Building Audience InsightsThe Data Value Pyramid
Presented by Marc RossenDirector of Media Strategy and Analytics, [x+1]
February 8, 2011
Before we start:Before we start:
System requirementsSystem requirements
Please press Please press *6*6 to mute your to mute your phonesphones
If you use this OS You can use these browsers
Windows XP SP2 Internet Explorer 8, 7 or 6 SP2
Windows Vista Internet Explorer 8 or 7, Firefox 3.5.7, Firefox 3.0.6, Safari 4.0.4, Safari 3.2.2
Mac OS X v10.5.8 Safari 4.0.4
Full system requirements available at http://office.microsoft.com/en-us/live-meeting/microsoft-off
Marc Rossen [x+1] Director of Media Strategy & Analytics
Marc Rossen is a marketing and business innovations strategy expert with more than a decade of experience helping companies innovate in digital marketing and analytics. At [x+1], he overseas thought leadership and innovation for agency and enterprise client relationships focused on audience insights and branding analytics in media. His work helps [x+1] clients transform the influx of data in digital marketing into actionable marketing gains.
Before joining [x+1] in 2010, Marc served as Digital Analytics Lead for the Procter & Gamble account at digital agency MediaVest. He also has worked at Publicis Groupe agencies VivaKi and Digitas on web-based analytic solutions, strategy and analytics for major brands in consumer products, automotive and financial services verticals. He started his career working on supply chain, growth and innovation issues for Staples.
3
Webinar #1:Webinar #1:
The Data Value PyramidThe Data Value Pyramid
Using data to effectively build insights is challenging, but has
huge payoffsMy goal today is help you understand how to use data effectively to gain actionable insights
The news is, almost daily, flooded with articles on data and the effects is has on business and our personal lives
With this data influx a problem arises of how to narrow down your data set to drive insights for your business
•What data is useful?•How do I simplify the data I have to gain insight?•How do a paint a picture of my audience?
5
Our Agenda TodayOur Agenda Today
Understanding the value of data and building a value framework that leads to actionable business results
•Actionable data•The role of context
•Findings •Insight •Action
Use our new framework to build an effective audience analysis
•An approach to building audience insights•Looking at the detail to derive tactical value
Agenda
Webinar #2
7
Where are we today?Where are we today?
We are still looking at data in a silos
If organizations are silo’ed, data is probably silo’ed•In today’s environment most people are looking at data in a silo without bringing multiple data points together to create a picture
Context is king•Therefore we are still struggling with deriving actionable data as we have context for the univariate data points we are looking at
9
The Broader Context
Data without Context…
10
…Is like looking at one corner of a painting
Puzzle Piece
How do we paint a picture of our audience?
At [x+1], we start with two core beliefs:
1.Gather all the facts together to identify the cause
2.Any one fact alone can drive you down the wrong path
11
1. With all the facts together we can identify the cause2. Any one fact alone can drive you down the wrong path
How do we paint a picture of our audience?
12
Metric Tool Set
Frequency was up 2x for
the week
Spend was up 2x
September 10
Los Angeles
No creative changes
Google Adx 2.0
Data without context
• Frequency• Spend• Exchange• Date• Geo• Day of Week• Etc…
The same applies true to us
How do we identify performance changes?To often one data point is chosen as the cause, however there is not context in place
13
What happened?
Building a framework:Building a framework:
The Data Value The Data Value PyramidPyramid
Building a value framework is imperative to build insights that lead
to action
Actionable Data
Finding
Insight
Action
Most organizations
are here
15
Organizations are running before walking
Most organizations look at a few data points and jump all the way to insight, take some action, and realize after some time performance has not changed. Why is this?
The [x+1] insight: Develop a roadmap that allows you to slow down and ask….do I have all the facts (e.g., data points) ?
16
By using a value framework you are forced to walk then run. In other words, you now have an analytic framework to asses your
data and determine if you have can paint an accurate picture
A simplified example of the framework
The [x+1] insight: Insights come from actionable data.
17
Actionable Data Finding Insight Action
RON CPA = $10RMKT CPA = $5
RMKT performed 2x better than RON
RMKT is more productive marketing tactic
Invest more $ into RMKT to drive campaign efficiency
Step One – Actionable Data
First, do not get overwhelmed by how much data you have!Next, take stock of all the data you have available and make a list
18
What happened on 12/7-12/8?
1. With all the facts together we can identify the cause2. Any one fact alone can drive you down the wrong path
How do we paint a picture of our audience?
19
Metric Tool Set
Frequency was flat
Spend was up 2x
December 7-8
CPM was flat
No creative changes
CPA down 2x
Data without context
• Frequency• Spend• Exchange• Date• Geo• Day of Week• Etc…
Step two - Findings
Once our data set is narrowed to what happened we can dig into univariates to derive findings that bring context
In our example its clear the data is telling us that our Google exchange inventory more than doubled and that this was caused by a large campaign ending
20
Step Three – Insights that drive action
By taking stock of your findings you can derive actionable insights
21
Finding Insight Action
Campaign had no operational issues (no pixel/creative issues)
Outside factors effected the campaign
No action needs to be taken as the increase in CPA was actually still within client constraints'. Due to lower spend CPA was already well below goal this room was still available to increase CPA while increasing reach.
Spend more than doubled while performance metrics dropped significantly
Spend caps were not hit despite spend doubling due to supply constraints
Exchange based inventory was effected for all exchanges
Supply constraints were prevalent for all exchanges
The value of a framework
Without a framework it’s common the following findings, Insights, and Actions would have taken hold
•Campaign frequency, spend, and/or inventory sourcing would have been optimized
•Unnecessary time spent diagnosing an issue that was not an issue
22
By using a value framework you are forced to walk then run. By running in this case, nothing needed to be done as campaign CPA
was still below goal!
Building Audience InsightsBuilding Audience Insights
Putting our framework in action
Taking a step back, we now understand:
•The value of using multiple data points for context
•How we build a value framework to derive insights that lead to action to improve performance
We are now ready to put our framework into action to build audience insights
24
Step one entails building our base set of actionable data
Actionable Data
Finding
Insight
Action
25
Audience insights are built by defining what makes your
customers stand out
What we are left with is defining characteristics that can paint a picture of who your customers are
26
Your Customers
The Internet Audience
These characteristics allow us to understand why performance trended the way it did and how it can be
repeated
Our CPA trend continually declined throughout the campaign as we optimized based on audience insights
27Source: DFP
AudienceOptimization
Audience Optimization
DMA
We then find audience characteristics that best describe your best customers
We find actionable data by isolating the attributes that drive
performance
28
Attribute Tool Set
Income Class
Income Producing Assets (IPA)
US Region
Ideal customers defined by:•GeoData•Internet connection speed•Internet browser•Computer Operating System•Income Producing Assets•Income Class•Social Life Stages•Presence of Children•Home Owner•Gender
By aggregating data into multiple points we bring performance into
context
We use multiple characteristics to define individual groups of your customers as each segment represents a different affinity to your product
29
People who live in Massachusetts, Florida,
Georgia, South Carolina, California,
and Pennsylvania
Elite, High, and Moderate IPA
Wealthy Income Class
Audience Segments aggregate multiple
characteristics
30
The client has 36 discreet audience segments across
GEO, Income, and IPA
Audience Segment
Key Takeaways• Don’t get overwhelmed by the data you have available• Take stock of your data set• Bring context with your data by looking at all actionable
data together• Paint a picture to define the problem you are solving• Assess your findings, derive insight, and define your
actions
31
Take Stock, Assess, Derive, and Define
Register for Webinar #2:Register for Webinar #2:
The Progression to The Progression to FindingsFindings
Thursday, February 10Thursday, February 10thth