Consolidated deck across HeroCon London '17, Superweek '18, SMX Munich '18 & DA Summit Greece '18 on the issues with marketing attribution with a recommendation alternative approach to use.
2. • Australian living in London
• Worked in Digital Analytics for over 10 years
• Founded L3 Analytics in 2010
• London based Digital Analytics Consultancy
• GA & GTM Certified Partner
• Merged with AEP Convert to form LeapThree in 2016
• Support with Google Analytics and Adobe Analytics
• Services from Set-up to CRO, Personalisation and Data Strategy
• Work with clients of all sizes and across all sectors
• Founder of MeasureCamp
• Co-founder of MeasureBowling
QUICK INTRODUCTION TO ME
12. “How is my <insert marketing channel> performing?”
“How should I split revenue between my marketing sources?”
“What is the right attribution model for me?”
THE QUESTIONS I GET ASKED IN WORKSHOPS
13. “How do I optimise the allocation of future marketing spend”
“How can I best spend my marketing budget
to make the most money”
THE QUESTIONS I SHOULD BE ASKED
Photo Credit: HikingArtist.com
via Compfight cc
14. THE BUSINESS PROBLEM THEN…
IS NOT
How to decide the right attribution tool or
model for your business
BUT
How to optimise your business marketing
spend & activity to maximise your future ROI
20. • Last click attribution model
• The player that scored the goal gets all the credit for the goal
• First click attribution model
• The player that started the play leading to the goal gets all the credit
WHO GETS THE CREDIT?
22. • Last click attribution model
• The player that scored the goal gets all the credit for the goal
• First click attribution model
• The player that started the play leading to the goal gets all the credit
• Weighted attribution models
• All players involved in the play leading to the goal get part of the credit
• Credit depends on the formula you have decided to use
WHO GETS THE CREDIT?
24. DATA DRIVEN ATTRIBUTION
Based on the data across multiple football games, how much did each
player contribute to plays where goals were scored and not scored
25. • Last click attribution model
• The player that scored the goal gets all the credit for the goal
• First click attribution model
• The player that started the play leading to the goal gets all the credit
• Weighted attribution models
• All players involved in the play leading to the goal get part of the credit
• Credit depends on the formula you have decided to use
• Data driven attribution models
• All players involved in all plays which did/didn’t lead to goals across the (very
long) season get credit for their contribution to goals scored
• E.g. formula to attribute the credit is calculated statistically
WHO GETS THE CREDIT?
26. • Last click models are flawed as gives credit to a single player only,
ignoring all other players
• First click models are flawed as gives credit to a single player only,
ignoring all other players
• Weighted attribution models are all flawed as one set of logic cannot
reflect the contribution of players to all goals scored
Note: invalidated simply by being able to create a model to arrive at
any desired answer you want…
• Is all this solved with the use of data driven attribution models??
SO WHICH ARE THE “BETTER” ATTRIBUTION MODELS?
28. 1. Customer journey mapping doesn’t include all touchpoints
2. Attribution models are based on allocating 100% of revenue /
conversions to all (known) touchpoints
3. Attribution models depend on historical data
THE THREE KEY FLAWS WITH ATTRIBUTION MODELS
1
2
3
29. According to data driven attribution tools
It is critical to record the full customer journey
THIS IS NOT POSSIBLE!!
REQUIREMENT FOR DATA DRIVEN ATTRIBUTION SUCCESS 1
30. RETURNING TO OUR FOOTBALL ANALOGY
Conversions
1
Remember how there
were only three players
in the story?
31. Home (computer)Work (or smart
phone, tablet, etc)
EXTENDING THE ANALOGY FOR MULTIPLE DEVICES
Conversions
1
32. Play started on
the other side of
the pitch & these
players deserve
credit too
Home (computer)Work (or smart
phone, tablet, etc)
THE JOURNEY CAN START ON A DIFFERENT DEVICE
Conversions
1
33. Ball forced out by
defender & other players
provided alternative
attacking options – also
deserve credit
Conversions
SOME TOUCH POINTS AREN’T EVEN ONLINE
Home (computer)Work (or smart
phone, tablet, etc)
1
34. DATA IS MISSING TOUCHPOINTS 1
If you can’t see all the players on the pitch, how can you accurately
calculate their contribution to goals scored??
35. • Very simplistic scenario
• High proportion of customers for a retailer research pre purchase
• 80% of eventual customers will have research visits
• 75% of these researchers do this on their lunch breaks at work (without logging
in) before purchasing at home
• The data will say that 80% of customers purchased on their first visit (20% no
research + 75% x 80% do research)
• The business strategy based on this data would be incorrect!!
• Not having the full customer journey breaks attribution
THE IMPACT OF INCOMPLETE CUSTOMER JOURNEYS 1
36. • A common response / point is raised
• “That is the same with all Digital Analytics data…”
• Within Digital Analytics tools, the sample reflects the population
• This data is not accurate
• Desktop conversion rate is 4.2% and Mobile conversion rate is 2.3%
• 63,036 sessions were from UK and 12,589 were from the US
• The Bounce Rate for visitors landing on the homepage is 36%
• But actions based on this data are the right actions
• (depending on your skill levels)
QUICK NOTE – NO IT IS NOT THE SAME!! 1
37. • Two scenarios taken from Gary Angel blog post - bit.ly/1tSBM8s
• First company is a motors dealership
• Does some research into customers and discover a website that is quite popular
with 20% of customers viewing pre purchase
• Due to this, start advertising with display ads on this website
• Great results, 20% of sales occur after viewing a display ad on this website
WHAT IS THE IMPACT OF MARKETING ACTIVITY? 2
38. • Second company offers annual subscriptions
• Very high retention rates, typically 85% renew subscription
• Company starts new email programme for existing subscribers
• Most customers open this email with no other marketing touchpoints
• 85% of customers who receive the email purchase a new subscription
• How much credit should these two marketing campaigns get?
WHAT IS THE IMPACT OF MARKETING ACTIVITY? 2
39. • Attribution models need to (or at least attempt to) include all touch
points prior to a purchase
• In these scenarios, the marketing campaigns get a lot of credit
• Marketing activity was correlated against the sales
• But it didn’t cause the sales
HOW MUCH IMPACT DID THE CAMPAIGNS CAUSE? 2
40. Correlation vs causation
Incremental conversions vs total conversions
Customers via marketing vs customer loyalty
HOW MUCH CREDIT SHOULD THESE CAMPAIGNS GET? 2
41. • Attribution models are forecasts built on historical data and statistical
modelling - logistic regression
• But situations change…
• If you can’t adjust the model based on known changes, the output is
going to be flawed
ATTRIBUTION OUTPUT IS BASED ON HISTORICAL DATA 3
42. CHANGES THAT COULD IMPACT PERFORMANCE
New Marketing
Campaign
New Product
Range
Competitor
Strategies
External FactorsNew Social
Media Platform
Change
Marketing
Campaign
3
43. 1. Customer journey mapping doesn’t include all touchpoints
• The maths can’t work if working on incomplete data sets
2. Attribution models are based on allocating 100% of revenue /
conversions to all (known) touchpoints
• We need to made decisions based on what spend caused what actions
• Some revenue / conversions will be generated without marketing, these should
not be included within calculations
3. Attribution models depend on historical data
• We need models that predict the future, not that explain the past
THE THREE KEY FLAWS WITH ATTRIBUTION MODELS
45. “How do I optimise the allocation of future marketing investments”
“How can I best spend my marketing budget
to make the most money”
BACK TO THE REAL QUESTIONS
Photo Credit: HikingArtist.com
via Compfight cc
46. …for the insights they can provide
ATTRIBUTION MODELLING TOOLS ARE STILL USEFUL
47. Evaluate performance of each campaign against the purpose for that
campaign – instead of trying to match to the end conversion
APPROACH FOR EVALUATING CAMPAIGN PERFORMANCE
1. Define what success means for
each campaign
2. Define a metric to represent
this success
3. Attach a financial value to the
success metric
4. Measure performance of the
campaign using success metric
5. Calculate the ROI of the
marketing campaign
48. • My advice on how to do this…
Forget data, analytics tools, machine learning, AI and even Excel
• Let’s talk about people…
• If someone is exposed to marketing campaign X, what is their desired
reaction?
SOURCE OF INSIGHTS FOR DEFINING MARKETING SUCCESS
49. • Visitor views information
about money transfers
• But if interested (now or in
future), will show this by
performing a calculation
• As that tells them the fees
and exchange rate
• Can you now measure and
evaluate against this
success action??
MARKETING CAMPAIGN FOR MONEY TRANSFER
50. • It is already in place for football players
• Why not for marketing campaigns??
DON’T SAY THIS CAN’T BE DONE...
51. • Similar to testing website elements, test marketing campaigns
• How to test campaigns
• Different campaigns in different geographical regions
• Hold out tests
TEST CAMPAIGNS TO CALCULATE TRUE ROI
53. • Similar to testing website elements, test marketing campaigns
• How to test campaigns
• Different campaigns in different geographical regions
• Hold out tests
• Switch off/on keywords
• Pick similar trending products & promote half
• Learn what impact of campaign really is
• Then use these insight to create your optimal marketing plan
TEST CAMPAIGNS TO CALCULATE TRUE ROI
54. • Causal models can be the most accurate approach for evaluating the
impact of online and offline campaigns
• They focus only on the two key factors, money spent and incremental
revenue received
• Ignoring the chaos in between
INCORPORATE CAUSAL AND/OR MEDIA MIX MODELS
SPEND REVENUE
55. • I am starting to visualise a tool (no, am not building this)
• And maybe this is what attribution tools already do
• Input is historical impact values for marketing campaigns
• Calculating based on interim success metrics appropriate to each campaign
• Or potentially using data driven solutions to provide these values
• Users are able to adjust variables due to known changing factors
• The tool forecasts (incremental) sales based on marketing spend
• The forecast is actually of customer lifetime value (profitability) by campaign
• Based on the output, users can adjust their marketing spend/activity
• Learning and improving as time goes on
THE DESIRED MARKETING OPTIMISATION TOOL
56. • Cons
• It is really hard to do and means you have to do hard thinking
• No buzzwords and shiny toys
• Pros
• You have control over your marketing
• You have a proper understanding of what does and doesn’t impact performance
This tool would be able to optimise future marketing spend & activity
in a way that truly maximises your ROI for the future…
WHAT WOULD THIS GIVE YOU
57. I can be found at
• peter.oneill@leapthree.com
• @peter_oneill
• www.linkedin.com/in/peteroneill
THANK YOU (AND QUESTIONS)