PROSAD: A Bidding Decision Support System for

roughhewnstupidInternet and Web Development

Nov 18, 2013 (3 years and 9 months ago)

93 views

Science
-
to
-
Practice Initiative

PROSAD: A
Bidding

Decision

Support System
for

PR
ofit
O
ptimizing
S
earch Engine
AD
vertising

Bernd Skiera, Nadia Abou Nabout


skiera@wiwi.uni
-
frankfurt.de

abounabout@wiwi.uni
-
frankfurt.de


Search Engine Advertising

2

Availability

of
video

presentation

and additional
exercises

Paper "PROSAD: A Bidding Decision Support System for PRofit Optimizing
Search Engine
Advertising"
was a finalist of
"The
Gary L. Lilien ISMS
-
MSI
Practice
Prize.“ A
video presentation and
the original PowerPoint
slides of
the
presentation are
available at
http://
techtv.mit.edu/videos/18315
-
prosad
.

Instructors can also contact the authors (
skiera@skiera.de

or
abounabout@wiwi.uni
-
frankfurt.de
) for a larger deck of slides and an exercise (including teaching note) that can be
taught and that involves two small data sets to further illustrate the decision support system.

Search Engine Advertising

3

Keyword

Organic search results

2

3

4

5

6

1

Paid search results

Rank

What is search engine advertising (SEA)?

Search Engine Advertising

4

Decision making
after
cooperation

1.
If


keyword profit after acquisition costs > 10


then

increase
bid by 30%

Rules
-
based decision making

2.
If


rank > 5

then

increase bid by 20%

3.
If


keyword profit after acquisition costs < 0



&
number of clicks > 100



&
rank <= 3

then

decrease
bid by 20%

Profit maximization

Search Engine Advertising

5

PROSAD

(
PR
ofit
O
ptimizing
S
earch Engine
AD
vertising)

Transactional

profit

Profit contribution
per conversion

Acquisition costs

per conversion

Number of
conversions

Max!

Search Engine Advertising

6

How does the bid influence transactional profit?

Transactional

profit

Profit contribution
per conversion

Acquisition costs
per conversion

Number of

conversions

Bid

Rank

Number of searches

Quality Score

Clickthrough rate

Conversion rate

1

2

3

4

Decision

variable

Search Engine Advertising

7

Percentage increase in
clickthrough rates

Percentage increase in
prices per click

Conversion rate

Profit contribution per
conversion

'
''
*
k
k k k
k k
Bid =PC CR.
+

 
 
Optimal bid

Optimal bid

1

2

3

4

k
PC
k
CR
'
k

'
k

Search Engine Advertising

8

Learnings from field experiment

ROI for lower budget:

+21%

Profit improvement per keyword per year:

+33.12


LOWER BIDS

Profit improvement potential of PROSAD for
SoQuero and its clients:

2.7


million

Search Engine Advertising

9

Summary


Number of rules
grows
quickly


Likelihood of contradicting
bidding suggestions
high


Choice of specific parameter values in rules
difficult

Rules
-
based
decision making
difficult


Profit function
equals number
of conversions times
profit per
conversion after
acquisition
costs


Estimation of functional relations between


rank and
bid


rank and
clickthrough
rate

Profit optimizing search engine advertising easily feasible


Reduction of
SEA budget
by 38%


Increase in ROI by 21 percentage
points

Results of field experiment support profit optimizing
SEA