SOCIAL MEDIA & MOBILE STRATEGIES FOR TRAVEL
San Francisco, March 2014
1
Barbara Pezzi
THIS SESSION
GET TO GRIPS WITH ATTRIBUTION MODELLING
A FEW STATS
IT’S COMPLICATED
IT’S COMPLICATED
IT’S COMPLICATED
http://services.google.com/fh/files/misc/multiscreenworld_final.pdf
IT’S COMPLICATED
http://services.google.com/fh/files/misc/multiscreenworld_final.pdf
IT’S COMPLICATED
IT’S COMPLICATED
IT’S COMPLICATED
IT’S COMPLICATED
IT’S COMPLICATED
HAVE YOU FIXED EVERYTHING ELSE?
HAVE YOU FIXED EVERYTHING ELSE?
DO YOU NEED IT?
DO YOU NEED IT? • There are limits on what can be tracked. If you can’t measure it properly, just say so and focus on what you can.
• Last click may not be perfect, but it is somewhat consistent and accurate and not necessarily enhanced by replacing it with a sequential attribution model where people assign gut values (or political values) to the steps.
• You need to account for payment models (affiliates, ppc, etc..)
• Is your organisation ready for it?
DIFFERENT CHANNELS FOR DIFFERENT GOALS
http://www.thinkwithgoogle.com/tools/customer-journey-to-online-purchase.html
THERE ARE OTHER FACTORS
OPTIONS • Considering the various funnel stages, multi-device browsing and off/on-line purchasing, media mix modelling (MMM) would be the more appropriate solution. It is however complex and expensive and only warranted for large advertisers.
• A cheaper alternative is to test campaigns introducing one channel at the time and measuring its impact.
• Look at channel sequencing and establish patterns (ex is Tripadvisor always at the start?) and impact. It’ll be more useful that guessing the % of revenue depending on path position.
OPTIONS
• Attribution analysis works best for high frequency purchases (ex retail/OTAs/large chains) and large advertisers. For smaller businesses, visitor segmentation can be sufficient.
• If you have to use an attribution model, apply segmentation to your analysis: ex new vs repeat buyers, subscribers vs non, etc…
• If using GA for attribution analysis, spend the time to create your own channel grouping
MULTI-CHANNEL FUNNELS & ATTRIBUTION IN GA
CREATE YOUR CHANNEL GROUPING
CREATE YOUR CHANNEL GROUPING
CREATE YOUR CHANNEL GROUPING
CREATE YOUR CHANNEL GROUPING
CHOOSE YOUR SETTINGS
CHOOSE YOUR SETTINGS
ASSISTED CONVERSIONS REPORT
TOP CONVERSION PATHS REPORT
ATTRIBUTION IN GA
ATTRIBUTION IN GA
ATTRIBUTION IN GA
https://www.google.com/analytics/gallery/#posts/search/%3F_.type%3DATTRIBUTION_MODEL%26_.start%3D0/
ATTRIBUTION IN GA
ALWAYS SEGMENT
ALWAYS SEGMENT
ALWAYS SEGMENT
ALWAYS SEGMENT
THE OTHER OPTION: VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
VISITOR SEGMENTS
Visit vs Visitor: $34,792 vs $125,131
code.google.com/apis/analytics/docs/tracking/gaTrackingCustomVariables.html
VISITOR SEGMENTS: CUSTOM VARIABLES
IN CONCLUSION:
• Get the basics right first: optimise all campaigns, ensure everything is correctly tracked, review conversion obstacles (high bounce landing pages), etc..
• Segment, segment, segment. • Before investing in a proprietary solution, do the math. The solution has to add value, not simply be an academic exercise.
• If you are going to invest time/funds in attribution analysis, ensure that internally everyone is on board and willing to re-allocate budgets.
• Acting on faulty models is worse than taking no action at all.
“"Not everything that counts can be counted, and not everything that can be counted counts.“
(Albert Einstein)