Integration and Automation in Practice: CI/CD in Mule Integration and Automat...
Google - measuring search's impact
1. Measuring Search’s Impact
Mark Riseley
Research Manager, Google EMEA
Presented on 3rd November 2011 at
“What works now - Marketing Effectiveness in the Data Age”
Google Confidential and Proprietary 1
12. How to measure the value of search
Surveys Tracking Experiments Modelling
Consumer insights Consumer insights + Prove incremental Influence of search
Attribution of sales impact of search ads and other drivers on
traffic / sales
Google Confidential and Proprietary
13. How to measure the value of search
Surveys Tracking Experiments Modelling
Consumer insights Consumer insights + Prove incremental Influence of search
Attribution of sales impact of search ads and other drivers on
traffic / sales
Shopper surveys Clickstream analysis Pre-post tests Econometrics
Unique phone numbers Regional split tests
Coupons Panel split tests
Check & reserve analytics Cookie split tests
Cross media
optimization studies
Google Confidential and Proprietary
14. POS Recruited Shopper Survey Approach
Objectives
1 Quantify the role of the internet in
buying television equipment in store
2 Identify the differences between those that
do and don’t use the web in the purchase
process
3 Understand purchase process of online
researcher
Methodology
1 357 offline TV purchasers @ Fnac
recruited via contest entry at point of
purchase (in store)
2 Follow up telephone interview then
conducted 10 days after purchase
Google Confidential and Proprietary
Source: FNAC Spain / Google 2010
15. Quantifying online to store activity
52%
WWW
52% of those buying a TV offline in-store at Fnac had
previously used the internet for research
Google Confidential and Proprietary 15
Source: FNAC Spain / Google 2010
16. Store Basket size higher for online researchers
Amount
Spent €701
€526
+ 33%
DID NOT
Those researching online
RESEARCHED
ONLINE
RESEARCH
ONLINE
spend on average 33% more
than those that do not
Google Confidential and Proprietary
Source: FNAC Spain / Google 2010
16
17. Unique phone number tracking methodology
Google AdWords: Product page of ABN Amro ABN Amro Contact Centre
Search for car insurance Display of special hotline Closing of contract or request
product, e.g. telephone number for users for final contract documents
“Autoverzekering” arriving via AdWords
ABN Amro:
Online closing of contract
Google Confidential and Proprietary 17
18. Example Results: ABN Amro Netherlands
Indexed Figures
Car Insurance Contracts through AdWords Credit Card Contracts through AdWords
400% 400%
350% 350%
300% 300%
250% 250%
254%
200% 200%
354%
150% 150%
14%
100% 100%
50% 100% 114%
50% 100%
0% 0%
Online Contact Centre Online + Contact Online Contact Centre Online + Contact
Centre Centre
Source: ABN Amro 2010 Google Confidential and Proprietary 18
19. Research Challenge: Finding Matched Test & Control Groups
Total audience
Test Control
Google Confidential and Proprietary 19
23. Search query data can tease out external effects: category seasonality
Google Confidential and Proprietary
24. Search query data can tease out external effects: product launches
Google Confidential and Proprietary
25. Search query data can tease out external effects: competition
Google Confidential and Proprietary
26. Mapping sales drivers
Business
Online world Offline world Outcome
+ Driver
Online Traditional - Driver
Search Media
Width of line
Store Refurbishment indicates size
of driver
SEM Holiday
Online (Paid Search) Email
Display
Short term
Web traffic Offline Sales Promo days
Online
Sales Competitive Sporting
Media
Rainf Events
Price, Base Sales all
& Seasonality
26
Source: Marketshare / PC City (Spain) Google Confidential and Proprietary 26
27. Econometrics helps optimize media mix
Historical
4,800 spend
Historical
4,800
spend
Weekly Sales (Units)
Weekly Sales (Units)
4,700 4,700
Optimal
4,600 spend
4,500 4,600
Optimal
4,400 spend
4,300 Increase spend in 4,500 Increase spend in
4,200 Paid search
catalogues
4,100 4,400
- 50,000 100,000 150,000 200,000 250,000 300,000 350,000 - 10,000 20,000 30,000 40,000 50,000
Weekly Media Spend € Weekly Media Spend €
4,800
4,800 Historical
Historical
spend
Weekly Sales (Units)
4,700 spend
Weekly Sales (Units)
4,700
4,600
4,600 Optimal
4,500 spend
Optimal 4,500
4,400
spend
4,300
Decrease spend in Increase spend in
4,400
4,200 Press
TV
4,100 4,300
- 100,000 200,000 300,000 400,000 500,000 600,000 - 10,000 20,000 30,000 40,000 50,000 60,000 70,000
Weekly Media Spend € Weekly Media Spend €
27
Source: Marketshare / PC City (Spain) Google Confidential and Proprietary
28. Search query trends can have predictive power
Google Confidential and Proprietary 28
29. Bank of England nowcasting with search data
Google Confidential and Proprietary 29
30. You can do this yourself...
Google Confidential and Proprietary 30
31. You can do this yourself...
Google Confidential and Proprietary 31
32. Conclusions
• Researching search is hard because it’s a pull medium, so
established methods don’t always apply
• Tracking and surveys can generate useful consumer insights around
search and the online to offline effect
• Experiments and econometrics can be used to establish the lift and
true ROI of search activity
• Search query trends can be used to represent external factors such
as seasonality, competition and product launches in models, and
has predictive power
Google Confidential and Proprietary 32