N-B-A (Next-Best-Action) marketing is an approach that uses real-time customer data and analytics to determine the optimal next action or communication for each individual customer across marketing channels. It aims to improve profitability through more customer-centric interactions. When implemented by O2, an early adopter, N-B-A resulted in a 9% increase in bill value, 75% response rate, and reduced customer retention costs in the first month. N-B-A marketing considers each customer's unique profile and preferences to identify the single best offer or message to provide at any given time, avoiding issues like campaign collisions seen in traditional marketing.
2. Introduction
Despite all the hype surrounding customer centricity, many organizations still rely on product-oriented
customer interaction. For example, traditional direct marketing in which each campaign is set up to
promote a single product. While this approach aligns well with current thinking about budgeting, staff
compensation, and organizational structures, it is at odds with customer centricity as well as the ultimate
corporate objective—profitability.
This paper discusses a new approach to customer communications called Next-Best-Action marketing
(N-B-A) that can yield significantly more effective marketing and results. N-B-A is part of the wider
next-best-action approach to many types of customer interactions such as risk mitigation, remedial
actionsfor churn and fraud, service provisioning, data collection, arrears, surveys and so forth. When
used formarketing, N-B-A has direct and immediate effect as sales and retention will improve with a
corresponding boost in revenues. At the same time, marketing costs will decrease while the possibilities
for brand expression and control grow. Add to these benefits such desirable ‘side-effects’ as improved
customer satisfaction and happier customer-facing staff, and everyone will wonder why this new approach
was not adopted as the standard long ago.
Adoption of N-B-A marketing has been slow for two reasons: 1) It is a multi-channel proposition, and
only recently has real-time and customer analytics technology advanced to a level that can support the
complexity of marketing across channels; 2) Organizational structures, corporate experience, and comfort
with traditional marketing make such transformational change appear difficult and potentially risky. In
addition to the discussion of the benefits to be achieved with the N-B-A approach, this paper will suggest
ways for companies to overcome corporate inertia and transform the way they define, monitor and control
the customer experience.
2
3. First Things First: Profitability
To demonstrate how the N-B-A approach actually delivers, it makes sense to look
at the benefits O2, an early adopter of both the methodology and the technology,
has achieved. O2 was one of the first to embrace N-B-A marketing for their 5,000
call center agents, 2 million online visitors per month and 470 retail outlets.
Now the United Kingdom’s leading mobile telecom provider, O2 was only the third
largest when the company began its N-B-A journey in 2004. Recently, Telefonica “N-B-A has direct and
acquired O2 to create the second largest provider in the world, but because O2 has immediate effect as
always zealously promoted and protected its brand, its brand is perceived to be a sales and retention
key asset that is likely to be adopted internationally. will improve with a
corresponding boost
At the 2005 Gartner CRM Summit, O2 reported a bill-value increase of 9%, a
in revenues.”
response rate of ~75%1, a significant increase in customer satisfaction, and
reduced costs for retaining customers, all in the first month after a customer had
been exposed to an N-B-A guided customer experience.
These benefits were achieved with no appreciable change in average customer
handling time. The results are all the more significant because customer
interaction was not exclusively focused on sales, with equal attention given to
branding, as well as customer care and satisfaction. And although the use of the
N-B-A application was not mandatory at the time in the call centers, O2 achieved
75% adoption among representatives.
The essence of how such numbers were achieved at O2 and elsewhere is in the
sophistication of the interaction and the corresponding quality of the customer
experience. With N-B-A, it is now possible to build a mini-business case (in
real-time when necessary) to determine the best action to take. Once the action
is taken and the customer’s response recorded, the business case is immediately
recalculated and the next best action is recommended. O2 and others are using
sophisticated arbitration metrics to balance insight into customer interests, risks,
loyalty, etc. with corporate priorities, such as revenues, costs, and branding.
O2 has recognized that while the company’s overall goal is profitability, this does
not mean that every interaction must be focused on selling. A better strategy is
to continuously maximize the relationship throughout the customer lifecycle. For
a very few customers, that may mean reaping as much as you can now as there
will be little future in the relationship, but for most customers it means that every
interaction, conducted through any channel, is used not necessarily to sell, but to
nurture the relationship. The end result, as evidenced by the O2 case, is a stable
increase in profitability and customer satisfaction.
1
4. The Next-Best-Action Model
To demonstrate how the N-B-A approach actually delivers, it makes sense
to look at the benefits O2, an early adopter of both the methodology and the
technology, has achieved. O2 was one of t The N-B-A model provides a corporate
“Decisioning Hub” that guides each inbound and outbound customer action
and communication for every channel and line of business. This centralized
decisioning capability ensures the following benefits:
“With N-B-A, it is now
``Consistency across channels possible to build a
``Collision avoidance mini-business case
``Cost alignment (in real-time when
``Optimization of interaction time necessary) to determine
the best action to take.”
``Natural conversations
``Effective monitoring
``Total control
Consistency and Collision Avoidance across Channels
Absolute consistency, across channels and over time, is achieved because the
Decision Hub executes a holistic strategy for each customer that recommends
the best action. Any customer response or relevant event will have a bearing on
what the next best action will be.
For example, a customer who logs into a company’s web site or calls into the call
center or is contacted through an outbound communication will be subjected to
the same (best) action, assuming that the channel is not a relevant factor in the
decision. The action may be to first address the customer’s question about an
outstanding balance and then to offer her the Premium Platinum card, or 30 free
minutes of international calling, but regardless of which channel is accessed for
communication, the same messages are delivered to the customer.
There is a second advantage to this approach—avoiding campaign collisions.
In traditional campaign management, it is a challenge to avoid targeting the
customer with two campaigns simultaneously. She may be selected for the
Premium Platinum card as well as the higher interest Ultimate Savings card.
Even if the two products are not mutually exclusive, they will compete for her
attention.
2
5. A traditional solution is to run a separate optimization after one campaign
is executed to detect that the customer has received the Premium Platinum
campaign and remove her from the Ultimate Savings campaign. With potentially
hundreds of live campaigns, this is a cumbersome process.
Alternatively, once the Premium Platinum proposition is executed, the customer
may (temporarily) be on the suppression list for other campaigns. The question
then becomes one of sequencing: what’s the best sequence of campaigns to
maximize the customer’s contribution to the bottom line? The fact of the matter
is that both of these solutions are trying to fix a problem that should not exist in
the first place.
Mortgage Mortgage Mortgage
Figure 1:
Propensity Eligibility Strategy N-B-A selling strategy
example.
Loans Loans Loans Retention
Best
Propensity Eligibility Strategy Matrix
Offer
Cards Cards Cards
Propensity Eligibility Strategy
3
6. The N-B-A model is an entirely different approach, representing a single,
continuous campaign that selects the optimal treatment, one step, or decision, at
a time. Arbitration between alternative treatments can be just-in-time, avoiding
the time consuming post-processing required to assign the customer to a single
campaign. With N-B-A marketing, the customer will be considered for a specific
proposition or treat¬ment during real-time interactions or when generating the
output file for batch processed outbound communications.
Consider an N-B-A model as illustrated in Figure 1, a simple selling strategy that
will be extended later in this paper. When fed to the Decision Hub, this strategy
(or Decision Logic) will make a decision concerning the best offer to make. The
decision is made in real-time, the moment the customer enters the website
or contacts the call center. In batch, the decision is made when the output file
is generated for outbound communication. In both cases the decision is made
just-in-time and will select only one (the best) proposition out of the three that
are possible, depending on the customer’s propensity to buy any of the three
products, her eligibility to do so, the desirability of making the offer now, and the
economics of the purchase, such as margin, contribution to branding, effect on
market share, and so forth.
The N-B-A strategy avoids collisions because under the organization’s arbitration
scheme there can be only one best proposition. Assuming the Decision Hub is
fast enough, it would be a waste of time, effort and data storage to set up three
campaigns for mortgages, loans and cards and then apply the same arbitration
afterwards. As will be shown later, a single Decision Logic strategy simplifies
campaign management, reduces costs and IT effort, and, importantly, does not
differentiate between inbound or outbound use. How to design and implement
such an overall strategy in a multi-product organization will be discussed later in
this paper.
Cost Alignment
Resources are scarce in every enterprise – both in terms of time spent in
communicating and money spent in negotiating terms or incentives to make
customers agree to a proposition. The ‘art’ in marketing is still choosing where
to spend scarce resources. N-B-A strategies and their underlying predictive
models2 enable an organization to be discerning in its use of resources.
Given the prediction of the customer’s level of interest, some time can be justified
in offering a proposition to those very likely to purchase if they are simply told
about it. More time and an incentive can be justified for those less likely to
be interested because they need to be persuaded. For those with even lower
interest, time can be spent not on selling the proposition but on explaining it
and developing the need or desire. Those with a minimal interest may fall below
a threshold that prevents the proposition from being offered. This hierarchy of
strategies for a proposition optimizes the use of scarce resources and aligns the
cost of the effort to the likely value of the customer.
4
7. The style of the relationship and the costs involved to both parties can be
similarly managed and optimized over time. For example, during the initial
phase of a relationship, the emphasis in the conversation may be on building a
mutual understanding and level of trust. Early propositions can be structured
to be particularly simple to understand and compelling to the customer, if not
particularly profitable for the enterprise. When the value of the proposition is
proven, the customer’s level of trust in what the enterprise offers increases, and
later propositions can be more complex, requiring an element of faith on the part
of the customer. Further in the relationship, when most needs have been met, the
customer can be offered a trade-off in the style of relationship – lower fees or a
discount in exchange for using a lower touch and lower cost channel. The N-B-A
approach enables any and every aspect of the relationship to be explored and
optimized.
Optimization of Interaction Time
The value of an N-B-A Decision Hub within the organization is that it can be used
to intervene at any time and make recommendations concerning the best action.
For example, it is estimated that the total customer interaction time is about
3 hours and 22 minutes per year for a multichannel bank (Figure 2).3 For the
customer, much of this time is consumed by waiting, and it is during these
waiting periods that the Decision Hub can step in and suggest the action that
will most likely be meaningful to the customer. For the bank, this means that
interaction time is used effectively.
AVERAGE
CHANNEL INTERACTIONS
PER MONTH
Figure 2:
Yearly customer interaction
time in banking.
Branch 1.7
Average 3hr 22min
interaction per year
Web 1.8
Call 1.5
ATM 8.4
5
8. It is estimated that marketing intervention during the 3 hours and 22 minutes can
provide value equivalent to a 40-fold increase in the marketing budget.4 It also
means that campaigns rarely need scheduling in the traditional sense of creating
a fixed timeline for contacting the customer in predefined ways. Rather, a single,
fluid, real-time campaign can be deployed that grabs available time for marketing
during inbound contacts.
In addition, the N-B-A Decision Hub can be used to decide on the best action
for outbound communications, starting with the decision on whether or not
to engage in outbound communication in the first place. If this decision for
a particular customer is positive, the outbound proposition can be entirely
consistent with the inbound proposition should the customer happen to engage in
an inbound contact first. As a result, outbound campaign costs decrease for two
reasons:
``Outbound communication becomes more effective by avoiding collisions;
``Inbound communication is used to make propositions that would
otherwise be made in the outbound channels.
Research has shown that customers are more receptive to offers made in the
inbound channel. One reason for this is that customers choose their moments
to engage with the company and are not contacted while relaxing at home or
otherwise engaged. The second reason is that humans are more susceptible to
offers made during a natural conversation, especially when the N-B-A Decision
Hub guarantees a dynamic give-and-take process. Industry analyst firm Gartner
estimates a factor-of-10 improvement in sales with a real-time, customer-
initiated, relationship-driven conversation over traditional outbound mailings
(Figure 3). All three features – real-time, customer-initiated, relationship-driven
– illustrated in Figure 3 can be supported by the N-B-A Decision Hub.
ENTERPRISE CUSTOMER
CAMPAIGN-DRIVEN
Loans
Enterprise-Initiated, Loans
“Intrusive”
1% to 5%
Propensity
Marketing-Driven Strategy
Response
Figure 3:
Effectiveness of inbound
versus outbound offers.
EVENT-DRIVEN
Loans
Customer-Triggered Loans
“Convenient”
5% to 25%
Propensity
Product as Service Strategy
Response
Leveraging Inbound
Requests in Real Time
Loans
Customer-Initiated Loans
“Appropriate”
10% to 50%
Propensity
Relationship-Driven Strategy
Response
6
9. Natural Conversations
For both the customer and the enterprise, achieving a natural conversation yields
greater satisfaction, better use of time, and ultimately greater profits. It is also
more conducive to having customers agree to a proposition because humans
are accustomed to a give-and-take style of dialog. It is important that both the
“giving” and the “taking” match customer interests and corporate objectives. “Fundamental to the
N-B-A paradigm is that
For example, achieving a natural conversation in a call center means recognizing
there is no such thing as
the customer’s context for the call and aligning the way the agent responds
with the way the customer is thinking. When the customer asks about a specific one-size-fits-all.”
product, the agent responds with a description of that product and others that
are worthy of comparison. When the customer asks about a particular type of
product, the agent responds with the products in that group. When the customer
expresses a need to do something, the agent offers the relevant products from
several product groups. And finally, when the customer describes some event he
or she is planning or experiencing, the agent again offers the relevant products.
This is more that just a simple prioritization of propositions—it is dynamically
focusing the conversation on the way the customer is thinking.
For many customers, a natural conversation would not stem from a “here’s what
we’ve got” offer of various products. Yet that is typically what happens in call
centers and particularly on websites. Given the context of the call or website
visit, the N-B-A Decision Hub enables the best propositions to be offered and
accepted or declined at an increasing level of specificity by initially offering the
most relevant product groups and then the most relevant products within the
selected group. Even at the product level, the recommended propositions can
be compared, allowing the customer to dismiss some of them based on their
features, while an individual proposition can be illustrated so that the customer
can comprehend its terms and implications and can tune the proposition to best
meet his or her needs. There can even be an element of negotiation, evolving
towards a deal that is a win-win for both parties.
7
10. Fundamental to the N-B-A paradigm is that there is no such thing as one-size-
fits-all. This extends to the way the relationship is conducted. Some customers
prefer to be presented with the relevant alternatives and to reach a decision
for themselves. Some prefer a consultative relationship in which they explain
their circumstances and needs so that the organization can recommend the
appropriate solutions. Some desire a mutually managed relationship in which
they negotiate everything – the deal, the products it contains, the terms and the
pricing. A natural conversation will mean something different to each group of
customers. Now that these kind of complex interactions can be expressed using
Decision Logic, consumer psychology can play an increasingly important role in
designing the optimal customer experience.
By continuously suggesting ‘what to do next’ during the dialog, the N-B-A
Decision Hub supports iterative and interactive forms of communication that
customers recognize as natural. As a result, they know how to conduct their part
of the communication and focus their attention on what is being offered rather
than being distracted, confused, or annoyed by the process itself.
Effective Monitoring
When all customer-facing decisions are made by the N-B-A Decision Hub, the
enterprise is in a good position to automatically capture every decision and
the basis on which it was made. In fact, this is affordable only with a central
decisioning capability, as the impractical alternative is to trace all legacy rules
and other logic that is embedded in all the different customer-facing applications
and processes. When the effect of the decision becomes evident (perhaps
immediately when the customer clicks a button; perhaps a year later when the
customer repays the loan), it is then possible to judge the performance of the
Decision Logic strategies fed to the Decision Hub.
Such a comprehensive monitoring environment plays three important roles as
follows:
``It operates in real-time, giving management an up-to-date view on
propositions made and accepted or rejected. This is a key capability as
management (product, channel, CRM) can exercise total and immediate
control over all aspects—or at least all marketing aspects—of customer
interactions.
``It allows marketers and others involved in designing the customer
experience to discover which elements of their strategies work and
which do not. This becomes even more important when the Decision Hub
is used–as it should be–for experimentation and trying out alternative
strategies in a champion/challenger fashion. Such feedback can be at
the level of whole strategies or individual business rules and predictive
8
11. models that make up Decision Logic. Figure 4 illustrates the use of
experimentation. In this case two retention strategies are randomly tested
once the decision has been made to try to retain a customer. This decision,
in turn, is based on the risk of defection and the expected customer value,
the combination of which determines the retention budget, if any.
``Comprehensive monitoring stores each decision as well as the basis
for that decision, allowing internal and external auditors to replay the
interaction and assess the underlying policies and assumptions. For some
areas of decisioning, this is becoming a legal requirement, such as the
Basel II framework that governs capital requirements in banking.
Retention Figure 4:
Strategy 1 Experimenting with retention
strategies via the decision hub.
Value Retention Champion/
Prediction Matrix Challenger
Exit Risk Retention
Prediction Strategy 2
9
12. Total Control
As is the case at O2 and other companies, strategies fed to the N-B-A Decision
Hub are developed and maintained by line of business management rather
than IT. The advantage of this approach is that the company can adapt its
strategies (and, by implication, its profitability and aspects of branding) on
the fly. It is important to note that this ability does not mean strategies should
be implemented casually. They will need to be tested before they go into
production as in any normal business practice. However, with the Decision Hub
as the central location for making all customer-facing decisions, changing the
instructions changes the way the company does business. For example, O2 is
making changes to their extensive customer experience strategies a few times
each week for a variety of reasons such as:
``Experiments (Figure 4) have shown that a particular challenger strategy
should be promoted to the default strategy.
``Agents in the call center complain about a conversational flow (script) that
is not producing the desired effect and requires modification.
``The strategy needs to be changed in reaction to an action by a competitor
and requires a same day response.
O2 can do all of these things because line-of-business management, including
CRM and marketing managers, have total control over the customer strategies.
While many changes require redeployment and retesting, some tactical changes
need to take immediate effect. Suppose a company is running out of stock for a
particular incentive. Operational management needs to be able to change the
Silver Pen incentive to the Leather Wallet instantly. Similarly, if a flu epidemic
strikes the call center staff, strategies may need to be adjusted to route inbound
contacts to other channels or reduce call handling time by shifting to faster-to-
complete propositions. With an N-B-A Decision Hub, such tactical changes can
be made by adjusting certain control parameters that are part of the Decision
Logic itself. Thus, rather than changing the structure of the strategy to any
degree, parameters used within the strategy are modified. The Decision Logic
does not have to be redeployed, as the extent of adaptation has been predefined
and pre-approved, and the effects are merely tactical.10
Similar to the modern Fly-by-Wire aviation approach where a pilot moves the
joystick to steer the plane without interfering with the mechanics, the company
becomes an agile and adaptive Fly-by-Wire enterprise with all checks and
balances firmly in place.
10
13. N-B-A Marketing Realized — Shifting
Operational Paradigms
The benefits of the N-B-A paradigm as discussed above are extremely valuable
to any customer-focused organization. And as the approach has been proven
by some of the world’s most successful brands, implementation is clearly
feasible, and will be discussed later in this paper. However, as with all market-
driven paradigm shifts, there must be similar shifts in operational paradigms
within your company to fully realize the potential benefits. Both technical and
organizational in nature, these shifts must be examined in some detail as many
large companies are organized in a way that resists this type of innovation and
change. Certain areas of the business will be impacted the most and may need to
modify their processes, and often, their ways of thinking. These areas include:
``Marketing & Organizational Structure
``Outbound Marketing
``Call Center
``Enterprise IT
Marketing — A Collaborative, Customer-centric Approach
Marketing faces perhaps the most important paradigm shift of all. Marketing
strategists have to abandon their natural product focus and adopt customer-
centric thinking. This has been the CRM mantra for some time, but the truth
is that existing CRM technology has never allowed a truly customer-centric
approach. N-B-A marketing, in contrast, forces everyone in the organization to
put the customer first.
All staff involved in thinking about customer interaction strategies, including
branding, product marketing, risk management and CRM, will need to answer
this Critical Question: What is the approach we will take to maximize the
relationship with each customer when contact occurs? The answer triggers
further questions and the full strategy unfolds. From the start, this rigorously
top-down strategy design is at odds with the traditional product-first approach
when setting up campaigns and selecting customers for the propositions.
The answer to the Critical Question immediately requires collaboration between
departments that may hardly interact. This lack of collaboration comes at a
great cost, confusing the brand and bypassing potential revenues. Without this
collaboration, the first decision made will be owned by the business function that
happens to own the interaction. Customer-centric interactions demand that a
company agree on the answer to the Critical Question.
The first step to answering the Critical Question is for the company to agree on
its priorities for the business. Is risk more important than selling? Is selling more
important than retention? Consider the simple N-B-A logic example in Figure
5 that could be executed in any channel to answer the Critical Question and the
11
14. Is there a credit risk,
and, if so, Credit Credit
Risk Strategy
how to respond?
Retention
Strategy 1
Is there an exit risk?
If so, should we invest
Value Retention Champion/ NEXT BEST
in retention? If so, Prediction Matrix Challenger
how to retain? ACTION
Exit Risk Retention
Prediction Strategy 2
Mortgage Mortgage Mortgage
Propensity Eligibility Strategy
If there are no risks,
what proposition should Loans Loans Loans Retention
Best
we make and how Propensity Eligibility Strategy Matrix
Offer
should we make it?
Cards Cards Cards
Propensity Eligibility Strategy
Figure 5:
Building customer centricity
into interactions.
12
15. subsequent questions that follow that first answer.
Figure 5 illustrates how to fit a single campaign to the customer, not fit the
customer to the campaign as with a traditional approach, where prospects are
assigned to predefined marketing segments. Segmentation, even when based on
a thorough customer analysis and not on ill-founded assumptions, is the ultimate
bottom-up approach to customer interactions. It is not uncommon, for instance,
to see organizations define hundreds, even thousands of customer segments,
while their options to differentiate propositions represent only a fraction of
that. Instead, if companies addressed the Critical Question first, segmentation
would naturally follow from the level of customer differentiation that is actually
operationally supported, reducing costs and enhancing transparency.
With the Critical Question as its starting point, the N-B-A Decision Hub applies
its strategies repetitively to the same customer. As the customer’s state is likely
to be affected by the recommended action, the next-best-action will be different
with each application.
To maximize bottom-line benefits, the Decision Hub considers operational
constraints as well as customer inputs to sometimes defer potential propositions.
At any time, the next-best-action may be determined to be no action. Perhaps
the Decision Hub decides it is best to wait until a later contact moment or not to
burden the customer with a proposition that is likely to be perceived as annoying.
Inside a company, strategy teams from multiple disciplines must come together
to help determine the answer to the Critical Question. Priorities should be based
on expected profitability, operational constraints, branding considerations,
relevant regulations, and company policies.
Having agreed on the Critical Question, the representatives of the different
departments then work on their internal Next-Best-Action sub-strategies, such
as for arrears management or sales. Within the department, this process should
begin with the runner-up question. For example, if the answer to the Critical
Question is to sell, which offer should be made first? Each following question is
answered until the sub-strategy is developed to the degree possible.
Not all sub-strategies need to be determined up-front. In fact, even the perfect
strategy would only be perfect for a limited period of time, as the market is a
fluid environment. Detail to basic assumptions can be added later, as the N-B-A
paradigm is perfectly suited to address a single business issue. For example,
the organization can start with N-B-A for fraud detection with the first decision
during an interaction about the likelihood of fraud. If the risk is deemed to
be satisfactorily minimal, traditional customer processes take over. N-B-A
strategies or sub-strategies for retention, arrears management or selling may
be added later, one at a time. O2 has been in production with N-B-A logic since
early 2004 and has achieved impressive results from the start with proposition-
13
16. to-acceptance conversion rates in the 50 percent range. Even so, O2 continuously
improves the logic to address more customer issues in greater detail. Results
continue to improve with conversion rates now in the 70 percent range.
Once the virtual strategy team signs off on the sub-strategies, they are
assembled into a corporate strategy. Technology helps to make this potentially
hazardous task safe and quick.5 The strategy can then be deployed and go
live. (See Implications for Enterprise IT Solutions section.) The monitoring
environment allows different departments to assess and calibrate the
performance of their respective strategies. (See Effective Monitoring section.)
The team discusses results and decides on strategy corrections and extensions,
priority changes and responses to feedback from monitoring or user comment.
New versions of the strategy, which are typically subtle changes, can be deployed
any time.
This is all possible because sophisticated N-B-A Decision Logic can be safely
created and maintained by non-technical users as it allows them to intuitively
combine predictive models that were once the realm of statisticians with a
completely new breed of user-oriented business rules that no longer require IT
ownership.
Traditional marketing campaigns are hardly portable because they start with
the detail and are very product specific. In contrast, N-B-A campaigns start with
a holistic view of the customer and add the detail last. As a consequence, the
skeleton logic does not vary much from one telecom company to another, or
from one bank to another bank. The implication is that best practice blueprints
can successfully be used to fast track the implementation of N-B-A marketing
solutions. Indeed, implementation time for early adopters of 8 months to
recuperate costs and 18 months for a full multichannel rollout, has shrunk to
about half that because of the availability of N-B-A blueprints.
In addition to the enterprise strategy team, switching to an N-B-A marketing
paradigm will require a high-throughput customer analytics department. High
throughput is necessary to drive successful prediction-based strategies. There
will be a large demand for high quality predictive models to help make decisions
about likely customer interests (in buying products, channel preference, likely
amount spent, etc.), risks (late or no payments, exit risk, fraud, etc.) and values.
With recent advances in predictive technologies, it is now less important that
14
17. the analysts have a PhD in statistics than that they fully understand the business
purpose of the analytics. Using these business-friendly tools, companies can
easily integrate the customer analytics department with the marketing or CRM
department.
A company like O2 easily requires a hundred different predictive models, each of
which would take weeks to develop using traditional tools, while they take only
hours using the latest technology, which is more business oriented and offers a
model factory rather than a model laboratory. The demand for predictive models
can be particularly great in sectors like financial services and telecom where
products are increasingly defined just-in-time as a dynamic combination of
features. Working together, the business analysts and strategy team will form
this model factory that can replace or augment the current modeling capability
relied on by many large organizations.
Fueled by the predictive model factory, an N-B-A marketing strategy will
enable companies to market to consumers based on anticipated behavior.
Traditional attribute-driven segmentation can only tell marketers how to sell to
a homogeneous group, it cannot describe exactly who to sell a specific product
to. With behavioral segmentation, an individual can become a “Segment of
One” in which non-linear interplay between individual dimensions can be used
to select exactly the right action or offer for exactly the right individual. For
example, someone making $100,000 per year in Oklahoma will probably have
a different spending pattern than someone with the same salary in Manhattan.
From a marketing perspective, salary cannot be linearly interpreted in order to
understand a particular person’s behavior. Similarly, a set of parents of a certain
age may buy a rubber duck for their child. The same couple may purchase a
rubber duck again for their grandchild. Segmenting this couple into a single
“rubber duck–buying” age category eliminates the possibility of selling to the
grandparents.
To create a successful N-B-A strategy, marketers will have to change their
thinking, from traditional attribute-based segments to multi-dimensional
behavioral segments, that are actionable, and exactly what’s needed to drive
N-B-A strategies.
15
18. The Role of Outbound Marketing
In traditional marketing organizations, outbound marketing is an independent
marketing silo tasked with increasing sales by getting product propositions in
front of prospects. In an N-B-A-centric marketing organization, this is considered
one (important) element of the total customer experience. The dialog with the
customer will touch many channels, and the outbound channel is one of them.
But rather than just being a direct sales channel, its job should be to encourage
customers to engage with the more effective inbound channels.
The real-time inbound channels, as was discussed earlier, are more suitable for
natural conversation and to exploit the fact that the customer, not the company,
has chosen the moment to interact. Because of this, inbound is more effective for
conducting the real business.
In general, the customer experience is the result of continuous interaction across
channels and over time. Directed by the Decision Hub all channels will ‘know’
what role to play and how to carry on with the conversation. For instance, an
inbound interaction may conclude with the customer’s agreement to be contacted
later. Outbound communication is required to keep that promise towards the
customers and ensure from a company perspective that leads are properly
managed and not wasted. As another example, certain lifecycle events like a
birthday may trigger an outbound action by the Decision Hub.
Some enterprises even choose to consolidate these kinds of triggers along with
most of their outbound campaigns into a holistic outbound customer contact
strategy. Rather than having a January mailing for product 1, a February mailing
for product 2, etc., all campaigns are competing for a customer (to be decided in
true N-B-A fashion) in a single monthly outbound campaign.
Next-Best-Action in the Call Center
McKinsey6 states that in a bank that realizes a ‘core product conversion rate’
of 4%, adding 5 extra agents to the call centre adds additional sales power
equivalent to that of a mature bank branch. A sales force to behold. The call
center is only one of the channels used by large organizations, but the effects of
N-B-A marketing can be felt perhaps most directly in this area. Where previously
call center agents could be dedicated to a task (service, selling or retention), the
N-B-A paradigm mandates that the customer take center stage. This means that
the agent will need to follow the lead of either the customer or the pro-active
recommendations from the company’s N-B-A Decision Hub.
The first action an N-B-A Decision Hub could take is to intercept the call while
it is being handled by the voice response system and route the call to the
16
19. appropriate queue or agent.7 Or, the Decision Hub can be implemented to get
involved by guiding the agent only once the agent has taken the call. Once in
place, the N-B-A Decision Hub becomes the agent’s partner8 in making each
customer experience as effective as it can be.
A good N-B-A strategy represents the call center’s best practices, which
empowers agents, especially those who are less experienced. With the high
turnover in call center staff, this is an important feature in its own right. In
addition, all agents like recommendations that meet the customer’s needs most
precisely as well as helping those with a sales incentive meet their targets. The
Decision Hub can alert less experienced agents of promising propositions they do
not know about, while reminding highly experienced agents of propositions that
are forgotten, easily overlooked or only rarely relevant. As a result, customers
and agents experience much higher levels of satisfaction.
By receiving continuous recommendations for the next best action from the
N-B-A Decision Hub, each interaction becomes a meaningful conversation that
fills gaps that normally occur when the agent is waiting for data from the back-
office. Because of this, average call handling time is not significantly affected,
even when a whole series of recommendations is made to the agent during the
call.
Another aspect outside the scope of this paper but worth mentioning is that
current technology allows for a call center application that is almost 100% driven
by Decision Logic, including presentation elements. This means that agents can
not only give input on the recommendation logic and expect a quick turnaround
(because of the independence of IT), but also on the look and feel of their
desktop.
Implications for Enterprise IT
The Next Best Action paradigm stands on the shoulders of three important IT
breakthroughs: 1) the adoption of enterprise-wide application servers like IBM’s
WebSphere™ and BEA’s WebLogic™, and the Service-Oriented Architecture (SOA);
2) the speed of even modest hardware today being sufficient to meet real-time
demands; and 3) the coming of age of artificial intelligence techniques, allowing
for the aforementioned model factory to help achieve better alignment between
IT and business.
Implementing the N-B-A model has many implications for IT, most of them quite
technical and out of scope for this paper. For the skeptical reader, it helps to
realize that several large companies have already implemented the concept to its
fullest extent. For the sake of brevity, the IT implications discussed here will be
17
20. limited to two—opening up operational processes for business intervention and
the data logistics necessary to support an N-B-A driven enterprise.
Effect on operational processes. Perhaps the most significant change to IT after
implementing an N-B-A Decision Hub is that business can take ownership of
operational processes. Or more accurately, intervene in operational processes
for marketing purposes.9 Although this intervention is done in a carefully
controlled manner, it still encroaches on IT realms that previously required hard
programming or business rule development (assuming the operational CRM
systems allowed significant customization in the first place).
In an N-B-A-based organization, IT is able to give business the means to take
control of the dynamic, decision-driven part of the process while they continue
to own the process flow itself (and the operational system interfaces defined
by IT.) If IT implements a Decision Hub and a trusted way of deploying Decision
Logic to this Hub, business can fulfill its requirements and adapt the Hub when
appropriate. However, business may need to do so more often than IT might
appreciate because market dynamics are more volatile than the corporate IT
infrastructure can afford to be. In fact, such business agility is a key contributor
to the success of the company. To maintain appropriate checks and balances, IT
must define Decision Logic criteria in advance and establish a process that forces
business to adhere to them. In brief, this problem is solved using meta-data, such
as advanced data dictionaries, to force business staff to design Decision Logic
that will be safe to deploy from a data logistics point of view.
Data logistics required for N-B-A. The other major implication from an IT
perspective is in data logistics. As the N-B-A model puts great value on exploiting
interaction opportunities in real-time channels, the biggest challenge is to find
a cost-effective way to manage real-time data. While the details are not in scope
for this paper, a few key observations can be made:
``For real-time decision support, not all data needs to be available in
real-time. More precisely, the data needs to be delivered in real-time but
does not always need to reflect the up-to-the-second customer situation.
A distinction between current real-time, near real-time and non
real-time data requirements is important. A pragmatic approach extends
the scope for real-time over time and starts with just the minimally
required data elements. Even this modest scope produces many of the
benefits discussed earlier. This is because customers are usually most
sensitive to the company’s response to the most recent information they
have given. For example, try repeating an offer if the customer just said
‘no’. Fortunately, the in-session data created during the interaction itself
18
21. is easier to come by than some of the data that needs to be retrieved
externally, such as the current account balance. In summary, the reason
is that the customer touch point itself collects the intra-dialog data and
can send it to the Decision Hub (at the same time as making a decision
support request for example.) Successive drops of the project will see a
migration of static customer data from near real-time to current
real-time, but only where the advantages of doing so are previously known
in order to weigh the costs.10 Data logistics roadmaps are available that
reflect best practices in this area.
``Contrary to expectations, making decisions, supporting calculations
and predictions just in time (realtime), is likely to make data logistics
easier, not more complex. If this seems paradoxical, consider dining
in a restaurant. The traditional prepare-in-advance approach (or ‘data
warehousing’ in IT terms) would be comparable to cooking all possible
courses in advance, ready for when a customer chooses that particular
dish. If this sounds far fetched and unappetizing, consider that product
propensities are usually calculated in advance and then added to the
customer data warehouse. The N-B-A alternative is to calculate the same
propensity to buy in real-time. This not only avoids the need to store all
propensities in a huge database, but also means that the calculation
will be made only for that customer for whom the propensity is relevant.
Moreover, it is possible to make the calculation using the very latest
information provided by the customer—a negative response to the
previous offer, the call reason or answers to questions.
The reason decisions, or more basic output like predictions and scores,
are usually calculated in advance with only the data retrieval as a real-time
process, is that not so long ago there was no other choice from a performance
perspective. Today, decision engine computations are of the same order as
database access in terms of speed. In other words, if data needs to be retrieved
from a database to arrive at a decision, the actual making of the decision does
not add significant overhead.
In summary, it appears the N-B-A way of thinking about customer interactions
has significant technical implications, but in the end will make life easier for
IT. They will become less of a bottleneck for marketing where agility is so very
necessary and the lack of it a significant source of frustration and a drain on
revenues.
19
22. Conclusion
Next-Best-Action solutions for marketing and other business areas represent
a necessary paradigm shift for companies that hope to increase profitability
while maintaining high levels of customer satisfaction. Technological advances
in predictive analytics and interface presentation have enabled new levels of
customer-centricity within an organization. As recent projects in this area have
been completed well within time and budget constraints, the biggest remaining
hurdle for companies in implementing such a radical shift in business focus
might be the new ways of thinking and new levels of intra-company collaboration
that are required.
In spite of these potential challenges, the quantitative, tangible business benefits
described in this paper are of such magnitude, that quick success will inevitably
engender even more enthusiastic and profitable collaborations between
departments. In fact, the first implementation of an N-B-A-centric marketing
organization achieved significant returns on investment within eight months. It
is too early to tell, but this paradigm shift may finally dissolve the artificial walls
that exist around departments and move customer experience to the forefront of
corporate awareness for entire industries.
Finally, Next-Best-Action thinking coincides nicely with the general trend of using
inbound channels to extend and partly replace outbound channels. Customers
are more likely than ever to choose their own moment and their own channel to
contact the company, then select, negotiate, and configure the product or service
of their choice. Analyst firm Gartner is not only anticipating the growing use
of inbound channels, but also an explosive demand for solutions that leverage
interactions through these channels to create an intelligent inbound channel
(Figure 6). As a Next-Best-Action solution adds value to every customer decision,
its bottom line contribution goes up with both the number of interactions and
their complexity. It appears this revolutionary approach will land on fertile
grounds. Implementation should have a high priority for organizations that want
to carefully control the customer experience and fulfill their brand promise as
well as for companies that are looking to make the most of customer interaction
time to increase revenues and optimize the use of resources.
Footer Notes
1. This high number can in part be attributed to the type of propositions made, such as selling a
relatively inexpensive texting (SMS) subscription. Perhaps more telling is the change in conversion
ratio compared to O2’s traditional approach using the same propositions: a factor-of-15!
2. Predictive models are essentially mathematical expressions that relate known customer data, such
as age, income, gender, and so forth, to expected behavior, such as the customer will buy product x.
Such models are usually developed by statistical or data mining software.
3. Banco Espirito Santo, Presentation at the annual Dutch Dialogue Marketing Association conference,
September 2001.
20
23. Figure 6:
TRADITIONAL INBOUND Growth in inbound channel
interactions.
INTELLIGENT INBOUND
NUMBER
OF CUSTOMER
INTERACTIONS
EVENT-DRIVEN
CAMPAIGN-DRIVEN
TIME
4. By making extensive use of data dictionaries and other meta-data to make sure the sub-strategies
can be safely blended. The technical aspects of this are outside the scope of this paper.
5. This could go as far as having the Decision Hub predict likely call reasons to optimize the voice
response menu structure in real-time and minimize customer navigation time. This use of real-time
N-B-A Decisioning is outside the scope of this paper.
6. The McKinsey Quarterly, May 2006: Using call centers to boost revenues.
7. This could go as far as having the Decision Hub predict likely call reasons to optimize the voice
response menu structure in real-time and minimize customer navigation time. This use of real-time
N-B-A Decisioning is outside the scope of this paper.
8. AOL, after a first implementation of this new paradigm in Germany, calls this form of advanced
decision support the ‘call center buddy’ (American Marketing Association webcast, June 8th, 2006).
9. This paper emphasizes marketing aspects. However, an N-B-A Decision Hub can be and is used for
different purposes like intelligent lending, collections, fraud, etc.
10. This cost/benefit analysis is greatly facilitated by the fact that all customer experience logic has
been consolidated. Simulating the business benefits of using up-to-date customer data in making
N-B-A recommendations would otherwise be next to impossible.