Download Decoding Predictive Marketing

Decoding Predictive Marketing
AN INTRODUCTORY GUIDE
ContentsING
PAGE
3 Introduction to Predictive Marketing
PAGE 10
Hidden Insights in CRM and
Marketing Automation
PAGE 13
Understanding Predictive Models
PAGE 18
Lead Scoring
PAGE 23
Account-Based Marketing
PAGE 25
Account Scoring
PAGE 27
Customer Expansion
PAGE 32
Conclusion and About Lattice
Introduction to Predictive Marketing
Given growing revenue targets, many top modern marketers are realizing that they need deeper intelligence to keep up with shifting buyer behavior.
Predictive marketing works by taking all the data in the world ­— including account-level information
about the businesses we sell to and the lead-level information about the people we actually sell
to ­— and applying modern data science to solve top marketing challenges. Some commons questions
and challenges include:
Who is going to be my next customer?
How can I find more of these ideal customers?
3
s
s
s
s
How do I convert them?
Here is a sample of the predictive attributes at the contact and account levels
that are hidden across a wide variety of sources.
EXTERNAL
Selected Attributes
Marketing Automation
Contact name, title, company, open rates, unsubscribes, web visits,
pages visited, lead score, video views, downloads
CRM System
Company, contact information, win/loss, deal value
Product Usage Logs
Features used, logins, session length, collaboration
Purchase History
Products purchased, prices paid, discounts, contract terms
Customer Support History
Complaints, resolutions
Public Websites
Job postings, grants, litigation, patents, contracts, locations, growth
Company Websites
Language(s), products, shopping cart, executive team profiles
Social Websites
Company and personal profiles, likes, comments, updates, friends/
connections/followers, usage
Media
News articles and stories, product launches, announcements,
press releases, litigation
Private Databases
Credit ratings, financial history, construction permits/
starts, deployed technologies
Predictive
analytics has
emerged as
possibly the
single-most
important
technology and
competitive
differentiator
for B2B
marketers
to adopt.
—Matt Heinz,
Heinz Marketing
4
s
s
s
s
INTERNAL
Source By applying new technology to the wealth of data at their disposal, marketers now have
access to predictive modeling without having to turn to a team of data scientists.
By better understanding buyer behavior and intent, marketers can score leads and
prioritize accounts in addition to selling more to their existing customers.
Upsell
It’s no wonder that predictive analytics is emerging as
central to the modern marketing organization.
By leveraging data science to make sense of all the data
in their midst, leading B2B marketers are marketing
and selling more intelligently. —Shashi Upadhyay, CEO, Lattice Engines
Cross-Sell
Retain
Picking Up Where Marketing Automation Leaves Off
5
s
s
s
s
Perhaps you’re thinking, “I’m already marketing efficiently by using marketing automation.” It’s true
that marketing automation software can streamline the marketing process from end to end. It can
also track the historic behavior of prospects and score leads.
Taking Modern Marketing to the Next Level
1990-2000
CRM systems emerged as
a must-have for companies
large and small.
2000-2010
Marketing automation
systems quickly became a
staple for companies to
digitally engage their
databases.
2010-2020
Progressive companies are
turning to predictive apps
to drive conversions.
Marketing automation has allowed marketers to collect more data on their prospects than ever before.
However, from a performance standpoint, the best that marketing automation can do is provide a view
into what happened in the past. The actions marketers can take on marketing automation data are
purely reactive. You learn something about a lead then you use that data to take action ­— send them an
email, add them to a campaign, alert sales that they’ve done something or score them based on those
actions. By contrast predictive marketing is proactive. It takes all that data into account and blends it
with data that is unseen by the naked eye, allowing marketers to guide their buyers' journeys based on
all that is knowable.
Marketers once
had to guess
where their
sweet spot was.
Now we can use
data science
to tell us.
—Meagen Eisenberg,
Vice President of
Demand Generation,
DocuSign
6
s
s
s
s
As marketing evolves from a cost center to a revenue driver, companies that have successfully
implemented CRM and marketing automation are now looking at what's next.
Predictive marketing can show how prospects engaged with various marketing channels, or which
campaigns performed better than others. It can help answer the following:
Which companies are the best targets?
What marketing activity is most likely to yield the best results?
How much new revenue could potentially be generated?
Reporting
IVE
DESCRIPT
100 customers
bought product X
TIVE
PRESCRIP
Companies with
increased hiring rates
will buy product X here
Sophistication of Technology
Pairing product X
with a Y at these
accounts will generate
$21 M in revenue here
Predictive marketing
combines predictive
and prescriptive
analytics to forecast
what will happen
and how to make it
happen.
7
s
s
s
s
Business
Value
Predictive
Analytics
Predictive
Insights
The Time is Ripe for Predictive Marketing Why is predictive marketing a must-have?
25 percent of all Fortune 500 companies and 76 percent of the largest SaaS providers
are using marketing automation. Overall adoption rates are above 50 percent for SMBs
and more than 70 percent in larger organizations. Many of the marketers who embraced
the promise of marketing automation have pushed the software to its limits. They’ve refined
their campaigns and messaging based on the information they’ve collected via the system.
They’ve improved efficiencies but are now looking for ways to optimize their performance.
25%
As more B2B organizations seek to win over entire buying committees involved in purchase decisions,
they are moving from pure contact-level to strategic account-level marketing. Marketing automation
was developed around the concept of a contact database. For that reason, most of these systems are
less adept at addressing entire accounts versus individuals. Yet marketers cannot afford to ignore the
wealth of buying signals or the account-level attributes that can provide key insight into the needs
of prospects.
of all Fortune
500 companies
use marketing
automation
76%
of the largest
SaaS providers
are doing
the same
8
s
s
s
s
Previously only the most sophisticated companies could make use of predictive analytics. If marketers
wanted to make marketing more predictive, they were forced to rely on a team of highly trained data scientists using complex analytic platforms to build predictive models from scratch. Since these data teams
often served as a shared resource across the organization, marketers often waited weeks or months to
have their requests fulfilled. Now, the power of predictive analytics is accessible to any company. A new
generation of predictive marketing applications is harnessing the power of machine learning to democratize their use by actual business users rather than by PhDs.
Take note from Internet giants like Amazon and Netflix. Both companies have become successful based
on developing recommendations from predictive modeling. In fact, Amazon notes that 35 percent of
its product sales result from its recommendation engines. Both of these companies combine profile and
behavioral indicators from thousands of signals from the Web, social media, news sources and
beyond to power their predictive models. In essence, they’re tapping into all the information that indicate when a customer is likely to need a specific product. For example, you may not be looking for a
shovel but Amazon knows that your neighbor just bought one, indicating you may need one too. By
using all the data in the world, every marketer — including you — can optimize his/her revenue funnel to
simultaneously improve conversion rates, increase revenue and improve lead velocity.
Key Takeaways
of Amazon’s
product sales
come from
recommendation
engines
•As marketing becomes a revenue driver, companies that have implemented CRM
and marketing automation are looking for new insights to take their modern marketing
efforts to the next level.
•Predictive marketing works by taking all the data in the world about accounts and
prospects — from both internal and external sources — and applying modern data
science to optimize conversions of all stages of the revenue funnel, and tackle other
top-of-mind challenges.
•Modern-day data science makes it easy for companies of all sizes to use the same
techniques Internet giants use to develop recommendations.
9
s
s
s
s
35%
What Insights Are Hidden in
Your Marketing Automation
and CRM?
As the adoption of CRM and marketing automation matures, it’s no surprise that
companies with these technologies are sitting on a wealth of data. Out of the box, CRM
and marketing automation are designed to capture and store a rich set of information on
customers and prospects. Many companies add additional customizations and integrations to turn these systems into robust marketing and sales data warehouses.
10
s
s
s
s
Good sales reps are already pouring through CRM to review contact and account
information prior to dialing a prospect. They are searching for opportunities created,
looking at recently won deals and surveying lost opportunities. Using robust APIs,
companies are connecting CRM to internal systems that provide product trial
and usage data and customer support data. They’re also mining social networks, looking for clues into buyer needs and trying to find connections.
While this behavior is effective, it is time not spent closing deals. The
best reps know how to take advantage of this full set of information to
determine whether or not a prospect is ready to be engaged. They want
to drive productivity by focusing their time on the highestvalue leads.
At the same time, marketers are creating laser-targeted campaigns within their marketing automation
platforms based on lead demographics and behavior. They have customized nurture streams for buyers at various stages of the funnel or by vertical, persona or product interest. Savvy demand-generation
teams are also enriching their leads with third-party data. Their goal is to target the right prospect with
the right message, at the right time and then pass them off to sales when they are most likely to convert.
Data-driven marketers are maniacal about measurement. They are keen to understand the attributes in
marketing automation that indicate a lead is ready to buy and is, therefore, ready to be passed to sales.
Let’s take a look at some sample positive and negative predictors of buying intent that
can be found within the CRM and marketing automation:
Sample Attributes of Buyer Intent from CRM
Existing customer
Product usage data
Trial products and trial dates
Customer support cases
11
s
s
s
s
Lost opportunities
Sample Predictors of Buyer Intent from Marketing Automation
Email response and opens
Content engagement
Website visits
Event participation
Webinar attendance
These are just some of the attributes that can carry predictive value. Companies continue to enhance the
breadth and richness of data they store in CRM and marketing automation. Predictive marketing can
leverage this rich data to help marketers pass on the most lucrative leads to sales.
Key Takeaways
•A wealth of data that is predictive of buyer intent is hidden in CRM and marketing
automation.
•Predictive marketing can turbo-charge your CRM and marketing automation efforts
to highlight the most sales-ready leads.
12
s
s
s
s
Understanding Predictive Models
Not All Predictive Models are Created Equal
Two key ingredients are required for efficient, highly predictive models ­— data and analytics. While the
data is crucial, the algorithms and analytics behind the predictive models are the engines that do most of
the heavy lifting and differentiate good predictions from great predictions.
Determining
the correct mix
of attributes for
inclusion
DATA NORMALIZATION
Ensuring that each
attribute maximizes the
contribution to
the model
PREDICTIVE MODELING
Mining the data to
“fingerprint” what
makes a good lead
or prospect
13
s
s
s
s
FEATURE SELECTION
FEATURE SELECTION Statistical models perform best when they incorporate
DATA NORMALIZATION While data represents a key input into any predictive
algorithm, it can take many shapes and forms. Some attributes like “number of email
opens” or “annual spend” are relatively straightforward to mine, whereas attributes like
“job title” or “geography” need pre-processing before they can truly shine.
MODEL EXECUTION The real value of machine learning comes out when the
models are finally selected and launched.
While data
represents a
key input into
any predictive
algorithm, it
can take many
shapes and
forms.
—Matt Pollock, VP,
Lattice Engines
14
s
s
s
s
the most optimized set of attributes (or “features”). It is typical to have thousands
of candidate attributes that could potentially be included in the pattern-matching
algorithms. To start, it is common to apply various statistical techniques to determine
which attributes should be retained and which should be discarded. Many predictive
models also look at the creation of derived attributes, which transform the raw data
in a native attribute into a form that is more meaningful in a predictive model. For
example, the founding date of a company is used as the basis for a derived attribute
called “company age” that is likely far more predictive than founding date.
Here is a sample of the techniques that can be used in predictive models.
LOGISTIC REGRESSION is a type of regression analysis used for predicting the outcome
of a categorical dependent variable. Logistic regression is very resource-intensive, consuming
a great deal of memory on a large data set; however, it is very stable and works particularly well
when you have continuous features or attributes like revenue data.
DECISION TREES are very powerful algorithms that help identify the best predictors.
Decision trees are intuitive to analyze and usually produce great results when applied to a
mixture of categorical (i.e. SIC CODE, industry vertical, location) and numerical attributes.
RANDOM FOREST is one of the techniques behind the recommendation engine in Netflix and
also a popular technique in the Hadoop framework. The main idea is to build a forest of many
decision trees over different variations of the same data set and take weighted averages of the
results. This technique is very powerful because it can effectively identify patterns across a large
noisy dataset. The technique is computationally expensive, but it can be easily run in parallel.
NEURAL NETWORKS are a composition of neurons combined together to describe a
data set. While machine-intensive, it is very powerful when you try to describe events that
are non-linear (for instance, a sales campaign that spans across multiple market segments).
Neural networks are typically used to identify very complex patterns.
K-MEANS CLASSIFICATION CLUSTERING can be very useful for prospecting. For example, take
the numerous existing customers and potential prospects in your CRM software. Clustering allows you
to find similarities between accounts and rank them according to the degree of similarity.
NAÏVE BAYES is a probabilistic classifier. It is very useful in identifying patterns and behaviors of an
15
s
s
s
s
account for cross- and upsell purposes. For example, this account bought products A and B, so the
probability of buying C is very high.
Top Considerations For Marketers Evaluating Predictive Apps
There has been an explosion of interest in using machine learning to build models to predict customer
behavior, and many companies offer products and solutions in this domain. As you go through the
journey to predictive marketing, you’ll want to keep the following considerations in mind.
2
3
Ensure you have the right data assets to solve your problem.
Organizing data can be a daunting task. It may seem like you can get an acceptable
answer with a small subset of the rows or columns, but you’ll want to ensure that you
have enough data so that the results are not misleading.
Understand what success looks like for your company.
Identifying success for any type of predictive model typically consists of weighing all of the
various factors of the model into a single “model quality factor.” Here are some examples of
how such model quality could be defined:
• The lift in the top 10 percent of scored recommendations
•The difference in the lift in the top 10 percent of scored recommendations and the
percentage of recommendations that fall in the lowest 30 percent
•The order value for the top 20 percent of recommendations
16
s
s
s
s
1
Think about the problem you are trying to solve.
If you don’t have a particular decision or marketing problem you are trying to solve in mind,
then it may not be time to investigate predictive marketing and analytics. Analytical models
are all built with some set of assumptions about what they are trying to predict and understanding the problem you are trying to solve will help ensure that you get value from
predictive modeling.
4
Think through what you will do with the output.
Knowing what you want to do with the output is critical from the start. You may uncover
insights that ultimately change a well-established process within your organization. Change
management is often an after-thought but should be considered at the beginning of any
predictive marketing initiative. Consider planning out communication or training schedules
if the need arises.
Data science has an enormous potential to bring immediate and significant impact to the
performance of a wide variety of business problems. Vendors in this space bring expertise,
software and data that can accelerate the impact of this trend on your business. It’s not simply enough to have a broad, high-quality set of buying signals or have a single, great predictive algorithm.
Key Takeaways
•Understand the problem you are trying to solve first. This ultimately helps inform the
model and ensures a higher rate of success.
•The data and the model matter. Some pre-processing work may be necessary in order
for predictive marketing to work.
•Understand what success looks like before you get started.
17
s
s
s
s
Lead Scoring
Numerous aspects of marketing could be vastly improved with better predictive
insights, but many marketers are finding predictive lead scoring is the best place
to start.
Why?
Many marketers have found their current lead scoring initiatives have failed to live up
to their expectations. According to a report from Decision Tree Labs, 44 percent of
companies using marketing automation have implemented lead scoring. However, on
average, survey respondents graded their lead scoring programs five out of 10. Why?
Most commonly, it’s a simple lack of good insight into what constitutes actual buying behavior. In fact, SiriusDecision reports that 94 percent of MQLs will never close
despite all of our efforts.
18
s
s
s
s
It’s no wonder marketers can increase their conversion rates only so much by
making do with the current crop of rules-based lead scoring engines. Traditional
lead scoring prioritizes leads based on various fit and behavior criteria in hopes of
getting a picture of a good lead. However, this approach taps into just a small percentage of data that could be gleaned from prospects and requires a heavy dose
of gut instinct and intuition.
Marketers are forced to make critical decisions about passing to sales based on a limited amount of
information. In a sense, this basic lead scoring is little more than a guessing game. As a result, many
marketers struggle to demonstrate tangible return on its investment in marketing automation.
A better option is to tap into the power of predictive lead scoring. This advanced lead scoring approach
augments the demographic and behavioral attributes that are part of basic lead scoring with thousands
of additional data points. Examples include whether the company in question recently received funding,
moved to a new location or hired new design engineers. In essence, predictive lead scoring empowers
marketers to build a sophisticated model that actually predicts which lead attributes matter most. This
approach allows them to:
Combine contact- and account-level attributes to get a complete
360-degree view of all buying signals — not just those captured in
marketing automation.
Uncover the true definition of a good lead through the
use of data science rather than intuition or consensus.
19
s
s
s
s
Determine the actual probability of each prospect
becoming a customer with unmatched precision.
Weaving Predictive Lead Scoring into the Buyer Journey
It’s important that any predictive tool fits into marketing’s existing workflow and tool set. Regardless of
how an organization views its revenue funnel, the key is to apply predictive lead scoring at each crucial
conversion point — especially the critical hand-off between marketing and sales.
The first conversion occurs when marketing passes a qualified lead to inside sales to further qualify and
accept it as a sales-ready lead (MQL -> SQL). With predictive lead scoring, marketers can be assured
they are only passing sales the contacts who are most likely to buy.
Traditional Lead Scoring versus Predictive Lead Scoring
A, B, C 10%, 30%, 40%
Let’s compare the probability to convert.
The traditional lead score doesn’t explain the difference in quality between
leads, where as the percentage is very clear and provides far better
prioritization.
20
s
s
s
s
As a result, sales will no longer waste time trying to track down and qualify contacts who would be better
served by a nurture program until they are actually ready to purchase.
→ THE DEMAND WATERFALL
The second conversion point
happens when the sales team
is tasked with qualifying a huge
volume of leads (SAL -> SQL).
Without a solid mechanism for
deciding where to focus, sales
either randomly follows up with
leads or cherry-picks leads based
on a very unscientific process.
Because of the time required to
contact so many leads, often a
large percentage of good leads
fall through the cracks while sales
is spinning its wheels with the bad
ones. If marketing can tell the
team how likely a given lead is
to convert, the sales reps can
prioritize their efforts using
science rather than chance.
The Demand Waterfall
From SiriusDecisions
INQUIRY
INBOUND
OUTBOUND
MARKETING QUALIFICATION
AUTOMATION QUALIFIED LEADS
TELEPROSPECTING ACCEPTED LEADS
TELEPROSPECTING
QUALIFIED LEADS
TELEPROSPECTING
GENERATED LEADS
SALES QUALIFICATION
SALES GENERATED
LEADS
SALES ACCEPTED
LEADS
SALES QUALIFIED LEADS
CLOSE
WON BUSINESS
21
s
s
s
s
Source: SiriusDecisions
Here is a quick look at how marketers can make predictive marketing actionable.
Different contact strategy by segment
40%
Probability to Convert
35%
30%
send to
sales/BDrs
25%
Predicted
20%
15%
send to
nurture
10%
Average
5%
0%
Accounts/Leads
Accounts
takeaways:
KeyKey
Takeaways
According to Demandgen report, most marketers are unhappy with their current
approach
to leadlead
scoring.
• rules-based
scoring techniques typically only account for one to five percent
of data available
on a prospect.
• Rules-based
lead scoring
techniques typically only account for one to five percent of
data
available on
prospect.
• Predictive
leadascoring
empowers marketers to build a sophisticated model that
actuallylead
predicts
which
lead attributes
matterto
most,
on uncovering
true that
• Predictive
scoring
empowers
marketers
buildbased
a sophisticated
model
buying
signalswhich
from internal
and external
data.most, based on uncovering true buying
actually
predicts
lead attributes
matter
signals from internal and external data.
s
s
18
22
s
s
s
s
•
• According
to to
Decision
Tree Labs, most marketers are unhappy with their current
approach
lead scoring.
s
s
Account-Based Marketing:
The Missed Opportunity?
For years, B2B marketing tactics have focused largely on individuals, as the majority
of channels such as email, phone and even events target at the individual level.
However, as technologies have evolved, the idea of account-based marketing
(ABM) has really drawn a great deal of appeal. After all, most B2B buying decisions
happen because of a company need and typically involve an entire team of participants in the buying process.
Unfortunately, one of the greatest assets in the demand generation technology
stack has emerged as a barrier to ABM. Marketing automation tools were fundamentally built around the concept of a contact, rather than an account. Even with
the account object and account-level attributes that can be stored within marketing
automation, marketers are fairly limited in terms of the data collection, scoring tools
and segmentation options they can rely on for creating account-level campaigns.
23
s
s
s
s
Recent improvements have come about to roll lead scoring up to the account level.
Some marketers are using third-party providers to help with data appending to
add better firmographic insights to help with segmentation. However, true
account-level scoring and more complex filtering and segmentation capabilities are
still lacking. Missing the ability to store and track richer, more dynamic account-level
data means marketing automation simply isn’t the ideal solution to advance the
ABM cause.
So Why the Sudden Interest in Account-Based Marketing?
For one, many of the newer, more innovative marketing channels rely on ABM to make them
effective. ABM can target many or all employees within a given account with personalization, display
ads, product-level campaigns and field marketing events. For example, with tools like Demandbase you
can target your content dynamically, based on the inbound company’s IP address. This creates powerful
messaging that is far more relevant if you know which account you are going after.
You can also dramatically reduce your display ad spending by ignoring accounts that aren’t relevant
to your business. For example, if you know a prospect account uses Salesforce.com, you can display
an ad that is relevant to that audience, whereas a visitor from an account using Siebel or Microsoft
Dynamics CRM would not be targeted. This increases effectiveness while also reducing costs — a
perfect storm of marketing effectiveness.
ABM provides an opportunity to fuel growth from marketing.
Key Takeaways
• As
B2B marketers, we need to remember that we sell to both people and companies.
• Many
existing marketing technologies are focused on contacts, rather than accounts.
• Account-based
marketing is critical when thinking about retaining or growing existing
84%
of marketers
noted that
ABM provided
significant
benefits to
retaining or
expanding
existing
customers
—Marketo
24
s
s
s
s
customer accounts.
Account Scoring
In a B2B environment, people generally don’t buy products or services their
company doesn’t need. In reality, it’s a combination of events and actions that sparks
a purchase decision.
For most considered to be in B2B buying cycles, the process begins with some kind of
trigger event within the company, followed by a reaction from a person or team to look for
a solution. The company need dictates a human-lead buying process. For example, a new
round of funding for a business might lead to an office expansion necessitating a slew of
different buying cycles for anything from office furniture to networking equipment.
So What Does All this Have to Do
with Lead Scoring? Quite simply, the standard practice of scoring purely at the contact level misses the much
bigger picture. Yet ignoring the behaviors of individuals within that organization also limits
your perspective. Only by blending the two will you get a complete 360-degree view of
your prospect’s buying signals.
25
s
s
s
s
When most companies start investigating account scoring, they are often building it by
using some limited firmographic data along with contact-level attributes. The most common approach is to aggregate or average the scores of individuals associated with the
account, but this is really just cumulative contact scoring ­— not true account scoring.
The Power of Blended Scoring
Most marketers look at accounts and contacts separately, and at best can create an account score that is
an aggregate of contact scores. By taking this blended approach, marketing can much more accurately
predict which leads to pass to sales, get signals much earlier in the buying process and ultimately create
much better alignment between marketing and sales.
Key Takeaways
• A contact level-only approach doesn’t account for the full picture.
• Smart
marketers are taking a blended approach to lead scoring, which combines
•
contact with account-level attributes.
Growth
trends, hiring, funding and technology usage are all sample account-level
attributes that may be predictive of buyer intent.
Marketers
need to
remember to
look beyond the
contact. There
are a ton of great
insights you can
learn at the
account-level
that can be used
to target
campaigns
and drive
more sales.
—Jon Miller,
VP Marketing
and
Co-Founder,
Marketo
26
s
s
s
s
Just like individuals, companies also exhibit digital body language. For instance, firmographic data may
tell you a company fits the right industry profile or size. What most marketers miss are the account-level
buying signals such as growth trends, hiring patterns, government grants, patent filings or technology
usage, just to name a few. These account-level activities are often the earliest buying signals, possibly
preceding contact-level activities by weeks, months or even years.
Customer Expansion
Improve Marketing Performance by Targeting
Existing Customers
Companies have improved their marketing performance through the adoption of B2B
marketing automation to reach and attract new prospects. Despite these tools and
technologies, very few marketing teams have applied them with the same type of rigor
for actually retaining and expanding customer relationships.
According to a recent survey by the DemandGen Report:
The act of managing multiple, disparate systems was
cited as the top obstacle to achieving customer marketing
goals followed by insufficient data.
27
s
s
s
s
In most cases, the sales or service groups still have exclusive ownership of expansion
opportunities. With such roadblocks, it’s no wonder marketing is so laser focused on
new customer acquisition.
A Great Opportunity to Improve Marketing Performance
Hiding in Plain Sight
For most organizations, existing customers drive 50 percent or more of revenue. Given the limited
capacity of individual sales reps, they must make instinctive bets about which accounts and products to focus
on. Compounding the problem, reps also tend to gravitate toward the products and messages they are most
comfortable with, meaning that many newer products, services or messaging get limited attention. So what
could these missed opportunities be costing?
New Customer
Acquisition
Existing
Customers
Churn/Attrition
Customer Marketing Opportunity
Unsell/Cross-sell
Traditional Demand Gen Opportunity
28
s
s
s
s
• Effective cross selling and upselling drive significant acceleration in both
revenue growth and renewal rates.
• Customer revenue potential is often three to five times larger than current sales
opportunities.
• Closing a new customer can cost between three to five times as much as retaining
or expanding an existing customer.
Revenue Contribution
Why is Customer Marketing Often Ignored?
Marketing automation platforms are fundamentally designed around customer acquisition. It’s as simple
as that. Features like prospect profiling, segmentation, event management and web analytics are often
tuned to gradually collect more information about leads or prospects, but lack capabilities for mining
existing customer information hidden in plain sight.
Customer Buying Signals are Hidden in Plain Sight
A survey conducted by DemandGen Report and Retail TouchPoints revealed that capturing and
integrating customer data is a key consideration for marketers, with more than 49 percent identifying it
as a top priority. However, some of the most valuable data doesn’t come from marketing automation
or CRM at all, but rather from transactional systems such as order management, call center or support
log ­— a rich data source unique to existing customers.
49%
of marketers
identify
capturing and
integrating
customer
data is a key
consideration
29
s
s
s
s
Beyond just activity history, marketers also need to look externally at account-level indicators such as
hiring trends, office openings, funding events or even social activity for hidden buying signals that could
represent good triggers for cross-sell or upsell.
Predictive Analytics Illuminate the Path to Improved
Marketing Performance
By combining contact- and account-level attributes, marketers can get a full view of the customer and
apply predictive analytics to identify not just the best opportunities for upsell or cross-sell, but also which
products or services represent the best fit. For organizations with complex product and customer matrices, arming sales with the right targets, the right products and right messaging can finally unlock the full
potential of that 50 percent customer opportunity.
Marketing
Automation
Purchase
History
CRM
External Data
30
s
s
s
s
BUYING SIGNALS FOR EXISTING CUSTOMERS ARE FRAGMENTED
Marketing no longer has the luxury of focusing exclusively on customer acquisition. With the right
combination of data and predictive analytics, marketers can offer sales reps higher productivity and
product leaders the proper attention on the full breadth of product and service offerings.
B
y tapping into customer expansion, marketing performance
can increase and help source a much larger share of total company
revenue ­— the other 50 percent.
—Brian Kardon, CMO, Lattice Engines
Key Takeaways
• In most companies, existing customers drive 50 percent or more of the revenue.
• Marketing automation is inherently designed to go after new customers.
• Sales reps typically gravitate toward the products and messages they are most
comfortable with, meaning that many newer products, services or messaging
get limited attention.
• A predictive marketing approach can help marketers identify opportunities for
customer expansion through account-level targeting.
31
s
s
s
s
Conclusion
There has been an explosion of interest in predictive marketing
and using machine learning to build models to predict customer
behavior. Data science has an enormous potential to bring immediate
and significant impact to the performance of a wide-variety of B2B
marketing problems, helping marketers go beyond modern marketing.
Still curious about predictive marketing? Contact us today.
About Lattice Engines
32
s
s
s
s
Lattice is pioneering the predictive applications market for marketing and sales.
Lattice helps companies grow revenue across the entire customer lifecycle with
data-driven marketing and sales applications that make complex data science easy
to use. By combining thousands of relevant buying signals with advanced predictive analytics in a suite of secure cloud applications, Lattice helps companies of all
sizes to stop guessing and start relying on predictive insights to increase conversion
rates and deal sizes by more than three times. Lattice is backed by NEA and
Sequoia Capital with headquarters in San Mateo, CA. Learn more at
www.lattice-engines.com and follow @Lattice_Engines.