognize a user’s as a unique visitor but even record the number of unique sessions, for
example Amazon.com stores a bit more information on users. This
is what an Amazon cookie stores and how it configures itself on users’ machines: session-id-time
954242000 amazon.com/
session-id 002-4135256-7625846 amazon.com/
x-main eKQIfwnxuF7qtmX52x6VWAXh@Ih6Uo5H amazon.com/
ubid-main 077-9263437-9645324 amazon.com/Amazon stores a main user identification number,
a unique identification number for each session, and the time the session started on a user’s machine, as well as an
x-main value, which could be the type of book a user previously purchased – this is of value since
targeted recommendations can be generated every time that user logs on to Amazon. So cookies can be used to make relevant recommendations to users,
which are mutually beneficial to both web sites and users. They make possible the placement of targeted content, products
and services. They can be installed on browsers by either marketers or enterprises – they have also evolved to new levels
of sophistication in that they are being placed on consumer browsers by ad networks as well as ISPs. This is how cookies get
installed and used: 1. The process starts when a user selects the URL of a web site into his or her browser, for example
this can be when the user clicks on a site from say the search engine results or a bookmark on their browser. 2.
The browser will check the computer’s hard drive for a cookie associated with the web site the user just clicked on.
If it finds the appropriate cookie, the browser will send the cookie information along with the URL to the web server which
set the original cookie. If the browser doesn’t find a cookie, no data is sent. 3. The web server receives the request for a page.
It then checks to see if a cookie was sent as well. If so, the web server can use that information to
tailor the web page specifically for that user, such as book recommendation by Amazon. 4. If the web server didn’t receive a cookie, it knows that user has never visited
the site before. The web server then creates a new ID for that user in the web server’s database and sends a cookie
in the header for the web page to that machine. The machine then stores that cookie on that user’s hard drive. From
that point on that user will be uniquely identified when they land on that particular web server. Sites
can use cookies to store user preferences
so that the site can look different for each visitor, for example when msn.com prompts a user for their zip
code it can be used it to customize their weather reports: WEAT CC=TX%5FEl_Paso%2DSouthwest®ION= www.msn.com/ Marketers and enterprises use databases to
store the products and service – as well as the content users have selected from their site, pages they have viewed
from that site, information users have provided to the site via online forms, etc. All of the information
is stored in that site's database, and in most cases, a cookie containing those users’ unique identification numbers
that is stored on those users’ machines. As previously mentioned ad networks can also create cookies that
are visible on multiple sites. DoubleClick is the most famous example of this, many sites use DoubleClick to serve banner
ads on their sites. This is how most ad networks like DoubleClick work, they place small (1x1 pixels) GIF files, also known
as beacons on a site that allow DoubleClick to load cookies on users machines. DoubleClick can then track
their movements across their multiple client sites. Some networks can see the search strings that users’ type on search
engines, thus providing them information about user preferences, needs, products, services and content they are looking for. Because ad networks can
gather extensive information about users’ behaviors from multiple sites, DoubleClick for example can form very rich
profiles. These are still anonymous, but they
are still valuable buckets of consumer behaviors. DoubleClick and companies like it are in a unique position
to do this sort of thing, because they serve ads on so many sites enabling them to perform cross-site profiling which is not a capability available to individual sites,
because cookies are site specific.
Ad networks and behavioral analytics service providers have the ability
to collect consumer data from their client’s ad servers and email databases, and then merge that information with their
client’s subscription, contest and registration form databases. These profiles contain both demographic and behaviors
information to provide marketers with a picture of a client’s consumers by allowing for behavioral analytics to take
place and the creation of important segmented clusters. Of course an enterprise can do the same by merging the same data sources
whether it’s their cookie and form databases with other operational legacy systems. Cookies are the key to
tracking and creating behavior profiles – while forms are excellent sources of customer provided demographics –
by merging them an enterprise can leverage all that valuable information to provide them relevant content or information on
specific products or services. The key is to use cookies along with other user provided information such as their zip code,
age, gender or other personal information to create clusters of segments or buckets of behaviors enabling the profiling of
consumer tribes. As we have discussed cookies are frequently
used to carry out session management, to identify users and to store user preferences. So far cookies have been limited to
the Web, however today many users use mobile devices to conduct searches and communicate with others. The
problem for using mobile cookies is that most devices do not implement cookies, for example Nokia only supports cookies on
60% of its devices, while Motorola only supports cookies on 45% of its phones. In addition, some gateways
and networks (Verizon, Alltel and MetroPCS) strip cookies, while other networks simulate cookies on behalf of their mobile
devices. There are also some discrepancies in the wireless markets around the world in which the ratio of devices not supporting
cookies vary, for example in South Africa 4%, UK 6% and in the US a whooping 43%. The support of cookies is greater in the Far
East, where wireless devices are more commonly used to access the Web. Behavioral analytics can also take
place on mobile devices via the use of mobile cookies, a practice already in place in Japan, so that whether watching a podcast,
a video, TV, clicking on a loan calculator or a GPS map – irregardless of the device cookies can be set for tracking
and capturing consumer behaviors for grouping into tribes. Behavioral analytics can absorb all of this
rich, IP-addressable information across wireless devices which has a greater penetration in Asia followed by Europe.
The future of wireless marketing, behavioral analytics and wireless cookies will continue to grow as more and more
devices are sold in the United States. For example, several years ago AT&T filled a patent for a wireless cookie in
anticipation of the value these mechanisms can provide; the following is an abstract of the application:
“Location information for a wireless device is obtained at the wireless device using a positioning unit such
as a global positioning system. The location information is stored on the wireless device. The wireless device operates a
client engine with a microbrowser or the like to communicate with a web server. The microbrowser and wireless device support
the reception of cookies from the web server. The user of the wireless device is presented with an option of adding the location
information associated with the wireless device to the cookie. If the user approves, then the location information is added
to the cookie information and transmitted to the web server. The web server may then utilize the location information and
present content to the wireless device that related to its location, such as weather, traffic conditions, or local hotel/restaurant
services.”
 AT&T patent for wireless cookie (Pub. No. WO/2002/102025) The big issue with cookies in mobile has been downloading
to devices and the stability of the cookie itself, and as previously mentioned there is the problem of carriers who have been
known to strip out cookies at the carrier gateway. Ringleader Digital offers a solution to these current limitations by its
new product, Media Stamp, which is comparable to an online cookie, but instead of being browser-based, it’s on the server
side. Media Stamp compiles profiles of each device by tracking mobile
online usage patterns across nearly one hundred discriminators that include device type, geography and mobile carrier information.
As with web cookies these wireless server cookies are anonymous. The “stamping” of mobile device enables
marketers and enterprises to identify unique visitors and track their clicks, impressions and acquisitions across all browsing
sessions, mobile sites and wireless carriers. The use of cookies is
also being performed by search engines such as AOL, Yahoo and Google. They set cookies based on what users
are searching for, such as trip, a car, content, etc., they differ in how long they retain those cookies. Some
of these search engines maintain their cookies for different periods, some look at what a user did a few days earlier to show
them ads about the same topic today. Google retains only the most recently searches – others set
their cookies to expire into a month, a week, or a day. Google goes only as far back as your current searches:
time is money.
C H
E C K L I S T:
A list
of issues to consider in your use of cookies for the creation of rich web data streams:
1. What is your privacy policy? Document and state in your site exactly what information
is being captured by your cookies, and your third-party vendors.
2. Are you using first-party
cookies? Do not rely solely on third-party cookies since they can be destroyed by anti-spy software and rejected by some browsers.
3. What is the purpose of setting cookies?
Are you using them to create a personal experience for your consumers, or to track unique visitors and sessions?
4.
What are you capturing? Where visitors are coming from, other websites, ad campaign, search engines? How often they visit
and what pages are most popular? Or, unique user attributes?
___________________________________________________________________________ Cookies, forms, beacons and even widgets make it possible for marketers and enterprises to deliver highly relevant
content to tribes of consumer segments, defined by their own behaviors, at the right time, irregardless of where they are
and what digital device they are using. Behavioral targeting mechanisms such as web and wireless cookies
can be used to anonymously monitor and track the content users read and the sites they visit by a designated unique identification
number or even an IP address as that user surfs the Internet or by their GPS location as they use their mobile devices while
they make their way across the physical world. These
cookies can be set by enterprises, marketers, ad networks, search engines, ISPs and wireless carriers. These
tracking codes, implemented as cookies, are the foundations for behavioral analytics enabling marketers and enterprises to
construct buckets of consumer profiles. These buckets of behaviors can represent sites visited, content viewed, searches done
via the web or wireless devices all of which can be stored in central databases and analyzed to predict multiple behavioral
patterns. All of these tracking mechanisms can be used to enhance web data streams but a relatively new
technology can make it even better: deep packet inspection (DPI).
Strangely
the improvement of behavioral analytics was made possible by a U.S. government-ordered Internet wire-tapping directive known
as the Communications
Assistance for Law Enforcement Act (CALEA) a United States wiretapping law passed in 1994.
CALEA allows ISPs
to block, shape and prioritize their traffic. However, CALEA was originally intended to preserve the ability of the FBI to
conduct electronic surveillance of Internet Protocol (VoIP) services. The core technology allowing for
CALEA is DPI –
which allows for intrusion detection and network security – but most importantly for enterprise and marketers it also
allows for the construction of very sophisticated models of consumers behaviors. DPI, which is also known as complete packet inspection
and Information eXtraction (IX) is a form of network packet filtering that examines the data rather than just
the header of a packet as it passes an inspection point, usually searching for viruses, spam, intrusions or some other predefined
criteria in the form of filtering IF/THEN rules to decide if the packet can pass. DPI
devices have the ability to inspect Internet traffic beyond the first level (packet header) to up to seven levels so they
can identify types of applications and the actual content of packets – so that rather than just
reading the header of an email, DPI can read its content. DPI devices have the ability to look at
Layer 2 through Layer 7 of the Open Systems Interconnection (OSI) model which is an abstract
description for layered communications and computer network protocol design. DPI allows enterprises, ISPs, marketers, service
providers and governments to implement a wide range of applications. Of course for our purpose it allows for behavioral analytics
of web data streams at a highly precise level. Layer 7 is the application layer, so that the actual messages sent across the Internet can be decomposed by DPI whether
it is from a browser like Firefox, a VoIP provider like Skype, or a BitTorrent program like Vuze. Because
ISPs route all of their customers' traffic, they are able to monitor web-browsing habits in a very detailed way allowing
them to gain information about their customers' interests, which can be used by enterprises and marketers for behavioral
analytics. Consumers
have been segmented by the traditional data points such as their zip code or FICO cluster – however, with DPI they can
now be segmented by seven layers of detail such as what network they used – this adds a totally new dimension to behavioral
analytics. DPI data offers a new dimension to understanding and serving consumer in a more relevant manner. DPI technology solution providers include Feeva Technology, NebuAd, Front Porch and Phorm, while DPI hardware providers
include Procera, Narus and Ellacoya. ISPs that use DPI include Knology, Charter Communications, Wide Open West, and Embarq
most of which provide bundled cable television, high-speed Internet, and telephone service. The following
is a brief description of what each DPI vendor offer: The Procera devices can detect more than 300 application protocol signatures. Harbor Network’s
Ellacoya also claims its hardware can look deeper than protocol and is capable of identifying particular traffic from such
sites as YouTube or Flickr. Narus offer four suites of hardware mainly aimed at network security but their
boxes like those of Procera and Ellacoya also perform DPI from Layer 2 to layer 7 and as such could be use for behavioral analytics of web and wireless data
streams. DPI hardware providers
such as Procera, Ellacoya and Narus allow for the construction of models across a grid of applications so that an enterprise
or marketer can create buckets of behaviors based on say the type of browser or website used by consumers – they add
a new dimension to behavioral analytics: they are the new zip codes, except here the attributes of the models have to deal
more with the seven layer features of data packets coupled with their associated human behaviors. An added
advantage to these DPI boxes is that they operate in real time, with web data streams throughput of up to 30 Gb.
Feeva Technology works to identify, target and deliver relevant and useful information to their users, in collaboration
with online media, content, advertising and search services. Their software platform provides DPI services on hotspot, hotzone
and municipal WiFi networks. Feeva was
recently awarded a US patent, this is the abstract: “According to some embodiments of the present invention, a system, apparatus and
method of network operation and information processing, including data acquisition, data processing, data provision, and/or
data interoperability features is presented. In some exemplary embodiments, the method includes registering users logging-on
to a computer network and gathering user-related information from users. In one or more embodiments, user-profile and location-centric
information for each user may be gathered and/or processed in connection with processing targeting and content information.”
Front Porch provides
sidebar and pop-up advertising or service messages to ISPs. Their technology enables an ISP to insert its own messages to
be presented to users as they use their browsers, such as customer service notices. Their technology can
be used by both wireless and broadband ISPs and can be used to manage their ad campaigns or billing messages.
Front Porch is using DPI to build a categorized dossier of interests of consumers, primarily in Asia but is expanding
quickly into the United States and other countries as behavioral analytics becomes more prevalent with marketers and enterprises.
The
two companies that are clearly looking to partnering with ISPs for behavioral analytics using DPI are Phorm and
NebuAd, how they go about it however is a bit different one relies on cookies while the other uses JavaScript,
both however also use DPI. Their DIP equipment performs a couple of things, first they insert their anonymous
cookie that uniquely identifies each ISP subscriber and reads every web page that the subscriber has asked for and creates
a profile of the subscriber’s interests based on a predetermined checklist. Phorm uses cookies and a “307 Temporary Redirect” command to re-route traffic
to get users tagged with their cookies. That is, if a user wants to go to www.jesusmena.com their ISP – if in partnership with Phorm – will intercept the request and route it to www.webwize.net a domain owned by Phorm who then issues the user a first-party cookie with a unique identification number.
Phorm then send another 307 temporary redirect telling it to go www.jesusmena.com the Phorm server also sends back a webwize cookie, but it’s placed in the jesusmena.com domain and becomes
another first-party cookie. Once the query arrives at jesusmena.com has a group of cookies including one from jesusmena.com
and one from webwize.net. The results of the original query are scanned by Phorm for key information used to create a browsing
profile of the consumer. NebuAd approaches the stamping of cookies via
ISP’s with a different angle, first it does not intercept the user’s query to www.jesusmena.com instead it waits until the query is answered as the last packet reaches the ISP, the NebuAd server injects
one packet to the end of the traffic from the jesusmena.com server. The final packet contains JavaScript, which causes the
user browser to go and retrieve a bit of scripting code from the NebuAd web site which places a cookie on the user’s
machine. Both the Phorm and NebuAd cookies track all users activities in order to create buckets of clusters for segmenting
users into unique tribes for advertisers to target. Today, with the advent of the web, mobile, chat, text, blogs
and email, new real-time behavioral analytics are required for a faster and more relevant way to interact with consumers as
events take place. Data warehouses were built for reflection not reaction, which is what is required for
behavioral analytics. To enable enterprises to intelligently interact with their current and future customers
– for sustained streams of revenue growth and superior customer service – networks of behavior models need to
be created and linked. Website provides a gold mine
of consumer data, everything from browsing behavior and patterns, to demographics, transactional histories, sources of online
traffic, the effectiveness of search marketing, changes in conversion, keyword drivers, cross selling propensities and average
order values. Metrics and adjustments to consumer behaviors are paramount, the key challenge is deciding on what core business
drivers each marketer or enterprise needs and wants to capture in order to drive the growth profits for their company or client.
Most importantly enterprises need to leverage the information
they currently posses as part of their overall marketing and customer service efforts. Enterprises need
to form a strategy for capturing and analyzing the behavior of future customers – they need to ask themselves “who
are my customers now, and who is likely to be my future one?” Knowing the core features of its customers
is critical and crucial to enterprises, which behavioral analytics can provide. Every enterprise has streams of transactional
and behavioral data flowing to it 24/7 but few of them are able to mine them simultaneously as events take place – enabling
them to make relevant offers to their new and existing customers – at the moment they interact with them, irregardless
of the medium: text, phone, email, web, or storefront. Careful and strategic planning by enterprises can
leverage customer behaviors enabling to mutually benefit both.
Digital enterprises however must be aware of the valuable intelligence
that is flowing to them from their current and future customers – they must be proactive and aggressive in formulating
a strategic plan for capturing and leveraging all of these streams of customer data. With every customer
‘event’ consumers are communicating with companies their needs and desires. Behavioral analytics
is leveraging these customer events, most of which start at their website but cascade across other operational systems within
an enterprise. Marketers and enterprises have the option to collect personally identifiable information from their
clients and customers, but it is really not needed for behavioral analytics. The process is focused on users’ behaviors
at a site to create a group of anonymous user profiles and to cluster them into tribes with others that have displayed similar
behaviors. As such the process of behavioral analytics is not about tracking users as individuals, but
as members of a behavioral-based group -- thus ensuring that the user's privacy is strictly protected. In
this manner relevancy and privacy can be executed and extended to consumers and clients. As far as a web data
stream strategy goes – total transparency is the only recourse for enterprises and marketers – in other words
don’t make claims that cannot be delivered – the Web and social networking can destroy an ad campaign or a company
within days if not hours. This is because the Web is a rapid self-evolving marketing ecosystem in which
the consumer drives demand, product design, service features, price structure and total production. Today, the 24/7 consumer and marketplace drive the success of products or services. Knowing this ahead
of time will ensure a successful use of web data streams for marketers and enterprises. Designing and implementing
a framework for creating and leveraging streams of consumer behaviors requires advanced planning on how cookies or DPI will
be defined, designed and constructed for both marketers and enterprises. Key building blocks to leveraging
web data streams includes having a business vision, goals or missions, objectives or milestones, as well as tactics, techniques
and a plan: § Business Vision: What the marketer or enterprise will strive for, more often
then not it is increased sales or profits, but it should also be improved customer service and relevancy §
Goal or Mission: This can be qualitative
in nature, such as “improve the consumer experience.”
§
Objective or
Milestone: This must
be quantitative, objectives are the metrics by which thing that must be done (cross-selling, up-selling ratios for example)
to accomplish a goal or mission. § Tactics: These are the executable action items needed to carry out the
overall strategy, and are used to achieve objectives. § Techniques: These are the systemic procedures by which
tasks are accomplished. § Plan: Lastly, we arrive at the set of tasks that lead to the achievement
of all the objectives. eLoyalty.com
has trademarked the term of Behavioral Analytics™ for a call site alert system, which automatically identifies customers who had multiple unsatisfactory contact center interactions over a
short period of time. The challenge for enterprises is being able to construct the framework for leveraging
behavioral analytics for identifying profitable, satisfied and loyal customers across all channels, but most importantly at
their web site via the strategic use of their web data streams.
|
