Florian Gramshammer, MD, Impact
The quality of (big) data has for a long time been a crux for CMOs. Instead of creating a true view that positions their brand to make sound decisions, CMOs are creating data Frankenstein monsters by stitching data together from different warehouses, assembling disparate findings, and failing to optimize decisions around the customer journey. In this session, Florian Gramshammer, MD of Impact EMEA, will discuss the importance of examining, cleansing, and accounting for your data carefully in a way that will inform sound decisions about your paid marketing and media spend; how CMOs can become data stewards within their own organizations and lead their teams with confidence; and how to establish whole, pure entities that will promote industry-wide innovation
How consumers use technology and the impacts on their lives
Fix, don't stitch: be a steward of your marketing data
1. FIX, DON’T STITCH
Be a steward of your marketing data
How to Create Pure, Actionable Data and Avoid a Data Frankenstein Monster
With Impact CEO, David A. Yovanno
25. A Franken-stack has:
Lack of data transparency
Aggregate level data only
Non-scalable tools and systems
No de-duping or normalization
No cross-device consumer journey
26. A Franken-stack has:
Lack of data transparency
Aggregate level data only
Non-scalable tools and systems
No de-duping or normalization
No cross-device consumer journey
27. A Franken-stack has:
Lack of data transparency
Aggregate level data only
Non-scalable tools and systems
No de-duping or normalization
No cross-device consumer journey
28. A Franken-stack has:
Lack of data transparency
Aggregate level data only
Non-scalable tools and systems
No de-duping or normalization
No cross-device consumer journey
29. A Franken-stack has:
Lack of data transparency
Aggregate level data only
Non-scalable tools and systems
No de-duping or normalization
No cross-device consumer journey
32. OMNI-CHANNEL MEDIA MIX + ADTECH / MARTECH
Person-Oriented
Data Lake
Tracking ⬝ Verification ⬝ Identity ⬝ Pathing ⬝ Normalization
33. Tracking ⬝ Verification ⬝ Identity ⬝ Pathing ⬝ Normalization
OMNI-CHANNEL MEDIA MIX + ADTECH / MARTECH
Real-Time Marketing Intelligence
Person-Oriented
Data Lake
ATTRIBUTION
MODELING
REPORTING/
VISUALIZATIONS
ANALYTICS APIs
Data In
Data Out
34. To build a Super-stack, you need:
● Complete data ingestion
35. To build a Super-stack, you need:
● Complete data ingestion
● Event-level tracking
36. To build a Super-stack, you need:
● Complete data ingestion
● Event-level tracking
● Fraud verification
37. To build a Super-stack, you need:
● Complete data ingestion
● Event-level tracking
● Fraud verification
● Person-oriented view
38. To build a Super-stack, you need:
● Complete data ingestion
● Event-level tracking
● Fraud verification
● Person-oriented view
● Forecasting and scenario planning
39. 3
1
2
Digital media mix
Point of sale
Offline media mix
Other vendorseCommerce systems
Reporting/
Visualization
Attribution models
Predictive analyticsDynamic reporting / dashboards Machine learning algorithms
APIs
Data Lake
Event level tracking and ingestion
Verification ⬝ Identity ⬝ Pathing ⬝ Normalization
58. 58
A more compelling narrative is to tell a story on an actual use case of a
how marketer goes through the process to make an ad spend decision.
These are major bets and most marketers are using inaccurate data to
place these bets. Attribution was supposed to be the silver bullet - but for
the most part has gotten limited adoption. Why? There is a lack of trust in
the results. Why? There is no consensus on the accuracy of the data
being fed into the attribution model. Why? As example GA says you had
100,000 google display clicks, DCM says you have 110,000 clicks, Google
Ads says you have 90,000 clicks, and your attribution vendor tracking
80,000 clicks. Who is right? Maybe they all are - it all depends how you
you define a “click”. However, this assume all of the display ads are
tagged correctly - what happens if they are not, what happens is tag values
are being stripped, what happens if certain landing page URLs do track the
clicks. In these cases the google displays clicks may get attributed to a
different channel - thereby making your attribution results inaccurate. The
scary part is that most marketers are not even aware that the results they
are using to make ad spend decisions are based on inaccurate data.
59.
60.
61.
62. “The abundance of data from marketing
tools seduces marketers with a false sense of
precision, while marketers lack the tools,
skills, and processes to produce a holistic
analysis.” – Forrester
63. 4
Identity ⬝ Verification ⬝ Pathing ⬝ Normalization
OMNI-CHANNEL MEDIA MIX + ADTECH / MARTECH
Real-Time Marketing Intelligence
Person-Oriented
Data Lake
REPORTING ANALYTICS VISUALIZATIONS
ATTRIBUTION
MODELING