BusinessOnline, Docusign, and Bizo address how Marketers everywhere are being challenged to assess ROI for each and every one of their marketing investments. They know last-click attribution doesn’t work: It doesn’t account for how display, email, search, social and other digital tactics all work together to drive prospects through the marketing funnel. So what is working?
The End of Business as Usual: Rewire the Way You Work to Succeed in the Consu...
B2B Marketing Metrics and Attribution That Works
1. B2B Attribution and Metrics
Richard Roberts
SVP, BusinessOnline
rich@businessol.com
2. Simple Definition
What is Attribution?
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
The practice of allocating proportional
credit to all marketing communications,
across all channels, that ultimately lead
to the desired customer action.
6. Why It’s Worth It
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Why integrate web activity with line-of-business activity to
get closed-loop attribution?
It gives you a powerful, data-backed methodology for
looking back on anonymous visitor activity and linking it to
known leads or sales.
This gives you a full picture – the ability to measure which
activities, campaigns or even content assets are best
leading to sales.
You can then optimize programs based on revenue rather
than front-end metrics like visits or leads.
11. Primary Attribution Model Types
Rules-Based
Single Touch
Type
Assigns 100%
credit to the last
or first exposure
Assigns credit to each
interaction based on specific
business rules
Statistically Driven
Assigns credit to each
interaction based on a
data driven model
First/Last Touch
Approach
Basic
E-Mail
Display
Search
E-Mail
Display
Search
33%
100%
33%
33%
27%
49%
24%
Even Weights
Custom Weights
Time Decay
Positioned Based
Advanced
Regression Model
Probabilistic Model
12. Pitfalls of Each Model
Single Touch
Type
Rules-based
Statistically Driven
Pitfall
Ignores the bulk of the
customer journey,
overvaluing the first or
last touch
Undervalues other
influencers (and
interactions), including
sales efforts
Basic
E-Mail
Display
Search
E-Mail
Display
Search
33%
100%
33%
33%
27%
49%
24%
Assigns arbitrary values
to each specific
marketing tactic
Subjective and lacks
analytics rigor to
determine weights
Advanced
Can be expensive to
execute
Some Marketers feel
uncomfortable with an
algorithmic, “black
box” approach
16. B2B Sales Funnel
Stakeholders in Buying Process
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Engagement Score
1800
1600
1400
1200
Analysts
1000
Marketing
800
IT Team
600
C-Level
400
200
0
Month 1 Mo. 2
Mo. 3
Mo. 4
Mo. 5
Mo. 6
Most of these stakeholders were anonymous
17. Go from Unknown Visitors to
Known Companies
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
~5% of visitors were known
Individual score of 130
18. Go from Unknown Visitors to
Known Companies
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
~5% of visitors were known
Individual ‘factor’ of 130%
50% known via IP mapping
19. Go from Unknown Visitors to
Known Companies
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
~5% of visitors were known
Individual ‘factor’ of 130%
50% known via IP mapping
Learned that 8 people from
certain company are engaged
Company factor of 725%
20. Company Level Attribution
Key Advantages
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
• Create success metrics based on Company status
• Example: evaluate marketing on aligning with
CRM opportunities
• Perform “Look Back Analysis”
• Analyze behavior of visitors from companies
that bought or are deep in the pipeline
• Develop company-level lead scoring
Notas do Editor
Non-linear
Challenges:Need to develop a new custom database schemaNeed to move all data into this locationScale & performance can be an issue (non-aggregated Web data)
Different channels or campaigns have different roles so you should apply the appropriate metrics depending on those goals.
Each attribution approach, in the end, wants to uncover that true value of the marketing interaction – an email open, search query, display clickthrough,etc. The first approach assigns 100% credit to the last or first exposure. It’s the most commonly used approach but we don’t even really categorize this as attribution. In fact, it can do more harm than good.The second approach assigns attribution weighting based on predetermined rules — examples of rules-based models include even-weighted, custom-weighted, time-decay or position-based. Theseallow you to assign fractional values to each interaction.We prefer thecustom weighted model that’s based on periodic look-backanalysis. The third approach assigns weights to each interaction based on statistical regression or probabilistic models. These advanced, data-driven models typically provide the most accurate picture of the customer journey.
First, the Simplistic approach ignores the bulk of the customer journey. It assumes the customer’s primarily decision factor was that first OR last interaction, thus resulting in assigning all the credit to that interaction. Based on what we know, that’s not only false, but grossly irresponsible for a marketer to assume this. Additionally, the simplistic approach grossly undervalues the key influencers in the customer journey.Second, the rules based approach allows the marketer to assign values to each tactic based on business rules - what they know about behavior of channels, of the business, of their industry. However, their “business rules” is based on what they think is correct about the channel-based on last interaction. If they do assign via time decay model, well, that just assumes that every customer, every interaction-is influenced the same way, which is false. So, in the end, unless doing the custom weighted model based on periodic look-backanalysis,this approach lacks analytics rigor and may not be a better approach than simplistic measurement. Finally, the algorithmic approach makes some marketers feel uncomfortable. Uncomfortable wit the black box approach to attribution. Uncomfortable with knowing the truth, with giving up their data to a model they don’t understand. And not being able to add their “gut feeling” input on decisions. Given the different approaches and pitfalls, we at BusinessOnline feel that the algorithmic approach is ideal but the creative custom weighted approaches can get you 80% accuracy on a holistic view of customer interactions.
B2B sales cycle: Many prospects from a single company research yet in different BUs, or departments
identify which company anonymous visitors are likely from and combine them together to get an aggregate view of a company’s total activity and engagement
identify which company anonymous visitors are likely from and combine them together to get an aggregate view of a company’s total activity and engagement
identify which company anonymous visitors are likely from and combine them together to get an aggregate view of a company’s total activity and engagement