Docstoc

Behavioral Targeting & Personalization using Audience Segmentation

Document Sample
Behavioral Targeting & Personalization using Audience Segmentation Powered By Docstoc
					BEHAVIORAL TARGETING & PERSONALIZATION USING
          AUDIENCE SEGMENTATION




                          TECHNICAL WHITEPAPER




                             © 2011 Ohana Media

                              All Rights Reserved




       New Jersey. California . Hyderabad. Mumbai.• www.ohana-media.com
                                           Table of Contents


Table of Contents                                                                                             1	
  
Executive Summary                                                                                             1	
  
Why Segment?                                                                                                  2	
  
       All Users are Not Equal                                                                                 2	
  

       Actionable Insights from Segmented Data                                                                 3	
  

       Segmentation from a Publisher’s Perspective                                                             5	
  

Creating Audience Segments                                                                                    8	
  
       Rules Based Segmentation                                                                                8	
  

       Types of Rules                                                                                          8	
  

Using Audience Segments                                                                                     10	
  
       Understanding Audience Behavior                                                                       10	
  

       Personalize On-site Content                                                                           10	
  

       Behaviorally Target Ads On-site                                                                       11	
  

       Personalize Ad Creatives Off-site                                                                     12	
  

       Re-market to Audience Off-site                                                                        13	
  

       Extend Reach for Publishers                                                                           13	
  

       Personalize Email Communication                                                                       14	
  

Integrating Audience Segments with External Systems                                                         16	
  
       Integration with Ad-Servers                                                                           16	
  

       Integration with Ad Exchanges & Supply Side Platforms                                                 16	
  

       Integration with Content Management Systems                                                           17	
  

OhanaQB Audience Clustering Engine                                                                          18	
  
       Handling Big Data                                                                                     18	
  

       Real-time v/s Offline Processing                                                                      18	
  

Ohana Media                                  Behavioral Targeting & Personalization Using Audience Segmentation

                                                       i
       Behavior Segment Lookup                                                                   18	
  

       Content Analysis                                                                          18	
  

About Ohana Media                                                                               20	
  
       Our Mission                                                                               20	
  

       What We Do                                                                                20	
  




Ohana Media                      Behavioral Targeting & Personalization Using Audience Segmentation

                                           ii
                               Executive Summary
In the ‘offline’ world, the concept of customer segmentation is a very familiar one to
marketers. It is an accepted practice for companies to offer differentiated solutions and
messaging to customers based on variables such as age, geography, past purchase
behavior etc. In the web analytics domain however, most reporting and analysis still
happens at an aggregate level. This absence of sophisticated audience segmentation tools
for online audiences represents a huge opportunity to improve user engagement and the
cost efficiency of online marketing. The OhanaQB Audience Clustering/Segmentation
Engine addresses this by giving marketers and publishers the tools to segment, analyze
and then target audiences online using a wide range of tools.

OhanaQB supports a variety of rules to create audience clusters using a combination of
AND & OR conditions. This segmentation enables marketers to then take customized
actions for specific clusters of audiences. These actions include:

   •   Understanding audience behavior

   •   Personalizing on-site content

   •   Behaviorally targeting ads on the site

   •   Delivering personalized ads on other sites

   •   Re-marketing to audiences after they leave the site

   •   Extending reach for publishers

   •   Personalizing email campaigns

Each of these actions can significantly improve the efficiency of campaigns and level of
user engagement.

The OhanaQB Audience Clustering/Segmentation Engine is built keeping in view the
need for efficient data storage and processing of terra-bytes of click-stream, demographic,
campaign, content analysis and custom event data. The whitepaper concludes by
describing the high level architecture of this clustering engine.




Ohana Media                            Behavioral Targeting & Personalization Using Audience Segmentation

                                                 1
                                   Why Segment?
All Users are Not Equal
In the ‘offline’ world, the concept of customer segmentation is a very familiar one to
marketers. It is an accepted practice for companies to offer differentiated solutions and
messaging to customers based on variables such as age, geography, past purchase
behavior etc.

In the web analytics domain however, most reporting and analysis still happens at an
aggregate level. For example, a common aggregate metric a marketer might be looking at
is bounce rate. While an aggregate bounce rate may be a broad performance indicator
when compared to the industry average, it does not give marketers any specific insights
that are actionable.

However if users were segmented based on where they came from - e.g. direct traffic, paid
search, natural search, social media, email newsletters, banner ads – one may see that
bounce rates are significantly higher in one or more segments. Let us assume the bounce
rate is far higher in the case of paid search - the immediate action that one can prioritize is
to look at the paid search campaigns carefully to identify why users are leaving the site
without further engagement. Perhaps the paid search landing pages are not informative
enough? Maybe the keywords being used are the wrong ones to focus on? Alternatively
the search ads may be misleading and not communicating the real proposition?

This absence of sophisticated audience segmentation tools for online audiences represents
a huge opportunity to improve user engagement and the cost efficiency of online
messaging. OhanaQB addresses this by giving marketers and publishers the tools to
segment, analyze and then target audiences online using a wide variety of features.

The basic principles of online segmentation can be illustrated by looking at outcome data
such as conversions, transactions and ticket size for a hypothetical online e-commerce
store XYZ.com. As illustrated in the chart on page 3, many e-commerce providers track
aggregate transactions and conversions on their site.




Ohana Media                           Behavioral Targeting & Personalization Using Audience Segmentation

                                                2
Actionable Insights from Segmented Data
While looking at the overall conversion rates in the purchase funnel might be useful there
aren’t many immediately actionable insights that can be gleaned from these metrics. For
example, clearly not all users convert at the 1% average.




Ohana Media                         Behavioral Targeting & Personalization Using Audience Segmentation

                                              3
Let us assume XYZ.com creates segments of users based on the sort of products they
navigate to, once they are on the website.

Now XYZ.com can look at metrics at a segment level, and find out how each segment
behaves - this can generate actionable insights. For example, which segments convert best
and have the best profitability could be an input into the strategy for buying traffic - if the
Apparels & Fashions users are more profitable for XYZ.com, maybe a larger share of the
media budget needs to be allocated to attract users who would be interested in this
category. This type of segment-based customer analysis can become much more granular,
giving marketers very specific actionable insight. For example, OhanaQB supports a wide
variety of customizable rules to create audience clusters using AND & OR conditions. This
enables marketers to take customized actions for specific clusters of audiences. This is
discussed in detail later in the whitepaper.




Ohana Media                           Behavioral Targeting & Personalization Using Audience Segmentation

                                                4
Segmentation from a Publisher’s Perspective
We have discussed the benefits of granular audience segmentation from a marketers’
perspective. Let us now look at it from a web publisher perspective. For a web publisher,
the most important business driver is improving yield on inventory. Audience
segmentation is one of the most effective ways smart publishers can increase their
revenues from their existing inventory.

Let us take a step back and look at the airlines industry, since like web publishers, they
own perishable inventory. Airlines segment their seats into different classes - business
class, economy etc. Each of these segments has different needs and willingness to pay.
Airlines maximize their yield by charging different prices to users in different segments,
instead of using a flat pricing for all seats on a plane.

Similarly, most publishers already use basic segmentation in their pricing strategy. For
example, they may charge different amounts for a Run-of-Site (ROS) versus homepage
inventory or Run-of-Category. However, this sort of segmentation is purely based on
inventory or page views, and not on the specific characteristics of the user.

Alternately (and increasingly smart publishers are adopting this), publishers can create
segments of their audience - and start selling Audiences rather than Page views. Using
Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                               5
OhanaQB, a publisher could create custom audience segments on their website - collect
metrics like page views, ad impressions available, engagement metrics etc, and approach
advertisers with a list of available segments to advertise to. For example, a sample rate-
card could look like this:



              SEGMENT            AVAILABLE IMPRESSIONS                            CPM


 Auto-Intent                    5.4 mm impressions/0.5mm                           $ 22
                                uniques per day


 Real Estate-Intent             9 mm impressions/0.3mm                             $ 18
                                uniques per day


 Shoppers                       2 mm impressions/0.1mm                             $ 10
                                uniques per day



Following this strategy has four distinct benefits for publishers:

   •   Advertisements shown to users are more targeted, especially when users visit
       generic news or user generated content pages, because they see ads relevant to their
       past behavior. This increases engagement rates with the ads, and ads on the site
       perform better for the advertisers.

   •   They can create many more segments than basic inventory classes - allowing them
       to have granular differential pricing and charge a premium for targeted audience
       segments.

   •   Yields on low value inventory like Run-of-Site (ROS) or Entertainment/User
       generated Content pages can be increased significantly. This is because instead of
       charging a low Cost-per-Thousand impressions (CPM) for these inventory pockets,
       publishers can target a high value user belonging to the Auto-intent segment -
       when he is visiting the entertainment section, and charge a high CPM for this user.

   •   They can use their own audience data to buy additional inventory off ad-
       exchanges/supply-side platforms to augment their campaigns. This enables them
       to make money on their audience, even after they leave their site.




Ohana Media                           Behavioral Targeting & Personalization Using Audience Segmentation

                                                6
Owing to these benefits, some ad-servers have started including bare-bones audience
segmentation functionality into their ad-servers. For example, OpenX has introduced
Audience Segmentation as part of its core ad-serving offering. The questions publishers
should ask while evaluating audience segmentation tools that are part of ad-servers are:

   •   Can I use this data outside my ad server ? Can I easily integrate my audience
       segments with other data?

   •   Is the process of creating segments comprehensive and relevant to my business?
       Can I create a variety of different behavior segments?

   •   Can I tie in on-site behavior data with intent data from my acquisition campaigns?

   •   Is the segmentation real-time?

   •   Can the audience segmentation tool capture custom data and use it to build
       segments?




Ohana Media                         Behavioral Targeting & Personalization Using Audience Segmentation

                                              7
                        Creating Audience Segments
Rules Based Segmentation
Theoretically, online audience segmentation can be done either though user-defined rules
or by using automated system generated algorithms. OhanaQB adopts a user-defined
rules approach. While both approaches have their merits, user defined segmentation has 3
significant practical benefits:

   •   Automated segmentation often yields a set of users who are similar in behavior
       according to a machine learning model, but users of the audience segmentation tool
       i.e. marketers, sales, ad operations - are not able to articulate the unique
       characteristics of the segment. This makes it difficult for real life use - especially in
       explaining it to decision makers or, in the case of publishers, to advertisers.

   •   A user defined rules-based approach forces discipline in thinking through the
       business and audience need and characteristics.

   •   Rules-based systems are simpler to implement and use.

In the section below, we will look at the types of rules supported by the OhanaQB
Audience Clustering Engine. In OhanaQB, each segment/cluster is a logical combination
of one or more rule instances. The logical operators supported are AND & OR.

For example cluster X could be defined as, Cluster X = (Rule A OR Rule B) AND Rule C

Types of Rules
OhanaQB supports multiple types of rules - let us explore the rules supported by
OhanaQB as an illustration of the usual ways to segment an audience:




Ohana Media                           Behavioral Targeting & Personalization Using Audience Segmentation

                                                8
Came From Rule: All visitors who came from specific web sites as shown in the referrer
header.

Visited Page Rule: All visitors who visited a specific URL or a page with a specific title.

Geography Rule: All visitors who are located in a specific set of geographic locations
based on IP-Geography mapping.

Keyword Rule: All visitors who arrived at the site after searching on specific keywords or
used specific keywords in internal site-search.

Event Rule: All visitors who performed a custom event a specific number of times. Each
event can be passed a set of attributes as name value pairs, and segments can be created
based on specified range of values of a given attribute. This rule can be used to create
segments of users like “All users who abandoned shopping cart after adding products
worth more than $3000”, if total price of products in the shopping cart is passed as an
attribute of a custom event.

Smart Rule: All visitors who showed High/Medium/Low affinity towards content
containing specific keywords or variations.

Category Rule: All visitors who visited content pieces in a specific content-sensed
category.

Ads Clicked Rule: All visitors who reached the site after clicking on specific campaigns or
ad groups

Profile Rule: All visitors who fall in a specified demographic parameter like gender, age,
household income etc. This rule can only be applied if you choose to pass demographic
data about your registered users

Each Rule can be applied on a dataset within a given range of days - we recommend not
creating clusters/segments beyond a 90 day data-window - beyond 90 days the value of
the data is questionable because of issues like cookie deletion. Each rule is also associated
with attributes like frequency and recency.




Ohana Media                           Behavioral Targeting & Personalization Using Audience Segmentation

                                                9
                          Using Audience Segments
The power of audience segmentation lies in the multiple ways in which it can be used to
engage and target users. OhanaQB has the functionality to support multiple actions:

   •   Understanding audience behavior

   •   Personalizing on-site content

   •   Behaviorally targeting ads on the site

   •   Delivering personalized ads on other sites

   •   Re-marketing to audiences after they leave the site

   •   Extending reach for publishers

   •   Personalizing email campaigns

Each of these actions can significantly improve the efficiency of campaigns and level of
user engagement, and are described in more detail below.



Understanding Audience Behavior
Marketers can analyze all their key metrics at a segment level to uncover specific
actionable insights. This can lead to a clearer understanding of how different types of
users behave on the site. Do users with intent X spend more time on specific sections of
the website versus users with intent Y? Does intent X convert better on certain types of
offers? How many users does the site get each day which have intent X ? How many of
these visitors come back to the site?



Personalize On-site Content
One can show different content and provide a different browsing experience to a
particular user on the site based on what is known about that particular user. For example,
it is possible to integrate OhanaQB with an existing CMS with a simple script. When a user
comes to the site, this script calls the OhanaQB Behavior Lookup Server, which returns
back with a list of segments that the user is a member of. One can then program the CMS
to show specific pieces of content to users who belong to specific segments.

Alternately, one can also use a feature of OhanaQB that lets you target different content
pieces without having to modify code in the CMS. For this, one needs to place the
OhanaQB Audience and On-site script on all pages of the website, login to the OhanaQB
interface, and associate one or more on-site personalization targets to a particular segment.
Let us assume that the content to be personalized - maybe a snippet of HTML or an image
Ohana Media                            Behavioral Targeting & Personalization Using Audience Segmentation

                                                10
- has an HTML ID of “xxx”. While creating the on-site personalization target on OhanaQB,
one needs to provide this ID, and an alternate HTML that one wants rendered when a
visitor is a member of a particular audience segment. The OhanaQB onsite personalization
script will do a client side over-write of the targeted HTML content with the new
personalized content.




Behaviorally Target Ads On-site
The OhanaQB Behavior Server can be integrated with existing ad-servers easily. Most ad-
servers allow either or both of these options:

   •    Create campaigns targeted to specific cookies set using the Ad-server domain.

   •   Create campaigns targeted to specific keywords passed on to the ad tag as a
       parameter.

When a user is part of multiple segments - for example, if a user is part of Auto Intent and
Shopping Intent, the decisioning on which ad to show to the user is left to the ad server. If
the ad-server has campaigns targeted to both of these segments, typically the ad-server
will choose one of the campaigns based on factors like a) expected eCPM b) delivery status
on guaranteed/non-guaranteed campaigns c) frequency caps etc.


Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              11
The Ad Operations team needs to be aware of the segments available, and the values of
cookies/keywords which they need to use while setting up targeting criteria on the ad-
server.




Personalize Ad Creatives Off-site
Dynamic Creatives can be used to tap into data about user’s behavior to show them a
tailored message, when one comes across the user on a different site. OhanaQB will serve
the HTML for the creative, but it can be trafficked in the publishers’ own ad-server. If the
code for rendering the creative is HTML, it needs to call the OhanaQB Behavior server and
look up the user segments and details on past events performed by the specific user, and
use that to render a personalized banner for this specific user.

Let us look at a sample example - User X visits travel advertiser A’ site, searches for
airline tickets between Mumbai and Delhi in the next 1 week, and leaves without
completing the transaction (in order to later shop around for prices on other websites ).
User X then visits a news website on which advertiser A has bought inventory based on a
high bid for this particular user who has a specific cookie. The creative that gets served to
user X, can call the OhanaQB Behavior server, look up the user’s origin & destination cities
and show the user specific deals on Mumbai to Delhi flights if booked in the next 3 hours.
The creative can be coded to take the user to the appropriate dynamically generated
landing page for this specific offer as well.


Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              12
Re-market to Audience Off-site
Advertisers spend thousands of dollars trying to get intent-rich traffic to their website. At
any given point, a large number of these visitors leave the site without taking an action -
making a purchase, subscribing to an email list, filling up a lead form etc. Re-marketing is
a technique used to reach out to visitors to the website who have not converted yet, and
show them banners when they are on other websites, which allows one to bring them back
to the destination site. In most cases re-marketing shows a lift of 200-500% in conversion
rates versus regular media buys.

While re-marketing/re-targeting has been around for 10+ years, this technique has gained
prominence over the last two years because of the increase in liquidity in the display ad
marketplace. The challenge in re-marketing earlier was to have access to enough inventory
in order to increase the probability of finding the same user on other websites. With the
increased usage of Ad Exchanges and Real-time Bidding by advertisers, access to large
amounts of inventory, on which to find a re-marketable user, is not a problem anymore.

OhanaQB makes the re-marketing process simple by:

   •   Increasing the granularity at which re-marketing can be done. Marketers can create
       segments retroactively, and test out the ROI on re-marketing to different segments.
       This is usually not possible while trying re-marketing with most ad networks - one
       would typically have to insert a different pixel for every area of the website that
       one wants to use as a separate catchment area for capturing users for re-marketing.

   •   Providing ready-made access to inventory across ad exchanges and various real-
       time bidding enabled inventory sources.

   •   Allowing the use dynamic creatives in conjunction with re-marketing to increase
       the effectiveness of the campaign.

   •   Helping to re-market to users who have clicked on specific intent-rich search
       keywords or other intent-rich campaigns one may have run



Extend Reach for Publishers
Publishers with high intent audiences - for example, vertical sites or e-commerce related
websites with significant amount of internal site search traffic, can use audience
segmentation through OhanaQB to generate incremental revenue. Typically, such
websites are fully sold out on their own inventory, because of availability of limited
number of intent-rich impressions. Most of these publishers already command high
eCPMs because of the targeted nature of inventory available with them. These publishers
can create their own audience segments, and buy ads off various inventory sources, on the
same audience, and re-sell the same to advertisers. The proposition that these publishers
can take to advertisers could be “Buy our audience, on our site, as well off our site”. Many
Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              13
of these publishers currently work with data brokers, and sell their data, but we believe
that its critical publishers control, own and monetize their own data. The power in the
display advertising marketplace is shifting from media owners to data owners, and one of
the ways high quality publishers can ensure they are not marginalized, is by retaining
ownership of their data and monetizing on their own. Moreover, this allows publishers to
not be limited by the amount of inventory on their site and the limited time that their
audience spends on their site.



Personalize Email Communication
Opt-in email to users is one of the best converting mediums. Unfortunately, in many cases,
marketers send emails to users that they are not interested in. This leads to dropping open
rates, click through rates, and subsequently an increased probability of these emails
getting marked as spam. To avoid this, its imperative that emails sent to users should be a)
targeted to specific users based on their interests b) personalized based on what one
knows about the user. In practice, this has been difficult to implement, without creating
custom applications that collect and segment user data, and push that data to the emailing
systems.

With OhanaQB, it becomes extremely simple to automatically segment an email list based
on user behavior. OhanaQB integrates with email providers like Mail Chimp to make this
process as simple as possible.

Below is an example of the steps that one needs to take to get started with behavioral
email using OhanaQB, with MailChimp as an example:

   •   Create a mapping between an audience segment/cluster in OhanaQB and a new
       mailing list in Mailchimp.

   •   Setup auto-responder rules or target specific campaigns to different lists defined in
       MailChimp.

   •   OhanaQB, through an API level integration ensures that whenever a new user gets
       pushed into a segment, if the system knows their email address - the user will be
       automatically pushed into the correct mailing list.

Continuing our example of XYZ.com from earlier in the whitepaper- let us say XYZ has an
audience segment defined - “Luxury Watches” - with the segment definition being -
“Anyone who has visited at least 3 pages in Luxury watches section in the last 7 days”.
XYZ’s marketing manager can create a mailing list for LuxuryWatchRecentIntent on
Mailchimp, and add an auto-responder with “Send an email to any new addition to the
list 1 day after addition”. The auto responder is configured to send a mail with the top 10
best selling luxury watches, and offers deep discounts on purchases within the next 48
hours. Now, when a user comes to XYZ.com and spends some time looking at luxury
Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              14
watches, she automatically gets added to the LuxuryWatchRecentIntent mailing list, and
the next day would get an email specifically talking about the top selling luxury watches
on XYZ.com. This makes the email personalized, relevant and timely.




Ohana Media                         Behavioral Targeting & Personalization Using Audience Segmentation

                                             15
     Integrating Audience Segments with External Systems
Integration with Ad-Servers
This is primarily done through either setting the ad-server cookie or passing a
keyword/name value pair to the ad tag as discussed earlier in this document.

Integration with Ad Exchanges & Supply Side Platforms
The most basic form of integration that is supported by all ad exchanges is through a
“piggyback pixel”. In this scenario, for each audience segment, one would need to create a
corresponding pixel on the Ad Exchange. This pixel needs to be called from the client
browser whenever a user is part of a particular segment.

While using OhanaQB, this process is automatic - OhanaQB automatically instantiates
piggyback pixels and calls the correct Ad Exchange pixels corresponding to the segments
that a user is member of. OhanaQB also maintains an updated mapping between the
audience segments defined and the pixels instantiated on the ad exchange. When the pixel
is called, a cookie is set by the exchange on the ad-serving domain used by the ad
exchange. This pixel is usually a 1x1 blank gif or a JavaScript pixel, and its purpose is to
set a cookie that the ad exchange can then use to target campaigns to that particular user.

This process can be time consuming, but OhanaQB simplifies this and manages all these
processes in the background, making it easier for the end user. OhanaQB users can create
audience segments, and launch campaigns targeted to them from the OhanaQB interface
without the need to create or manage these pixels.

Google started supporting re-marketing in the Google Display Network a few months
back. The Adwords API does not yet have support for instantiating and managing re-
marketing pixels from Google. However, marketers can use OhanaQB to manually
piggyback pixels from Google or any other ad network, even if they do not have API
support for pixel instantiation and management.

With rapidly increasing adoption of real-time bidding, supply side platforms like Admeld,
ContextWeb, Pubmatic etc., have emerged as a viable option to find audience members for
re-marketing. One of the advantages of real time bidding, is that it avoids the need to call
multiple pixels from the inventory source, as described above. In case of RTB, OhanaQB
maintains a mapping between OhanaQB audience cookie and the cookie set by the supply
side platform on a particular user. This means, that as an advertiser, there is no need to
call an additional piggyback pixel when a user is visiting the site - OhanaQB manages the
mapping between this user’s unique identity on the website, and on the supply side
platforms.




Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              16
Integration with Content Management Systems
This is done by calling OhanaQB Behavior Server to look up the segment membership
information for a given user, and then programming the CMS to show different content
pieces based on which segment a user is a member of. This has been discussed earlier in
the document.




Ohana Media                         Behavioral Targeting & Personalization Using Audience Segmentation

                                             17
                OhanaQB Audience Clustering Engine
Handling Big Data
The OhanaQB Audience Clustering/Segmentation Engine is built keeping in view the
need for efficient data storage and processing of terra-bytes of click-stream, demographic,
campaign, content analysis and custom event data. Our platform runs on the Amazon EC2
cloud that lets us scale the volume of data our platform can handle, on-demand. The
platform handles processing and storage of both structured and unstructured data at scale.
The audience clustering engine is built using a stack of open source technologies including
Hadoop, PIG and Nutch.

Real-time v/s Offline Processing
When a user visits V a particular page P on a website, the OhanaQB audience script
passes back a number of pieces of information to the Ohana Behavior server, and also does
a lookup on the segments that the user belongs to, and sets the appropriate cookie values.
There are two different ways in which a user is classified into a particular segment - real-
time bucketing and offline classification. OhanaQB provides a variety of rules that can be
used for creating segments - some of these rules can be evaluated at real time, while some
requires extended offline processing. In case of segments whose membership can be
evaluated in real-time, the OhanaQB platform associates the user with the particular
segment within 75 ms. Segments which require offline processing - for example, when a
user visits a page P for the first time, and we do not have any content analysis performed
on that particular page in a short historical window, the OhanaQB platform does offline
processing to update each user’s membership of this segment - however, this does not
happen in the same visitor session. Real-time bucketing is extremely important in real life
use-cases because many of the marketers users might not be repeat visitors - systems
which depend purely on offline processing, are unable to store segment membership
information in cookies for the users who came to the website only once - and hence leave
aside personalization and re-marketing opportunities on a large percentage of users to the
site. At the same time, offline processing allows building richer user profiles by doing a
deeper level of analysis.


Behavior Segment Lookup
As part of the OhanaQB platform, the marketer is provided scripts/APIs to lookup the
segment membership of a particular user, with a latency of 200 ms. This information can
be used to export segment membership data of any user or sets of users to external
systems - for example, a CRM system or an email system.

Content Analysis
OhanaQB Audience Clustering categorizes every page of content which has the OhanaQB
Audience Script installed, and does further content analysis at a keyword level in order to
Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              18
power the Smart Rules and Category Rules in the system. This is especially useful if you
have a lot of user generated content on your site, which is not stored under pre-defined
sections of your website corresponding to each category.

Below is a quick overview of the different components of the OhanaQB Audience
Clustering Engine:




Ohana Media                         Behavioral Targeting & Personalization Using Audience Segmentation

                                             19
                              About Ohana Media
Our Mission
To simplify Digital Marketing in Emerging Markets.

What We Do
We are a technology driven advertising company based out Hyderabad, Mumbai, New
Jersey and California. Our key offering is the OhanaQB Audience Marketing Platform, of
which the Audience Clustering Engine is a key component. OhanaQB integrates across all
digital touch-points of a marketers audience - search, social media, display, email and
direct traffic and helps deliver highly tailored marketing experiences across channels,
leveraging audience intelligence and cross-channel optimization techniques. We offer
OhanaQB as a Software-As-A-Service offering to marketers, agencies and publishers, in a
self-service or assisted self-service mode. We also deliver full digital marketing solutions
to marketers - all our solutions are delivered using the OhanaQB platform, with
appropriate custom solutions and human intervention where required.

Reach us at aloha@ohana-media.com to know more about how we can help you increase
the ROI on your digital marketing efforts.




Ohana Media                          Behavioral Targeting & Personalization Using Audience Segmentation

                                              20

				
DOCUMENT INFO
Description: In the ‘offline’ world, the concept of customer segmentation is a very familiar one tomarketers. It is an accepted practice for companies to offer differentiated solutions andmessaging to customers based on variables such as age, geography, past purchasebehavior etc. In the web analytics domain however, most reporting and analysis stillhappens at an aggregate level. This absence of sophisticated audience segmentation toolsfor online audiences represents a huge opportunity to improve user engagement and thecost efficiency of online marketing. In this whitepaper we look at how audience segmentation data in the online world can be used for behavioral targeting and personalization of marketing messages.