tutorial 10 (computational advertising)

of 17 /17
Computational advertising Kira Radinsky Slides based on material from: Ronny Lempel

Upload: kira

Post on 29-Nov-2014




1 download


Part of the Search Engine course given in the Technion (2011)


Page 1: Tutorial 10 (computational advertising)

Computational advertising

Kira Radinsky

Slides based on material from: Ronny Lempel

Page 2: Tutorial 10 (computational advertising)


Placement in Search Engines

With the increased economic impact of the Web, it is crucialfor businesses to have high Web visibility

With the absolute reliance on search engines for finding online resources, Web visibility translates into attaining high placement in search engines’ results for queries related to the business

With searchers rarely browsing more than two result pages per query, “high placement” means showing up among the first 20 results of the relevant queriesThe first page is much better thoughPreferably “above the fold”… and being #1 would be great, thank you.

Page 3: Tutorial 10 (computational advertising)


Heat Maps

Warm colors indicate higher CTR (click-through rate)

Page 4: Tutorial 10 (computational advertising)


Means of Achieving High Placement

1. Being a near-unanimous top page for a query, having appropriate links and anchor text from the “public”Very difficult in competitive settings

2. Designing “search engine friendly” sites, carefully choosing text, and developing connectivity• Search engine optimization (SEO) firms can help• A thriving business (2005: $643M in the US, $275M in

the UK)• Had their own competition with the phrase “nigritude

ultramarine”3. Spamming (both text and links)

Some “black hat” SEOs also resort to this4. Advertising with the engines, and having your message

shown alongside the algorithmic search resultsSpending money to gain mindshare and visibility

Page 5: Tutorial 10 (computational advertising)


The Advertising Market

• Internet advertising is a $1010 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

Page 6: Tutorial 10 (computational advertising)


The Advertising Market

Page 7: Tutorial 10 (computational advertising)


The Advertising Market

Page 8: Tutorial 10 (computational advertising)

Roles and Goals in Computational Advertising

Roles 1. Ad agency /Advertiser

• a service business dedicated to creating, planning and handling advertising for its clients

• handle overall marketing and branding strategies and sales promotions for its clients.

• typically search engine

2. Publisher

3. User

Goals1. Find the "best match" between a given user in a given context and a suitable


2. Short term monetization and long-term engagement and health of market for publisher and ad agency

3. ROI for advertiser (actual sales, brand recognition, etc.)

Page 9: Tutorial 10 (computational advertising)


Internet Advertising Types:Sponsored Search

textual ads served onsearch results pages,triggered by query termson which advertisers bid

Page 10: Tutorial 10 (computational advertising)


Internet Advertising Types:Contextual Ads (a.k.a. Content Match)

textual ads served on thirdparty (“publisher”) pages,triggered by content of pageon which advertisers bid

Page 11: Tutorial 10 (computational advertising)


Internet Advertising Types:Display Ads

Graphical elements (e.g. banners) ofstandard sizes, served on publisherpages, mainly targeting audiencedemographics• Guaranteed delivery (GD): advertiser

buys a display campaign months in advance, requesting ## impressions to certain demographics in a given time period; ad agency pays fine for under-delivery

• Non-guaranteed delivery (NGD): advertiser bids in real time for impressions, with no guarantees that ads will be shown

Page 12: Tutorial 10 (computational advertising)


Search Advertising Business ModelsCPI – Cost per Impression

• Also known as CPM, cost per 1000 (mille) ad impressions• Advertiser is charged whenever ad is shown• Very popular in the mid-90s (all those banners on top of search result

pages)– Still used, mainly for building brand recognition– However, Display Advertising is much more prevalent today on general

Web pages, with targeted audience segments

• Very simple mechanism– But requires trust between search engine and advertiser

• Risk is all on the advertiser• Not spammable• Highly complex from an algorithmic point of view, with guaranteed

delivery contracts to target demographics in the mix

Page 13: Tutorial 10 (computational advertising)


Search Advertising Business ModelsCPC – Cost per Click

• Advertisers bid for keywords that will trigger the ad’s placement, and are charged only when users click on an ad and are transferred to the advertiser’s site– Actual price is set by the engine according to the auction

mechanism– Most keywords sell rather cheaply. In 2005, for example, 50% of

fees < 25 cents, 80% of fees < 50 cents• The main business model of search engines since the turn of the

century– Transformed the search industry into a profitable business

• Still a relatively simple mechanism• Shared risk/win-win situation – the success of an ad campaign is in

the best interest of all parties (search engine, publisher, advertiser)– On the search page itself, the engine is the publisher

• Trackable by all parties, which seems to overcome the trust issue– However, invalid clicks are an issue

Page 14: Tutorial 10 (computational advertising)


CPC Issue: Invalid Clicks

• Loose definition: superfluous clicks on an ad-impression– Causes increased expenses to the advertiser (and increased revenue

to search engines/publishers)

• Click fraud: clicks being done maliciously, with intent to increase the costs of an advertiser (competitive gain) or the profits of an ad-hosting site (financial gain)– Sometimes by “click-bots” or other types of software

• Search engines don’t charge advertisers for clicks that are determined to be invalid– However, the intent behind a click is not easily determined

• High invalid click rates will threaten the viability of the CPC business model

Page 15: Tutorial 10 (computational advertising)


Click Fraud is a Big Deal

• Independent click-tracking companies claim that in some ad campaigns the percentage of fraudulent clicks was around 30%– That’s certainly not a typical number

• The search engines claim that the percentage is much smaller, but do not report numbers

• The engines take click fraud seriously and have established engineering teams tasked to the problem– The Click Quality team at Google (36 people in mid-2006), the Click-

Through Protection team at Yahoo, and the Click Quality team in adCEnter Microsoft

– Pattern recognition, data mining and human analysis is at work– But advertisers say the engines aren’t doing enough, and need to act

with more transparency• Both Yahoo! And Google have settled class action law-suits in 2006 by

offering tens of millions of dollars in compensation to advertisers who claimed they were charged for fraudulent clicks

Page 16: Tutorial 10 (computational advertising)


Search Advertising Business ModelsCPA – Cost per Action

• Also known as cost per conversion

• Search engine/publisher gets a fee only if the user, upon clicking an ad, completes some transaction at the advertiser’s site– Doesn’t necessarily mean that the user buys something

• Amazon were the among the pioneers of this model by paying referral fees when revenue was generated

• Risk is all on the publisher/search engine

• Complex mechanism that requires the advertiser to share some business data with search engine– But more difficult to spam

Page 17: Tutorial 10 (computational advertising)


Search Advertising Business Models

Popularity Risk bourn by

Simplicity Spammable

CPI Exclusive model in Mid-90s; main model in use today for display advertising

Advertiser Simple No, but requires trust

CPC The most popular model since the turn of the century

Shared by all parties

Still pretty simple


CPA Still small, but on the rise

The engine or publisher

No No, but requires sharing of business data