dmexco: The State of Marketing Attribution
Transcript of dmexco: The State of Marketing Attribution
The State of Marketing Attribution
Shane Murphy
@adroll
@shanemurfy
Agenda
1 Attribution the theory
2 State of Marketing Attribution Report Highlights
3 The Case for Incrementality
4 How to get started
About AdRoll
35k Customers of all sizes globally
10+ Years of growing online businesses
True Business Impact Made Easy
Online - Action Offline - Desire
When the internet started life was easy
● See athlete wearing shoes
● See tv ad for shoes
● Try shoes on in store
● Search for shoes online
● Click on ad
● Purchase shoes
Online - ActionOnline - Desire
● See ad with celebrity wearing shoes on
● See friend bought shoes on Facebook
● Go to nike.com and check sizes
● See retargeting ad for shoes
● Search for Nike.com
● Click on branded search ad
● Complete purchase
Now it’s far more complex
We started with rules rules rules
Last Click - everyone’s first model
Last Click - everyone’s first modelFirst Click - the mirror image
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equally
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyPositional - location, location, location
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyTime Decay - short term memory
80%+ of Marketers use a Rule based Model
50%+ use a click based model
But it’s not that simple
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyLack of correlation between clicks and conversion
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyLack of correlation between clicks and conversion
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyLast Click models pull investment from upper funnel demand generating channels
AdRoll internal campaign data
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyAnd ignore the multiple other steps that help drive a conversion
AdRoll internal campaign data
% Revenue driven with conversion path +1 step
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyAnd to make it worse
The top quartile of clickers are5.5x more expensive than less
clicky users
“In order to get more click conversions, you need
to drive more clicks. In order to drive more clicks,
you may be optimizing away from some of the
highest converting inventory in the campaign.”
- James Moore - CRO simpli.fi
“Rule based or last/first touch approaches are remedial approaches that falsely boost campaign and channel performance and do a poor job of measuring cross-channel impact”
- Forrester
We can not truly help our customers unless we both
think differently about attribution
Key FindingsState of Marketing Attribution
N=987 respondents: 74% in house marketers, 26% agencies
Regions covered: North America, Europe, Japan and Australia
www.adroll.com/dmexco2017
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyAbout the research
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyLess than half European marketers run on all campaigns
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyClick based models still most used
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyTechnology, data structure, knowledge and time are key barriers to adoption
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyMost people think First and Last Click models are effective
Last Click - everyone’s first modelFirst Click - the mirror imageLinear - touchpoints matter equallyOver 50% of marketers use Google for their Attribution modelling...
Marketers are waking up but lots of work to do
AttributionThe Case for Incrementality
There is no ideal way to do attribution
How you optimise campaigns is directly linked to likely incremental impact on your business
Click
Based
Conversion
Based
Incremental
Based
Optimise to
expensive
clickers
Optimise to
high intent
likely
converters
Optimise to
actual uplift but
hard to do
Click CPA Conversion Rate Lift Incremental Revenue
$46 -14.72% -$26,479.00
$102.75 +12.36% +$56,216.00
We took some of our top customers and they agreed to test optimising to incrementality
Click
Based
Increm.
Based
They all saw an increase in conversion rates
+ 15.79% + 18.5% + 36% + 5.62%
Brand A Brand B Brand C Brand D
I know this because I did it in 2014
AttributionOK so how do I get started
You don’t need to run from the start
Last Click Last Touch Add time
decay
Algorithmic
Incrementality
Multi-Touch
Track as much as you can
Pick a partner to work with or get an analyst involved
Questions?
Visit Hall 7 Booth C061Tweet me @shanemurfy