Ad Serving Technology

It provides a mechanism for optimizing bid prices, ad placement, ad targeting etc.

Significant methods include Behavioral Targeting, Contextual Targeting, and  Optimization through Experimentation

A. Behavioral Targeting

– Using a profile of prior behavior on the part of the viewer to determine which ad to show during a given visit. For example, targeting car ads on a portal to a viewer that was known to have visited the automotive section of a general media site.

B. Contextual Targeting

– Inferring the optimum ad placement from information contained on the page where the ad is being served. For example, placing Mountain Bike ads automatically on a page with a mountain biking article.

C. Optimization through Experimentation

– Using experimental or predictive methods to explore the optimum creative for a given ad placement and exploiting that determination in further impressions.

Ad Network

It is a company website that want to host ads (i.e., Publishers) with advertisers who want to run ads.

For example :

Universal Target Access
With Conversant Media, brands get access to the best opportunities on virtually any device across more than 1 million sites and apps. Brands can reach and connect with an individual at ideal moments virtually anywhere online:
• 20+ billion impressions a day across 25 leading ad exchanges, including the top 10 exchanges
• Integrated buying relationships with Facebook and Twitter
• Deep partnerships with more than 6,000 leading digital publishers
• SDK integration across more than 20,000 mobile and tablet apps

A. Behavioral Targeting – displays Ads Flow

To understand how Google Ad Words work internally read the document ‘Google Search and Advertisements’

Business Process Flow :

Beacon – is an async, non-blocking, real-time tagging approach which helps capture most important site analytics data to enable search , recommendation , optimization, ads etc.

Multiple Beacons Calls for 3rd Party Site Analytics , Ads, Marketing
• Collect Web data
• Map data to corresponding schema of 3rd party server ( Data Distribution Rule)
• Web request to send data to 3rd party server through Partner API Wrapper

  • Third-party Servers: Boomerang , Omniture, BazaarVoice, Google Analytics, Turn Audience , CoreMetrics , Blue Kai , Channel Intelligence, 24/7 , OAS

1. For example : Common Omniture Flow :

• Key : Product Items , var : s_omni.prop1 , used_for : Product Sell , populate_reports : custom traffic > Product Sell Page type , Where : Set on every page On page load ,

• Key : Campaigns , var : s_omni.prop2 , used_for : Campaign Pool , Reports : Custom Conversion > Campaign Pool , location :

This variable is set to the last N non-Direct traffic vehicles (also called the channel) that a user touched in the last 30-days or since the last purchase, separated by a “special char”, ordered by first to last.

Note that if the same vehicle is used twice in a row the value in the vehicle stack will not be shown twice.

Vehicles: eml=Email, aff=Affiliate, dat=Data feed, soc_ac=Social active (tagged), soc_pa=Social passive (untagged), cse=Comp shop, org=Organic, ref=Non-affiliate referrer, sem=Paid search, seo=Natural search

For SEM and SEO traffic the vehicle also includes whether the search query was Exact Brand (_eb), Branded (_br), Non-branded (_nb), and Unknown (_un).

Example of different types of flow

Visitor first came to site via a non-tagged Social site (soc_pa, or Social Passive), then via a paid search using a branded term (sem_br), then via an email link (eml), and finally by typing in an exact brand search such as “Kohls” and clicking on an unpaid link (seo_eb).

Paid Search Detection is a feature within the Admin section of SiteCatalyst that allows paid search to be differentiated from natural search. Paid Search detection looks for a specific parameter that is present on paid search traffic, and is not present on natural search.

2. For example : Commision Junction Flow :

Add the Commision Junction ID in the Redirect Domain of your site.

CJ_ID is Commission Junction Id for the Affiliate
Are you pumped up and ready to leverage the Commission Junction Network? To get started, we need to ask you four relatively painless questions.

• Do you have a live website that generates sales or leads for your company?
• Can consumers purchase products or request quotes directly on your website?
• Do you only want to earn money by generating sales or leads for other retailers or companies?
• Do you want to earn money from merchants by placing ads on your website?

3. For example : TURN Audience Flow

Technical Process Flow :

Beacon – can be implemented by extending Yahoo Boomerang !
Customer → Website → Beacons [fires events (page views, eCommerce actions, search metadata ) ] → Data Analyzer (filters information) → Hadoop

a) mWeb → Anivia (data capture) → > Ominiture User Targeting
➢ Splunk / SumoLogic / Log Analyzer
➢ MQ → Hadoop Sink
b) Site → Third-party Ominiture User Targeting
c) Site → Beacons → MQ → Hadoop Sink

B. Contextual Targeting Flow

Steps :
→ (Data Source) Keywords : select the ones with higher conversion rate
→ Quality Filter (dedupe , remove blacklisted items)
→ Product Association (say top selling products from Analytics API )
• Associate docom breadcumb , Leaf Category, Vendor brand
→ Correct Grammar (remove trademarks , common modifiers )
→ Apply Ad Group (site taxonomy – page category , vendor brands , general terms )
→ Ad Optimization (multiple permutations to get high quality score)
• Achieve relevance : consumer intents to be represented in optimized ad
• Make ad more relevant by changing important keywords
• Once perfect relevance is achieved, test which words/phases/intents are revenue generation
→ Deploy to the Platform