Bhanu C Vattikonda Vacha dave Saikat Guha Alex C Snoeren Organic results Sponsored results Query 2 Huge and growing industry 20 3 Source IABPwC Internet Ad Revenue reports ID: 447167
Download Presentation The PPT/PDF document "Empirical Analysis of Search Advertising..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Empirical Analysis of Search Advertising Strategies
Bhanu C. Vattikonda, Vacha dave, Saikat Guha, Alex C. SnoerenSlide2
Organic
results
Sponsored
results
Query
2Slide3
Huge and growing industry
20%
3Source:
IAB/PwC Internet Ad Revenue reportsSlide4
Organic
SponsoredSlide5
Brand Ad
Competitor Ad
Are common advertising strategies profitable?Slide6
Metrics for campaign analysis
Click-through-rate (CTR)Measures ability to attract users but not quality of usersCost per acquisition (CPA) or cost per conversion:Does not capture profitability (e.g., in case of organic results)
Profit-per-impressionRequires access to sensitive financial information from advertisersSlide7
User Flow
7
Search impression
ConversionSlide8
Search engine
dataUser click information Cost of a particular advertising strategyNumber of conversions that an advertiser receives
8
How much is a conversion worth to the advertiser?Slide9
Value of a conversion
Advertisers specify bid value for an adIndicates how much they are willing to spendWe assume that advertisers are rational and want overall advertising to be profitable
Value of a converison,
9
Slide10
Net acquisition benefit (NAB)
We introduce a new metric--NAB--captures profitability of a particular strategyAdvertiser spends
on a particular strategy Leads to
conversions is value of a conversion
Profitability
, normalized for impressions
10Slide11
Computing NAB
Computing NAB for a strategy requires:
Conversions
Cost of the ad strategy
Value of a conversion (
)
11Slide12
Comparing ad strategies
Compute NAB for each advertising strategy Compare NAB to see which ad strategy is more profitable If
then strategy
is better than
12Slide13
Common ad strategies we analyze
Cannibalization Advertising for own brand namePoaching Advertising for a competitors brand name Mobile ad extensions Using ad extensions (e.g., call extension) on mobile devicesSlide14
Organic
SponsoredSlide15
Search engine
datasetMonth long click log data from a search engine Billions of clicksMillions of ads Conversions for both ad and organic clicksSeveral million dollars of ad spend
Bid that the advertiser placed for each ad click 15Slide16
Advertising for own brand name
Advertise for brand name NAB is computed over search impressions where user searches for brand name and ad is shown Do not advertise for brand nameNAB is computed over search impressions where user searches for brand name and ad is NOT shown
16Slide17
Advertising for own brand name
Conversions
Total number of conversions through ads and organic results
Number of conversions through organic results
Yes
, advertise
No, do not advertise
Users search for brand and advertiser ad is shown
Users search for brand and advertiser ad is not shown
Impressions
Cost
Cost of campaign
0
Target cost
--- but irrelevant
17Slide18
Advertising on own brand name
18Slide19
Incremental benefit is limited
32% advertisers net loss
19
Why do advertisers advertise for their brand name?Slide20
Click-through-rate is misleading
Click-through-rate not correlated to campaign performance 20
CTRSlide21
Summary
We introduce NAB which can be used measure profitability of an ad campaign without access to profit information from advertisersAdvertisers often target the wrong users Common strategy of targeting own brand name can be a net loss Commonly used metrics can be misleading
21Slide22
Thank you
22Slide23
Measuring campaign performance
Value of conversion for advertiser is unknownAdvertiser diversitySix orders of magnitude difference on total ad spendAd strategy comparison has to be advertiser centricQuery level analysis leads to poor statistical significance
When comparing ad strategies we should aggregate similar queries