1. Next Best Action for
B2B account based
marketing
Edmond Sam • 2019
2. Edmond Sam | 2019
Framing the Business Problem
2
Stated Goal
● Personalizing Next Best Action suggestions based on the Healthcare
Professional (HCP)
● HCPs are individuals, thus have their own preferences
B2B marketing problem but
different
● Right content at right time to persuade all decision stakeholders
● Pharmaceutical not directly connected with patients
● Treatment information controlled
● Treatment consumption influenced by payer
3. Edmond Sam | 2019
Health Care Professional Journey
3
Metric Measurement
Coverage
How many protocol
decision
stakeholders?
Awareness
How many know
about treatment?
Engagement
How much time
spent? (e.g. webinar)
Reach Target / Attendees
Impact
Target business
outcome
4. Edmond Sam | 2019
Reference
Decision Model
For Next Best Action
4
5. Edmond Sam | 2019
1. Use cases that
could be
supported in
the component
5
Plan Build Run/Measure
“I want to get the
right story to the
HCP”
● Review funnel &
persona status
from system
● Define goals for
next marketing
iteration
● Set baseline
● Build & update
personalized
promomat to
persona and
organizations
● Measure HCP
funnel movement
● Use system data in
Marketing Mix
Model
“I want my team to
achieve business
goals.”
● Approve journey
policy for new
treatment in
system
● Set baseline
● Review system
journey policy
based on goals
● Update biz rule
● Review system
suggestions with
teams
● Review feedback
from HCP
“I want to help my
HCP” ”
● Update HCP
data
● Review and
prepare for
upcoming
messages
● Accept/Reject
suggestions.
Deliver content.
● Capture feedback
for system
6. Edmond Sam | 2019
2 - Traits and
preferences of the
HCP
personalization
6
7. Edmond Sam | 2019
3 - How these traits and preferences can be determined and calculated
7
HCP
data
available
● Reverse engineer path to successful
outcome
● Reinforcement learning for Markov
Decision Process
○ Individual with discrete states
○ Sequential decision making over
timeline (state, action)
○ Bellman optimality - max reward
○ Solve for optimal policy (π*)
No HCP
data
● Default persona
● Ask HCP, expert opinion by account
owners
● Look alike modeling using 3rd party
like Axciom or LiveRamp. “Find more
of this person”.
Inspired by Grid Dynamics Blog
8. Edmond Sam | 2019
4 - Evaluate the
on-going “validity”
or relevance of a
Next Best Action
suggestion?
8
Influenced
revenue
● Marketing Mix Modeling
● Multi linear regression
analysis
○ Sales, Media
○ Market (e.g.
population with
indication, payer
data)
● HCP touchpoints as media
● Revenue lift downstream
HCP funnel
movement
● Define expected next step
in journey after touchpoint
● Set baseline on variable
● Feedback or questions after
event
● Increase in time spent as
“engagement”?
● Regular survey for
sentiment and mindshare
9. Edmond Sam | 2019
5 - Success criteria
for the HCP
Personalization
Component?
9
External
● Financial targets: Revenue - R&D expenses
● Customer Satisfaction score
● Net promoter score (NPS)
● Case studies of quantifiable customer success
● Evidence of landing and expanding in target
vertical/geo
● Increasing # of pilots
● Robust pipeline on customer requests
Internal
● Team engagement (e.g. 360 review)
● “Internal NPS”
10. Edmond Sam | 2019
6 - How would you evaluate the value compared to the engineering time for this
component? Prioritization Scorecard.
10
Description Score
Roadmap
alignment
Innovation Overall solution: Agile marketing optimization in HCP
journey
● Does it enable part of process previously
unavailable?
● Does it improve existing step?
● Does it prove secondary option?
5 - Enable
3 - Improve
1 - Secondary
Customer
request
Measurement of pervasiveness
● Ideas portal with customer voting
● Veeva customer success team
5 - 50%+ customers
3 - 25% +
1 - 10%+
Integration ● Improves user experience in E2E process
● Non-functional requirement
○ Availability, Performance, Supportability
5 - Verifiable with client
3 - Quantifiable
1 - Unconfirmed
Ability to
execute
Technology
complexity
● Brand new infrastructure / technology stack
● Technology risks
5 - Same stack and data pipeline
3 - One is new
1 - New stack and pipeline
11. Edmond Sam | 2019
6 - Engineering
Time
2x2 Prioritization
11
12. Edmond Sam | 2019
7 - Additional
enhancements or
roadmap items
12
Plan Build Run/Measure
Studio to visualize and/or define:
● Funnel
● Decision model
● HCP Journey policy per
persona
● Persona discovery
Promo materials
● Auto scoring against
persona
● Suggestions on content
and format
Setup A/B testing
● Random sampling
● Test vs. Control variables
● Tracking of outcome
A) Prescription/consumption
data
- Collaborate with HCP &
pharmacies.
- Acquire app or license data from
existing consumer services.
-Solve known consumer pains in
exchange for data.
1. Upload picture of
prescription
2. Education on treatment
3. Reminder for treatment
4. Food compatibility
5. Auto refill
B)Trusted, “opt-in” benchmark
marketplace