Transform Advertising Attribution Overview
-
Upload
evan-dunn -
Category
Data & Analytics
-
view
342 -
download
0
Transcript of Transform Advertising Attribution Overview
DATA DRIVEN ATTRIBUTION
“It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” Sherlock Holmes in “A Scandal in Bohemia”
ATTRIBUTION STARTS with data
visi6ng a webpage
clicking on a banner ad
searching on the internet
making an inquiry
pos6ng on social media
placing an item into a shopping cart
making a purchase
watching a TV show
redeeming a coupon
booking a trip
using an app
paying a bill
Data is the byproduct of consumers doing th ings !
CURRENT ATTRIBUTION APPROACHES FIRST TOUCH
Assigns all of the revenue credit to the first media channel to touch the customer.
LAST TOUCH
Assigns all of the revenue credit to the last media channel to touch the customer
LINEAR
Assigns equal revenue credit to each media channel to touch the customer.
TIME DECAY
Assigns increasingly more revenue credit to the most recent channel to touch the customer.
MOST MODELS TOO SIMPLISTIC MINIMAL COMPUTATION OF OFFLINE TO ONLINE EFFECTS NO ACCOUNTING FOR EXTERNAL FACTORS COMPUTES CUSTOMER INFLUENCE BUT NOT ROI
TRANSFORM ADVERTISING genome
data driven model
environmental data
social sen6ment data
compe66ve media spend data
What is the true value of TV adver6sing?
How do offline ads drive online ac6vity?
What is my cost per lead? Per acquisi6on?
What is the best media mix?
What will a par6cular marke6ng budget generate in terms of inquiries or sales?
What media budget is required to generate my sales targets?
ADVERTISING genome COMPUTATION ENGINE
YIELD CURVES Computation model that determines the point at which incremental input does not produce corresponding incremental output.
Saturation rate for various forms of outbound advertising such as TV, radio or display.
COVARIANCE MATRIX Computation model that determines interrelationships between what may otherwise appear to be random values and variables. Effect or intermediate influence of one media channel on another.
Based on tools and models used by Wall Street to manage portfolios that have multiple types of securities such as bonds, stocks, cash, etc.
OUR genome data engine
bought data
3rd party web data
Client Business Data
public data
census data
labor stats
weather industry data
social sen6ment
interac6ve surveys
DATA LAKE INGEST -‐ TRANSFORM
GENOME
DASHBOARDS
MARKETING AUTOMATION
MEDIA TYPES • Television • Website • Print • Radio • Paid Search
• Retargeted Display • Social • Aggregator • In-game • Pre-role • Other (e.g. yellow pages)
GEOGRAPHY • By zip code • By DMA • By state • By urban area
WE MAKE ATTRIBUTION ACCESSIBLE
1. Sophistication = Complexity The Advertising Genome is highly sophisticated but hides complexity from users via a use-case driven graphical application.
2. Expensive, Lack of ROI The Advertising Genome generates meaningful ROI by optimizing media spend to generate more measurable results.
3. Lack of Data The Advertising Genome comes integrated with our Data Lake, which is over 20 terabytes of 3rd party data necessary to map the intricate relationships between customers, products, markets and media channels.
44% of interactive marketers do not have a measurement
process in place, not even post-click
30% of marketers still use first or last touch metrics, even though most know it is too
simplistic to be effective
Only 11% of marketers use any form of algorithmic attribution
tools
addressing issues that have kept most companies f rom using at t r ibut ion !
The ADVERTISING genome IN ACTION The predic6ve customer
acquisi6on model The actual customer acquisi6on
analy6cal dashboard
The ADVERTISING genome
Use-case based apps
External data integration
Intuitive user interface
Dynamic slicers
Built-in reporting & analytics
Custom UI ATTRIBUTION genome
Advanced graphic user interfaces
Integration of social intelligence data
Cross-channel intermediate influence
Real-time updates
Custom screens based on user roles
External data essen6al to adjust for changes in customer sen6ment, global/na6on/regional economic ac6vity, compe6tor ac6vity and scores of other environmental factors.
Formulaic computa6on provides greater than 95% accuracy of predicted cost per ac6on versus actuals.
User adop6on increases exponen6ally when “a]ribu6on” is integrated into day-‐to-‐day applica6ons such as media budge6ng planning, repor6ng and ROI analysis.
DATA DRIVEN ATTRIBUTION