From Click Consult's Benchmark Search Conference 2017, The Bridgewater Hall, 21st September. Presented by Sabrina Garufi, Agency Development Manager, Google.
8. Two free solutions to tackle this challenge
cross-device intra-channel multi-channel
AdWords Search Attribution: Google Analytics Multi-
Channel Funnels:
9. Two leading Attribution types
Rules-Based Attribution Data-Driven Attribution
Advertisers use pre-defined rules Mathematical models automatically apply
credit
12. DDA calculates the incremental impact on conversion
rate
3%
conversion rate
best tech gifts
1% incremental impact
2%
conversion rate
top rated tablets Google pixel c
top rated tablets Google pixel c
13. DDA is customised to your account and includes all
user journeys
Powered By Data Scope Data Source Scalability
Dynamic algorithms
assign credits to
touch points based
on fractional credit
Takes converting
and non-converting
paths into account
100% based on advertiser’s
own data. Conversions and/
or GA imported goals.
(learns over time)
Auto updated
based on
performance
Fixed, static rules
assign credits to
touchpoints
Takes only converting
paths into account
Advertiser can choose
model but no customization
is available
Must be
manually
revisited
Data-Driven
Rules Based
14. Attribution Model Mobile Mobile Desktop
Last Click Model DEFAULT TODAY 0 (Ignored) 0 (Ignored) 1 Conversion
Data Driven Attribution 0.3 Conversion 0.1 Conversion 0.6 Conversion
Last Ad Click on Converting Device
Desktop
Ad Click
Mobile
Ad Click
Desktop
Conversion
Mobile
Ad Click
This also applies to Cross-Device
22. Proprietary + ConfidentialProprietary + Confidential
Our industry is still in its infancy, and it’s
important for us as a company to grasp
every opportunity that might provide us
with a competitive advantage.
Targeted CPA and Data-driven
Attribution have given us lower CPAs,
and allowed us to compete and thrive
in a competitive and volatile space.
Neil Jones,
Digital Acquisition Manager,
Purple Bricks
“
”
Source: Think with Google blog post
+28%
conversions
23. Clothing retailer De Pauli drives 88% more sales with
DDA & automation
Increase marketing
efficiency
Implement AdWords
Attribution and Smart
Bidding
88% more conversions
-32 % CPA
Goa
l
Approac
h
Results
25. How to achieve success using a new attribution
model?
Switching away from last-click
attribution requires re-thinking
time ranges to analyze data
Changing attribution model alone
does not change performance
Use Appropriate Time Ranges
- 30 days recommended
- Don’t use most recent days
Impact results from acting on the data
- Revisit your bids & budgets
- Consider autobidding
26. How can you act on your new DDA model
Automated Bidding Improvements in generic
and decreases in brand
Adjust your budgets
Test new upper funnel
terms is suggested.
27. 1. In Tools > Conversions, click on the conversion
action for which you’d like to change attribution model
2. Select “Edit Settings” to change default
attribution settings
Tools > Conversions
Tools > Conversions > Conversion Actions
It takes 2 minutes to implement in Adwords
Work with advertiser across the UK , helping them to move away from Last Click
And get the most out of the solutions that Google has for measuring complex customer journeys
Most of you know this already
When digital advertising was taking off 10 years ago the customer journey was much simpler
Customer would search and purchase on the same desktop
And of course there wa no option to use any other model but last click
So by default really - it became the industry standard
Not the world we live in any more
Today the customer journey is more fragmented
90% move across devices
38% move across digital channels
What this means is that they are throwing away the value of 75% of their paid clicks
Imagine the insights and performance they could gain by looking at the whole journey
Add in examples
What this means is that they are throwing away the value of 75% of their paid clicks
Imagine the insights and performance they could gain by looking at the whole journey
Add in examples
Really the point of all this is adjusting your measurement for growth - if you’re using last click you are ignoring the channels that customers first use to discovery your brand
There is no single, solution to attribution. Attribution is not a perfect science yet. Nobody knows exactly how much one click is worth yet. It’s simply a way for advertisers to get better answers on the value of media investments than their current strategy allows
Each methodology, each tool and each test should be designed with the intention of improving performance across the entire customer journey.
From our experience, two strategies tend to emerge - rules based attribution and data-driven attribution
RULES-BASED
Overview:
The first strategy is rules-based modeling. Advertisers using rules-based attribution are all about simplicity and incremental progress.
Instead of relying on elaborate models to assign credit, they build and test individual rules that help to explain their how marketing channels work together.
There are 5 default rules-based models which we’re sure most of you have heard of which you can find across AdWords, Analytics, and DoubleClick: first click, last click, time decay, position-based, and linear (visuals on slide)
Pros & Cons:
These models are often seen as highly subjective and arbitrary, mostly because it’s humans rather than algorithms defining what logic is to be used.
While progress is quick and iterative with these models, they but can be difficult to scale over time - especially as rules grow more and more complex.
Please note that these models are based on correlation and don’t tell us if a channel had to be part of the path in order for the conversion to occur.
DATA-DRIVEN
Overview:
The second strategy is Data-Driven Modeling. With data-driven or algorithmic attribution, advertisers use advanced mathematical models to calculate the impact of each interaction.
Pros & Cons:
Models can take time to develop and implement, which may not be the best fit for customers looking to take action immediately.
Much easier to scale and maintain and are highly dynamic compared to static rules-based models (updated on a weekly basis)
Can be risky if used, especially if data isn’t clean, not x-device
Closer to causality than correlation as they take both converting and non-converting paths into account
[CLICK]
The entire industry is rapidly moving to data-driven attribution:
SmartPath (Kenshoo)
Best Fit Attribution (Adobe)
Visual IQ
Convertro
For Facilitators: This would be a great place to bring in market specific numbers (% of advertisers using each strategy). US numbers below
(Optional) Some US advertiser Statistics (Source: AdRoll, "State of the Industry: A Close Look at Retargeting and the Programmatic Marketer" conducted by Qualtrics, Dec 16, 2014)
58% are using first/last click models
34% are using multi-touch models
5% are using algorithmic/custom models
3% are using ‘other’ undefined modles
Add animation - highlight 2 preferred
Uses converting and non converting paths to decide what a keyword is worth
Tailored to you and your customers behaviour
Not static , constantly updating
Customised to your account and customers
Users all user journeys , not just converting -
Constantly updating
Last two are key differentiators for DDA:
X-device: no other ecosystem can leverage a deterministic graph built off of 7 products with over 1B MAUs
40% signed in UK users - understand how customers click on ads and build this into our attribution model - unique for each advertiser, based on your customers behaviour
Even if you working with another attribution provider - DDA in adwords can still add value in understand cross device
Last click & last device- data driven is multi-touch and multi device
Adoption has been growing exponentially
These advertisers have better insight into what is really growing their businnes
Search queries no longer growing - you need to find unique pockets of value
If you are using last click - you dont
Dont just have to relay on Googles data
•But we also know there’s strong demand to move beyond a last-click attribution model•We know that, because here’s what happened when we launched AdWords Attribution last year (and this is the actual adoption graph)
•Thousands of large advertisers are now using it to optimize their search ads in AdWords
•And if you’re not yet using AdW
ords Attribution, we urge you to start doing so
XMas is coming and with it, the hugh retail events like Black Friday & Cyber monday
Last click this is what we see
DDA we can see that almost 2/3rds of the revenue can be attribution to searches that took place 3-4 weeks before the event
On the day of the sale, of course purchases spike as discounts to kick in
They do the research days in advance to decide what they want, and may have already made the decision days before the event
Still some window of time before this surge in purchases
Perfect opportunity to switch
What is this insight worth to you?
Last two are key differentiators for DDA:
X-device: no other ecosystem can leverage a deterministic graph built off of 7 products with over 1B MAUs
Our DDA model is best-in-class, built off of the client’s own data for converting and non-converting paths
GOALS:
Generating qualified leads at an efficient cost-per-acquisition Understanding the real value of online campaigns and keywords throughout the funnel
METHOD:
Replaced their Adwords Last-Click Attribution Model with the Adwords Data-Driven Attribution Model with a set Target CPA
RESULTS:
28% Increase in conversion rates on generic campaigns
1.7x Improvement in conversion rate on mobile generic campaigns
The additional data also lead to valuable time-savings for the marketing team who are now able to focus on the brand’s future keyword strategy.
Last two are key differentiators for DDA:
X-device: no other ecosystem can leverage a deterministic graph built off of 7 products with over 1B MAUs
Our DDA model is best-in-class, built off of the client’s own data for converting and non-converting paths
First thing is that this is a different way of reporting so you need time to understand the new numbers - 30 days - perfect time before Xmas
Reporting alone does not change performance
So one peice of homework is to go back to your teams
Check if this is implements
And get moving on it over the next week or two so you can use the new model during the christmas season
•That’s why Google Attribution will look at all available interactions across all your channels
•And yes, it will also include impressions - which is particularly relevant for Display
•Remember that “last-click” is the worst possible attribution model for Display advertising: because “last” is bad for Display and “click” is bad for Display
•And over time, we’ll improve Google Attribution even further
•By expanding our coverage to include apps
•And by integrating with DoubleClick products, like DoubleClick Search
•Here’s how simple it is to set up Google Attribution
•You log in - and we automatically identify your Google Analytics and AdWords accounts
•You select the Google Analytics property you want to link to
•You select the right View in that Google Analytics property
•We then automatically link the AdWords accounts which are linked to Google Analytics
•And that’s it: your Google Attribution account is ready to go
A great solution - but not til 2018. Dont let perfect be the enemy of good and take advantage of data driven attribution for search & shopping that you already have available to you