Computational Advertising

Computational Advertising
Ahmad Shah Sultani
M.Sc. Computer Science
South Asian University, New Delhi
13-10-2014

Outline
 Advertising
 History
 Types of Advertising
 Computational Advertising
 Participants
 Business Models in Comp. Advertising
 Landscapes
 …

• Advertising is a form of marketing communication used to
encourage, persuade, or manipulate an audience (viewers,
readers or listeners; sometimes a specific group) to take or
continue to take some action.
 Awareness
 Knowledge
 Liking
 Preference
 Conviction
 Purchase
Advertising

Long History…
 Egyptians used papyrus to make sales messages and wall
posters, date back to 4000 BC.
 In ancient China, the earliest advertising known was oral, as
recorded in the Classic of Poetry (11th to 7th centuries BC) of
bamboo flutes played to sell candy.
 In Europe, as the towns and cities of the Middle Ages began to
grow, and the general populace was unable to read, instead of
signs that read "cobbler", "miller", "tailor", or "blacksmith"
would use an image associated with their trade such as a boot, a
suit, a hat, a clock, a diamond, a horse shoe, a candle or even a
bag of flour.
 In the 18th century advertisements started to appear in weekly
newspapers in England.

 Papyrus
 JAPAN, 1806
traditional
medicine
called
Kinseitan

 A Coca-Cola advertisement
from the 1890s
 A 1900 advertisement for
Pears soap

 On the radio from the 1920s
 Public service advertising in WW2
 Commercial television in the 1950s
 Media diversification in the 1960s
 Cable television from the 1980s
 On the Internet from the 1990s

Types of Advertising
 Television advertising / Music in advertising
 Infomercials
 Radio advertising
 Online advertising
 New media
 Press advertising
 Billboard advertising
 Mobile billboard advertising
 In-store advertising
 Coffee cup advertising
 Street advertising
 Aerial advertising

Classic Advertising
 Brand Advertising
 GOAL: Create a distinct favourable image

 Direct marketing
 Advertising that involves a "direct response”: buy,
subscribe, vote, donate, etc, now or soon

The Advertising Market
 Internet advertising is a business that is growing faster than old media
advertising (radio, TV, newspapers & magazines, mail, outdoors)
 Online advertising budgets still lagging in proportion to time spent online, which
is growing fast at the expense of old media.
 Seen as a driver to continued growth of online advertising market
•Traditional advertising:
 –Few, expensive opportunities
 –Targeting en-masse, by immediate context only
 –Difficult to measure effectiveness
•Internet advertising:
 –Billions of opportunities daily
 –Open to personalization via rich context of impression
 –Effectiveness is measurable: can measure click-through rates as % of
impressions, and conversions as % of clicks

US Online vs. Offline advertising
spend

Lots of computational this and that …
 Computational Biology
 Computational Chemistry
 Computational Finance
 Computational Geometry
 Computational Neuroscience
 Computational Physics
 Computational Mechanics
 Computational Economics
 …
All are about mixing an old science with large scale computing
capabilities

What’s computational about it?
 Classical:
 Relatively few venues –magazines, billboards, newspapers, handbills, TV,
etc.
 High cost per venue ($3Mil for a Super Bowl TV ad)
 No personalization possible
 Targeting by the wisdom of ad-people
 Computational –almost the exact opposite:
 Billions of opportunities
 Billions of creatives
 Totally personalizable
 Tiny cost per opportunity
 Much more quantifiable

 New scientific sub-discipline, at the intersection of
(involve)
 large scale search and text analysis:
 information retrieval: query-ad selection, learning-to-rank.
 statistical modeling
 machine learning: clustering, classification and regression.
 optimization: linear, integer, convex optimization.
 microeconomics: game theory, mechanism design, auction
theory.
 Classifications
 Recommender Systems

Find the "best match" between a given user in a
given context and a suitable advertisement.
 Examples:
 Context = Web search results  Sponsored search
 Context = Publisher page Content match, banners
 Other contexts: mobile, video, newspapers, etc.

Participants: Publishers, Advertisers, Users, &
“Matcher”
 The publisher is the owner of Web pages on which
advertising is displayed.
 The advertiser provides the supply of ads.
 The ad network is a mediator between the advertiser
and the publisher, who selects the ads that are put on
the pages.
 End-users visit the Web pages of the publisher and
interact with the ads.

Landscapes
 Three main types of textual Web advertising:
1. Sponsored search which serves ads in response to
search queries
2. Content match which places ads on third-party
pages
3. Display advertising (banner ads)
 Ads are information!

Sponsored Search
 context: a user issues a query.
 publishers: Google (AdWords), Bing (AdCenter),
Yahoo!
 max: publisher revenue, s.t. advertiser campaign goal,
budget, user satisfaction.
 marketplace: keyword-based GSP with cost-per-click
(CPC) pricing.
 system sketch: query analysis  ad selection and
relevance  click prediction  GSP.

Contextual Ads
 contextual ads: an extension of sponsored search;
 context: page content and user behaviour.
 publishers: content providers, and major search engines
operate the marketplace.
 system sketch: starts with keyword extraction in
absence of user query.

Display Ads (Banner Ads)
 display ads;
Graphical Ads
 context: page, application and user behaviour.
 publishers: content providers in display ad-networks operated
by Google (doubleclick), Microsoft (aquantive), and Yahoo!
(rightmedia).
 two types of display ads:
1. reserved: delivery guaranteed, contracts negotiated upfront,
pricing based on CPM (cost-per-(k)impression), e.g., brand ads,
direct response.
2. performance-based: max publisher revenue, s.t. advertiser
budget, real-time bidding on exchange, pricing based on CPC/CPA
(cost-per-action)/CPM.

Search Advertising Business Models
 CPM (Cost Per Thousand) or Cost per Impression
– Advertisers pay for exposure of their message to a specific audience. (M in the
acronym is the Roman numeral for one thousand)
 Typically used for graphical/banner ads (brand advertising)
 CPC (Cost Per Click) aka Pay per click (PPC)
– Advertisers pay every time a user clicks on their listing and is redirected to their
website.
 Typically used for textual ads
 CPA (Cost Per Action) or (Cost Per Acquisition)
– The publisher takes all the risk of running the ad, and the advertiser pays only
for the amount of users who complete a transaction, such as a purchase or sign-up.
 Typically used for shopping (“buy from our sponsors”), travel, etc.

Summary
Key Messages
 Computational advertising = A principled way to find the "best match“
between a given user in a given context and a suitable advertisement.
 Two main types of online advertising are graphical and textual
advertising.
 Sponsored search is the main channel for textual advertising on the web
 Advertising is a form of information.
 Adding ads to a context is similar to the integration problem of other types of
information
 Finding the “best ad” is a type of information retrieval problem with multiple,
possibly contradictory utility functions.
 New application domains and new techniques are emerging every day
 Good area for research + new businesses.

Many active research areas & open
problems
 query understanding
 content matching
 sentiment analysis
 online modeling
 massive optimization
 text summarization
 named entity extraction
 computer-human interaction
 economics of ads

References
 Computational Advertising course @ Stanford:
http://www.stanford.edu/class/msande239/
 Internet Advertising and the Generalized Second-Price
 Auction: Selling Billions of Dollars Worth of Keywords,
 Edelman, Ostrovsky and Schwartz
 From query based Information Retrieval to context driven
Information Supply, Andrei Broder
 Just in time contextual advertising, Anagnostopoulos et al.
 Internet Advertising and Optimal Auction Design, Schwarz

Computational Advertising

More Related Content

What's hot (20)

Viewers also liked (12)

Similar to Computational Advertising (20)

Recently uploaded (20)

Computational Advertising