improving sem results through multi-channel conversion attribution - kenshoo webinar

Kenshoo: Proprietary and Confidential Information |

Improving SEM Results through Multi-Channel Conversion Attribution

November 2012

Sponsored by

Kenshoo: Proprietary and Confidential Information | 2

Webinar information

Audio instructions• Listen via computer speakers or dial into the conference call using numbers in email

from [email protected]

Still can’t hear audio? Tell us please!

Webinar Information


Presenters

[Daniel Morgan]Head of SearchAccor

[Ariel Rosenstein]Director of MarketingResearchKenshoo

[Yossi Synett]Research AnalyticsTeam Lead Kenshoo


Agenda

Importance of Multi-Channel Approach

History of Attribution

Reinvestment = Attribution

Types of Attribution Models

Viewing Data Via Different Models

Accor Case Study

Choosing the Right Model


About Kenshoo

Controlled Automation Algorithmic Optimization

Holistic Attribution Infrastructure Integration

✔ ✔ ✔

✔

✔


Kenshoo.com/ForresterWave

The Forrester Wave is copyrighted by Forrester Research, Inc. Forrester and Forrester Wave are trademarks of Forrester Research, Inc. The Forrester Wave is a graphical representation of Forrester's call on a market and is plotted using a detailed spreadsheet with exposed scores, weightings, and comments. Forrester does not endorse any vendor, product, or service depicted in the Forrester Wave. Information is based on best available resources. Opinions reflect judgment at the time and are subject to change.

Forrester Wave™: Bid Management Software Providers, Q4 '12


Importance ofMulti-Channel Approach


More important now, more than ever

1996 2011245 millions U.S. internet users Average usage: 27 hours/month

Top rated TV show Sunday Night Football 10.7% of household per average

week

92% of surfers use a search engine

Average Facebook user 15.5 hours/month on site

Video game usage matures Average gamer is 32 years old 2 out of 5 gamers are female

20 millions U.S. internet users Average usage: 30 minutes/month

Top rated TV show ER 35% of household per average week

Source: OnlineUniversity.net


Increased mobile activity but results differ by device

Kenshoo Aggregate U.S. Client Metrics Q3 2012

TABLETPHONECOMPUTER

CLICKS 79% 11% 10%

AD SPEND 87% 6% 7%

CPC $0.48 $0.23 $0.31

CONVERSIONS 91.72% .061% 8.67%


Search engines performance varies

QUARTERLY U.S. SEARCH ENGINEAVERAGE CLICK-THROUGH RATES

Q2 2011 Q3 2011 Q4 2011 Q1 2012 Q2 2012 Q3 20120.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

1.80%

1.03%1.08% 1.10%

0.89%

1.01%

1.25%

1.09%1.18%

1.49% 1.49%1.56%

1.61%

Google Yahoo Bing Network

Source: Kenshoo Global Search Advertising Trends, © 2012, Kenshoo, Inc.


Search engines’ performance varies

QUARTERLY U.S. SEARCH ADVERTISINGRETURN ON AD SPEND BY SEARCH ENGINE

Q2 2011 Q3 2011 Q4 2011 Q1 2012 Q2 2012 Q3 2012

1.00

1.14

0.991.04 1.05

0.96

1.271.20

1.241.30 1.29

1.23

Google Yahoo Bing Network

Source: Kenshoo Global Search Advertising Trends, © 2012, Kenshoo, Inc.


Facebook makes customers reachableat all times of day

CONVERSIONS BY TIME OF DAY (FACEBOOK)

12:0

0 AM

1:00

AM

2:00

AM

3:00

AM

4:00

AM

5:00

AM

6:00

AM

7:00

AM

8:00

AM

9:00

AM

10:0

0 AM

11:0

0 AM

12:0

0 PM

1:00

PM

2:00

PM

3:00

PM

4:00

PM

5:00

PM

6:00

PM

7:00

PM

8:00

PM

9:00

PM

10:0

0 PM

11:0

0 PM

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

8.0%

Source: Kenshoo Social DSP partners


History of Conversion Attribution


The early days of PPC

2003 Yahoo!buys Overture

a company

2002 AdWordsgoes to PPC Model


The assumption: One Conversion = One Click


Keyword 1

Green T-ShirtGreat PricesT-shirt.com

Blue T-ShirtGreat PricesT-shirt.com

Keyword 2

Red T-ShirtXL Size

AvailableT-shirt.com

Keyword 3

Last-Click Attribution Model:• Solves conversion duplication issue• Assumes last click is most important

Discovering the need for conversion attribution


Reinvestment = Conversion Attribution


Reinvestment based on data

Mutli-Source Data Reconciliation

Mutli-Channel Data Connections

Attribution Model

Keyword Selection

Ad Copy & Testing

Campaign Structure

Sites and Content

Devices

Geographies

Algorithm Selection

Algorithm Configuration

Forecasting & Scenarios


GroovyTravel.com investment strategy(with Last-Click attribution)

Investment£600

Revenue

£500

cheap flights to Mexico

Revenue£1000

Investment

£200

GroovyTravel.com

£800.00

£1,500.

00

ROAS = 0.83 ROAS = 5

Total ROAS = 1.88


GroovyTravel.com investment strategy(with Last-Click attribution)

Investment£200

Revenue

£180


Revenue£800

Intended Investment

£600

Actual Investment

£400

GroovyTravel.com

£600.00

£980.00

ROAS = 0.9 ROAS = 2

Total ROAS = 1.63


GroovyTravel.com path to conversion

£450cheap flights to Mexico

£100GroovyTravel.com

£900GroovyTravel.com


£50cheap

flights to Mexico

GroovyTravel.com


GroovyTravel.com investment strategy(with Distribute-Evenly attribution)

Revenue£925

Investment

£600


Revenue£575

Investment

£200

GroovyTravel.com

£800.00

£1,500.

00

ROAS = 1.54 ROAS = 2.88

Total ROAS = 1.88


GroovyTravel.com investment strategy (with Distribute-Evenly attribution)

Revenue£1100

Investment

£800


Revenue£700

Investment

£300

GroovyTravel.com

£1,100.

00

£1,800.

00

ROAS = 1.38 ROAS = 2.33

Total ROAS = 1.64


Types of Attribution Models


Keyword 1



Keyword 2

Red T-ShirtXL Size


Keyword 3

Industry Default Model

Attribution Models: Last-Click


Attribution Models: Linear

60%25%10%5%

Attribution Models

Linear

U-Shaped

Distribute-Evenly

Single Click Only (First or Last)

Keyword 1



Keyword 2

Red T-ShirtXL Size


Keyword 3


Attribution Models: Non-Linear

40%10%10%40%

Attribution Models

Linear

U-Shaped

Distribute-Evenly


Keyword 1



Keyword 2

Red T-ShirtXL Size


Keyword 3


Attribution Models: Distribute-Evenly

25%25%25%25%

Attribution Models

Linear

U-Shaped

Distribute-Evenly


Keyword 1



Keyword 2

Red T-ShirtXL Size


Keyword 3


Attribution Models: Single Click Only(First or Last)

100%0%0%0%

Attribution Models

Linear

U-Shaped

Distribute-Evenly


Keyword 1



Keyword 2

Red T-ShirtXL Size


Keyword 3


Last Click Attribution Undervalues FB Performance

Increase in Facebook ROI for RetailAdvertiser Based on Revenue Attribution

• Last Touch

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch

20% 40% 60% 80% 100%




• Last Touch

• Prefer Last

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch

20% 40% 60% 80% 100%




• Last Touch

• Prefer Last

• U-Shaped

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch

20% 40% 60% 80% 100%




• Last Touch

• Prefer Last

• U-Shaped

• Distribute Evenly

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch

20% 40% 60% 80% 100%




• Last Touch

• Prefer Last

• U-Shaped


• Prefer First

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch

20% 40% 60% 80% 100%




20% 40% 60% 80% 100%

• Last Touch

• Prefer Last

• U-Shaped


• Prefer First

• First Only

FirstOnly

PreferFirst

DistributeEvenly

U-Shaped PreferLast

LastTouch


Case Study: Major US RetailerInterplay between search & social channels

20% of all conversion paths showed ad clicks from more than one channel (search and social)• Over 1/3 had brand queries in the path to conversion• Consumers discover brand on FB and then search for it by name

Logarithmic Distribution of Keywords in Multi-Channel Conversion Paths


Viewing Data via Different Models


Considering multiple channels usingKenshoo Attribution Model Simulator

• Leading US pet supply cataloger using last-click attribution

• Brought data from various channels (email, CSE, display) into Kenshoo

• Using Kenshoo Attribution Model Simulator, saw best and worst case ROI for each channel under different models

• Determined which model was best fit by seeing how different channels would benefit or lose from changing models

Total

1.90 2.00 2.10 2.20 2.30 2.40 2.50

2.27

2.12

2.47

2.24

2.31

2.29

ROI – Display

U Shape Prefer First Prefer Last

First Only Last Only Divide Equal


Case Study: Moving away from Last-Click

U.S. national service center chain increases ROI by changing attribution model following a multi-model attribution analysis

Key Lessons

Different regions have different behaviors

Last-Click model was holding back potential new to file customers

Prefer last model better allocated money

Kenshoo: Proprietary and Confidential Information |

82% Increase in Revenues Year-on-Year using Kenshoo’s U-shaped Model (Generic Keywords Jan-Mar 2011 Vs 2012)

41

About Accor

Present in 92 countries, with more than 3,500 hotels and 440,000 rooms, Accor’s brands offer hotel stays tailored to the specific needs of each business and leisure customer.

42

Introduction So Sofitel Bangkok

• Accor Group began working with Kenshoo in April 2011

• Accor run paid search campaigns across all 13 brands in over 80 countries on the Kenshoo Platform

• In combination with Kenshoo, Accor use 2 other tracking softwares – TARS & Xiti

• Campaigns are optimised using Kenshoo data but internal reports continue to use TARS

43

1. Accor’s Internal Tracking is based on Last Click Attribution

…Yet Search is so much more than an isolated

media channel

The Rest of Online Media

Above-the-Line Media

Search Engines

The Complex Challenge Ibis Leiden Netherlands

44

Search Engine

• The user clicks on an Accor Paid Search Ad

On-Site Search

• They arrive on the Accorhotels.com site and search for a hotel – No Reservation

Reservation on Sister

site

• They then access the Brand website directly AND Reserve No Internal Sales

Attribution

2. Cross-site conversions were untracked

The Complex Challenge Ibis Leiden Netherlands

45

Accor Turns to Kenshoo for a Solution...

What we found • Cross site conversion - SEA is directly affecting the direct access of all Accor’s websites• Maps , Affiliation, Email -The more we spend on SEA the higher our revenues

Direct SEA Sales; 66%

Cross-site Conver-sions at-

tributed as "Direct En-try" inter-nally; 18%

Other Search Cam-

paigns; 8%

Partners/Affil; 5%Maps; 4%

… Kenshoo showed that SEA is strongly linked

with all other channels

Total SEA Kenshoo Sales Broken Down by Accor’s Internal Tracking Attribution

46

No more conspiracy theories…

What we found:• Generic affects brand – 10% of Brand sales started searches on generics• Brand affects generic – Surprisingly 20% of brand searches generated generic clicks• Brand affects other brand – Multiple Accor Brand Searches e.g. ibis then etap,then accor• Generic affects generic – From broad searches to long tail e.g. “hotel paris” TO hotel “paris

with pool, excepts cats”

40%

40%

6.7%

6.7%

6.7%

U-Shaped Model - attributes greater value to the first and last clicks in the path with both the starting and the closing keywords getting the same credit regardless of the length of the conversion path.

Attribution Value

…Each Keyword’s Real Value Can Be Viewed & Acted Upon Automatically

47

Optimisation becomes easier…

…despite the complexity of the online purchasing funnel

But the next challenge is to see if those campaigns that don’t covert have a positive affect outside the cookie period?

Can Search Act as an Awareness tool?

No need to spend in hope - Accor can optimise campaigns by pausing any keywords or campaigns that do not contribute to any conversion.

48

Developing an online relationship is key…

Display,Offline,

PR

SEM,Natural Search,

Partners/Affil

Social,Natural Search

Email

Success still heavily relies upon generating awareness, which at the moment Search does not do well!

…Accor still see strong evidence that the old funnel still exists

48

49

The Results

Accor’s Search Team can justify higher spend on paid search

Comparing the period from 1st January 2011 to 31st March 2011 when Kenshoo tracking had not been implemented, to the same period in the year 2012:

• 82% increase in revenues • Only 36% increase in spend• 14% increase in revenues across all other channels • 33% increase from brand terms

Pullman Bangkok


How to choose the right model


Attribution Model Simulation: Inform your investment decisions

Track activity across multiple channels

Gain target and keyword-level insight including contribution in the conversion path

Empower strategic decisions for budget allocation, bid management, and keyword selection


Get insights

• Any single model will only show a single viewpoint, compare different models for the most comprehensive view

• Lower costs by pruning ads/keywords that have no touch-points in any conversion attribution model

• Expand assisting keywords to find new traffic generators and widen you market reach• Don’t miss the forest for the trees, apply real world categories and use your own

business logic to make attribution model decisions


Download the full whitepaper

Kenshoo.com/ConversionAttributionWhitepaper

http://www.kenshoo.com/ConversionAttributionWhitepaper/

http://www.kenshoo.com/ConversionAttributionWhitepaper/


The Yahoo! Bing Network is a must-buy

Excerpt from Kenshoo GLOBAL SEARCH ADVERTISING TRENDS, October 2012 Edition

“…with YBN consistently delivering better ROAS… the new imperative for marketers is, at the very least, to be active on YBN, if not make it the first place resources are allocated and maximized.


Thank you

[email protected]

improving sem results through multi-channel conversion attribution - kenshoo webinar

Technology