[Book Review] An Essential Guide to Artificial Intelligence for Marketers

Source:  Trust Insights

The marketing world has been going gaga over ChatGPT for the past four months. 

The generative AI application from OpenAI was released to the public on November 30, 2022, and analysts have estimated that it reached 100 million monthly active users in January of this year.

The capabilities of ChatGPT have amazed many users, including me. More importantly, the buzz surrounding ChatGPT has ignited an arms race among tech companies to develop and roll out new or enhanced applications enabled by artificial intelligence. 

For example, Microsoft - which had already invested $1 billion in OpenAI - recently confirmed that it will invest an additional $10 billion in the company, and announced that it is incorporating some ChatGPT-like functionality in its Bing search engine. Google, Salesforce, Hubspot, and a host of other firms have also recently announced new or enhanced applications featuring generative AI capabilities.

The application of artificial intelligence to marketing isn't new, but the intense reaction to ChatGPT by both the public and tech companies suggests that we are on the cusp of a step change in the use of AI in marketing.

Until now, many marketers have been able to function effectively with only a superficial understanding of artificial intelligence. But, as AI becomes increasingly integral to marketing at more and more companies, it will be imperative for marketers to have a better grasp of AI principles and techniques.

Christopher S. Penn, the co-founder and Chief Data Scientist of Trust Insights, and a recognized authority on analytics, data science, and machine learning, has written a book that will help marketers begin their journey toward a better understanding of artificial intelligence.

AI For Marketers:  An Introduction and Primer (Third Edition, 2021) provides a sound introduction to basic AI techniques and illustrates how AI can be used to improve marketing performance.

What's In the Book

AI For Marketers contains 18 chapters, but the book's content falls into four broad topic categories.

In the opening two chapters, Penn discusses the importance of AI in marketing and explains why marketers often struggle with AI. He argues that one reason marketers find AI challenging is that, " . . . AI and its prerequisites are deeply entrenched in mathematics and statistics - two fields which are not strong points for most marketers." (Emphasis in original)

Penn devotes four chapters to an explanation of the basic techniques of artificial intelligence. He defines AI and explains algorithms, models, and types of machine learning. He also covers the vital importance of good data and provides a useful data quality framework.

The third major topic category in AI For Marketers is a description of several practical applications of AI in marketing. Penn places his discussion in the context of problems that AI can help marketers address. For example, he explains how attribution analysis can help marketers forecast what strategies, tactics, and tools will deliver the best results, and how dimension reduction and feature selection can help marketers identify which data points are important.

Lastly, Penn discusses how companies can successfully incorporate artificial intelligence in their marketing efforts, and how marketers can prepare their careers for AI. He lays out seven steps that describe the process of becoming an "AI-first company," and he covers the people and process governance capabilities that companies need to be successful with AI.

My Take

Writing a book about the use of artificial intelligence in marketing is a daunting task because the field is evolving so rapidly that a book can easily become outdated soon after it's published.

Writing a book about AI for marketers is even more challenging because most marketers have little, if any, education or training in statistics or computer science, both of which are essential components of artificial intelligence.

Christopher Penn does an admirable job of addressing both of these challenges in AI For Marketers. He focuses on the core fundamentals of artificial intelligence and on the basic applications of AI in marketing. He also explains AI concepts and techniques in an informal, easy-to-understand way, making the subject accessible to marketers who haven't been trained in statistics and computer science.

In the book, Penn argues that marketers don't need to become practitioners of AI in the sense of learning statistics, data science, and machine learning. He uses the analogy of "chefs and farmers" to illustrate his argument. He wrote:

"Talented chefs take great ingredients and, using the right tools and skills, transform those ingredients into delicious food . . . However, what's the likelihood that the chef is also a farmer . . . Almost none . . . [Chefs] may have some sense of what's gone into an ingredient, but they're not the ones to focus on the details of the ingredient's creation . . .

The typical outcome of an artificial intelligence platform is a model that creates insights or makes decisions. The software . . . plugs into our marketing infrastructure and spits out highly refined products from the raw ingredients - data, algorithms, and analyses. The machines are the farmers, and we are the chefs." (Emphasis in original)

I would contend that Penn's argument goes a little too far. I believe that many marketers - particularly those in more senior roles - will need to delve a little deeper into artificial intelligence than Penn suggests.

Penn wrote that marketers ". . . should know what great data, algorithms, models, or decisions look like . . ." It's difficult - probably impossible - to determine whether an algorithm is "great" and fit for purpose if you don't know how the algorithm works and what its strengths and limitations are. While AI For Marketers is a solid introduction to the topic, it doesn't go quite deep enough to provide this level of information.

Even with this one caveat, I strongly recommend Chris Penn's book. AI For Marketers is a great resource for marketers who are beginning their journey toward a greater understanding of artificial intelligence and its expanding role in marketing.

One final point. The third edition of AI For Marketers was published in 2021 and therefore doesn't capture the tsunami of developments in AI that have occurred over the past few months. I subscribe to Chris Penn's Almost Timely Newsletter, and he has already provided numerous valuable insights regarding the recent developments in AI.

I have no inside information, but I suspect that a fourth edition of AI For Marketers is already in the works. In other circumstances, I might recommend that you wait a bit for the new edition of the book. But artificial intelligence is poised to become so important for marketers that I think you should read the third edition now and be prepared to read the fourth edition when it appears. 

The Promise and Perils of Focusing on "In-Market" Prospects

Intent data and predictive analytics have been hot topics in B2B marketing circles for the past few years. Simply put, intent data is information collected about the online activities of a person with the goal of using that data to identify or predict purchase intent. To make this prediction, intent data is processed using a software application with predictive analytics functionality.

Some providers of intent data and/or predictive analytics capabilities have been rather effusive in describing the benefits of their solutions. Consider, for example, these two blog post passages from firms operating in the intent data/predictive analytics space:

    "To avoid wasting time and money pursuing prospects that either already just bought the product from your competitor or are not serious about buying yet, your team should focus on the right people, targeting them at the right time by leveraging intent data, which will help you understand total active demand. Instead of a broad market of generic buyer personas, it enables you to find specific accounts that are active in your market."

    "The opportunity represented by intent data is obvious:  find in-market buyers before they enter the funnel by tracking their online behavior and content consumption on different websites. Get enough of a head start and you can land a deal before they even consider your competition, shorten your sales cycle, and cut your customer acquisition costs."

This is heady stuff because the ability to know which prospects are engaged in an active buying process could enable fundamental changes in the practice of B2B marketing.

The Promise

For example, suppose that your company has implemented account-based marketing. With intent data and predictive analytics, you could select ABM target accounts based on both fit (how well a prospect matches your ideal customer profile) and interest (whether a prospect is "in-market"). You could also frequently update your list of target accounts so that you have a near real-time view of which accounts are engaged in an active buying process.

This sounds like marketing nirvana, right? When you know which of your prospects are in-market, you can focus your marketing programs on this "low-hanging fruit," which should result in higher conversion rates, increased marketing efficiency, and as the blog passage says, lower customer acquisition costs.

The Perils

It's clear that some companies can reap substantial benefits from using intent data and predictive analytics in their marketing efforts. But intent data still has some important limitations that marketers need to understand. Those limitations have been widely discussed in articles and blog posts. For example:

• What Is Intent Data & How Can You Use It?
• How to Assess the Relevancy of 3rd Party Intent Data
• Intent Data is Great. Except When it Isn't

Using intent data and predictive analytics to focus marketing efforts on in-market prospects also presents a broader hazard. If taken to the extreme, it can lead marketers to ignore prospects that don't make the in-market cut. This is dangerous because it disregards an important aspect of how business buyers make buying decisions.

The conventional view is that a B2B buying process begins when a company's leaders or managers recognize a need or a problem and decide to do something about it. These "buyers" then gather information about the need or problem, evaluate possible solutions, and may or may not decide to buy a product or service to address the need or problem. So our traditional view of B2B buying is that information gathering, learning, and evaluation all occur after an intentional buying process is underway.

But business decision makers rarely begin a buying process with a clean slate. Every day, they are forming impressions of companies, brands, and products from touch points like ads, content resources, news reports, and conversations with business colleagues and friends. When something triggers an intentional buying process, these accumulated impressions become pivotal because they shape the initial consideration set.

The initial consideration set contains those companies and/or solutions that business decision makers immediately think of when they're faced with the potential need to buy something. And being included in the initial consideration set really matters. Research by McKinsey in the B2C space has found that brands in the initial consideration set can be up to three times more likely to be purchased than brands that aren't in it.

I suspect the impact is slightly less in B2B, but being part of the initial consideration set is still important because it all but guarantees that your solution will be one of those evaluated in the formal buying process. And, you have to be invited to the party before you can be asked to dance.

The importance of being in the initial consideration set explains why it would be a mistake for most B2B companies to focus their marketing efforts exclusively on in-market prospects.

At any given moment in time, a large majority of your most attractive prospects - those with high potential value and good fit - will not be engaged in an active buying process and would not qualify as being "in-market." These attractive prospects may not be likely to buy in the near term, but that doesn't mean they are unlikely to buy in the longer term.

If you focus your marketing efforts solely on in-market prospects, you'll be abandoning the opportunity to influence the perceptions and preferences of high-value future buyers.

Image courtesy of Fertile Ground via Flickr CC.

Why You Need to Be Careful With One Feature of the New Demand Waterfall

The new SiriusDecisions Demand Unit Waterfall has received lots of accolades since its introduction this spring, and the accolades are richly deserved. But one of the new demand stages should be labeled Use With Caution.

In May, SiriusDecisions unveiled the latest iteration of it venerable Demand Waterfall with great fanfare. SiriusDecisions calls the new version the Demand Unit Waterfall, and it's depicted in the following diagram:

Source:  SiriusDecisions

The first version of the Demand Waterfall was introduced in 2006, and over the past decade, thousands of B2B companies have used the waterfall model to track and manage their demand generation efforts. So it shouldn't be surprising that the introduction of the Demand Unit Waterfall has generated quite a bit of interest in the B2B marketing world. Here's a small sample of the reactions from pundits and practitioners:

• What's The Impact Of The New Demand Unit Waterfall? Hear From 30 B2B Experts (Demand Gen Report)
• How We've Implemented (Parts of) The Demand Unit Waterfall for ABM (Tony Yang, Mintigo)
• Diving into the new SiriusDecisions Demand Unit Waterfall (John Steinert, MarTech Today)
• SiriusDecisions:  The New Demand Unit Waterfall (John Common, Intelligent Demand)

For an in-depth discussion of the new Demand Unit Waterfall, I recommend that you watch the recording of this SiriusDecisions webinar.

The early reaction to the Demand Unit Waterfall has been overwhelmingly positive, and I agree with those who say that it more accurately reflects the realities of B2B demand generation.

• By focusing on demand units, the new waterfall recognizes that most B2B buying decisions are made by groups of people, not by individuals.
• By eliminating the waterfall stages that focused on the sources of individual leads and on the "ownership" of demand generation activities, the new waterfall implicitly recognizes that demand generation has become a team sport that involves marketing, business development, and sales throughout the whole process.

Use Caution With "Active Demand"

My biggest reservation about the new Demand Unit Waterfall is the possible implications of the Active Demand stage. SiriusDecisions defines Active Demand as the demand units that are showing evidence of acute need or buying intention. In other words, Active Demand refers to demand units that are currently "in-market" for the type of solution you sell.

SiriusDecisions analysts are suggesting that companies should focus their outbound demand generation activities on in-market prospects. Not surprisingly, this is the approach also advocated by many providers of B2B predictive analytics software.

This approach may work for some types of B2B companies, but it won't work for all. Here's why.

The identification of in-market prospects relies heavily on the use of intent data, which is data regarding the online behaviors of potential buyers. Intent data - particularly third-party intent data - can be valuable for some types of B2B companies, but it can be almost useless for others. To learn more about the limitations of intent data, take a look at this post by Jingcong Zhao at the Marketo blog, this post by Todd Berkowitz at the Gartner blog, and this white paper by Infer.

The bottom line is, business and marketing leaders should be cautious about relying on their ability to accurately identify in-market buyers. And they should be particularly cautious about focusing all - or even most - of their marketing efforts on such buyers. I've discussed this issue in two earlier posts. If you'd like to see my view on this issue, take a look at Why B2B Marketers Need to Care About "Casual Learning" and Why Marketers Shouldn't Go All In on In-Market Buyers.

Have We Really Improved Marketing Productivity?

The recent pace of change in B2B marketing has been nothing short of breathtaking. Over the past 10-15 years, new marketing technologies, channels, and techniques have appeared in rapid succession, and many of these innovations are now in widespread use. B2B marketing automation, content marketing, inbound marketing, and social media marketing are just of few of the technologies and techniques that have changed B2B marketing over the past decade or so.

By all indications, the pace of change is not slowing. During the past couple of years, many B2B companies have adopted account-based marketing, and many have begun using predictive marketing analytics technologies to support ABM and other marketing efforts. And just within the past few months, we've started to hear that machine learning and artificial intelligence will have a major impact on B2B marketing in the near future.

All of these innovations have promised to improve marketing effectiveness and efficiency, and numerous research studies purport to show that they are delivering a wide range of benefits. But have these innovations really improved the bottom-line productivity of B2B marketing? Can we show - in a credible and convincing way - that B2B marketing is more financially productive today than it was 10 or 15 years ago?

Obviously, these questions must be answered on a company-by-company basis. Some B2B marketers may be able to show that their marketing efforts have become significantly more productive over the past several years. But there is evidence suggesting that some aspects of B2B marketing performance haven't improved as much as we might have anticipated.

One indicator of B2B marketing and sales productivity is the efficiency of the demand generation process. Efficiency is usually measured by the percentage of potential buyers or leads who are "converting" from one lead stage to the next.

Many B2B companies use the Demand Waterfall model developed by SiriusDecisions to describe the stages of the lead-to-revenue process, and from time to time, SiriusDecisions publishes "average" and "best-in-class" conversion rates that link to the Demand Waterfall. The following table shows the conversion rates reported by SiriusDecisions for 2008 and 2014:

What is most striking about this data is that it indicates there was essentially no improvement in conversion rates - particularly the overall lead-to-revenue conversion rate - between 2008 and 2014.

The 2008 conversion rates largely reflect marketing productivity before many of the marketing innovations mentioned above had become widely adopted. But research has shown that by 2014, a significant number of companies were using these technologies and techniques.

• 60% of large companies (more than $500 million in annual revenue) had acquired a marketing automation solution. (Raab Associates)
• 93% of B2B marketers were using content marketing in some form. (Content Marketing Institute/MarketingProfs)
• 86% of companies were running inbound marketing programs. (Hubspot)

Of course, lead conversion rates aren't the only relevant measure of marketing productivity, and there may be a reasonable explanation for the lack of improvement shown in the SiriusDecisions data. For example, the 2014 conversion rates would not have captured the impact of the shift to account-based marketing that's occurred over the past couple of years. Nevertheless, this data should be a wake-up call for B2B marketers.

Senior company leaders are increasingly expecting marketers to demonstrate that their activities and programs are creating economic value for the enterprise and improving enterprise financial performance. Many senior leaders are no longer satisfied with the tactical performance indicators (campaign response rates, content downloads, etc.) that marketers have traditionally used to describe marketing performance. What senior business leaders really want to see is proof that marketing is delivering financial results and that the dollars they are investing in marketing are being spent as efficiently as possible.

The important point here is that the value of any marketing technology or method must ultimately be judged by whether its use improves marketing productivity. So that's what marketers must be prepared to demonstrate.

Top image courtesy of Kelly Teague via Flickr CC.

The Promise and Peril of Personalization at Scale

Marketers in virtually all kinds of companies are now intensely focused on improving the quality of the customer experiences their companies provide. Most marketers believe that the ability to personalize marketing offers and messages for individual customers at every touch point is critical to delivering outstanding customer experiences.

Numerous research studies have shown that personalized marketing can be highly effective. For example, in a 2016 survey of more than 1,500 US and UK consumers by Accenture Interactive, almost two-thirds (65%) of respondents said they are more likely to make a purchase from a retailer that sends them relevant and personalized offers. But there are also many well-documented examples of personalized marketing efforts that have failed miserably because they were clumsily executed, or because they were seen as a "creepy" invasion of privacy.

The ability to "personalize at scale" is highly dependent on data and technology. Many companies are already using data and predictive analytics technologies to automatically generate personalized marketing messages without human involvement, and many marketers believe that this practice will only become more commonplace. In fact, it's difficult to envision how personalization at scale can be achieved without a heavy reliance on data and technology.

But this creates a conundrum for marketers that I contend isn't fully appreciated. Automated, data-driven personalization will make personalization at scale more achievable, but it also increases the danger of getting personalization wrong.

An article by Charles Duhigg in The New York Times Magazine provides a great illustration of both the promise and the peril of automated, data-driven personalization. In this article, Mr. Duhigg describes how personalized marketing caused an unwitting father to discover that his teenage daughter was pregnant.

When I decided to write about this topic, my intention was to briefly summarize the content of Mr. Duhigg's article, but I quickly determined that a brief summary simply would not do justice to the subject matter of the article. I strongly recommend that you take the time to read the entire article if you are involved in developing personalized marketing programs.

For me, there are three important takeaways from the article.

• First, data and predictive analytics can enable marketers to develop marketing messages and offers that are highly relevant for individual customers and prospects. 
• Second, relevance alone is not enough to ensure that a personalized marketing program will be successful. It takes sound human judgment to evaluate factors that data and analytics simply cannot capture. 
• And third, sometimes it is more effective to make marketing messages less personalized, particularly when the subject matter touches a highly personal or otherwise sensitive topic.

(Note:  Mr. Duhigg's article is fairly long. The beginning and ending portions of the article contain the material that pertains directly to personalized marketing. The middle part of the article focuses on the power of habit in human decision making. The entire article is well worth reading.)

Image courtesy of Josh Hallett via Flickr CC.

A Much-Needed Reality Check on Predictive Analytics

Predictive analytics has become one of the hottest topics in B2B marketing over the past several months. In a survey last fall by Everstring, 25% of respondents said they were currently using some predictive tools, and another 47% said they were aware of predictive marketing and were investigating how to use it.

Two recent studies by Forrester Consulting reported even higher usage rates of predictive analytics among B2B companies. In one of these studies, 49% of survey respondents said they were currently using predictive analytics, and another 40% said they were planning to implement predictive analytics in the next 12 months. In the second study, 61% of survey respondents said they were currently using predictive analytics, and another 26% said they were planning an implementation within 12 months.

A new study by Econsultancy (in association with RedEye) provides a much-needed dose of reality regarding the adoption of predictive analytics and the challenges of using it effectively. The Econsultancy study was based on an online survey of nearly 400 digital marketers and e-commerce professionals that was fielded in April and May 2016. About half (51%) of the survey respondents were based in the United Kingdom, 23% were based in North America, and 22% were based in Europe. Respondents represented a wide range of industries.

Fifty-nine percent of the respondents to the Econsultancy survey work for client-side enterprises ("companies"), while 41% work for agencies, vendors, or consultancies. The survey results described in this post are based on the answers given by company respondents only.

The Econsultancy study confirms the strong level of predictive analytics usage. Forty percent of respondents reported that their companies are either currently using, implementing, or have budgeted for predictive analytics over the next 12 months. In addition, 80% of respondents said that the use of predictive analytics is "critical" or "very important" to the future of their organizations. Given this view, it shouldn't be surprising that 65% of the respondents said their company's budget for predictive analytics will increase in the coming year.

The Econsultancy research also revealed that predictive analytics is not a "magic wand that automatically guarantees sales." For example, 53% of the survey respondents from companies currently using predictive analytics said that their sales had significantly increased over the past year. However, 50% of respondents with companies that had "evaluated implementing predictive analytics and decided it's impractical" also reported a significant increase in sales over the past year. So, it's questionable whether the use of predictive analytics was the primary cause of the increased sales in those companies that are using predictive tools.

It's also clear from the Econsultancy study that you can't expect predictive analytics to be an overnight success. Only 23% of survey respondents rated their company as "competent" or "highly competent" at the use of predictive analytics, and 35% of the respondents strongly agreed that they "are yet to realize the benefits of predictive analytics."

Predictive analytics is becoming an important marketing tool for many large and mid-size B2B companies. But like most tools, it will take work and practice to maximize the value of predictive analytics in marketing.

Illustration courtesy of Skye D. via Flickr CC.

What Separates Top-Performing Marketers from the Field?

The most basic of all business questions is:  What drives high performance? For years, the effort to find the "secret sauce" for achieving high performance has probably consumed more brainpower than any other single business topic. It's been, and still is, the modern-day business equivalent of the quest for the Holy Grail.

The desire to understand what drives high performance is widespread in the marketing world, and dozens of research studies have been devoted to solving the marketing performance puzzle. Earlier this year, Salesforce published a research report that includes several interesting insights on the practices of high-performing marketing teams.

The 2016 State of Marketing report is based on a worldwide survey that produced 3,975 responses from marketing leaders. Twenty-six percent of the respondents were affiliated with B2C companies, 29% with B2B companies, and 45% with B2B2C companies.

For this report, Salesforce divided survey respondents into three cohorts:

• High performers (18% of the respondents) were those who were "extremely satisfied" with the current outcomes realized as a direct result of their company's marketing investments.
• Moderate performers (68% of the respondents) were those who were "very or moderately satisfied" with the outcomes produced by their marketing efforts.
• Underperformers (14% of the respondents) were those who were "slightly or not at all satisfied" with their marketing results.

As you might expect, the Salesforce survey revealed some major differences between high-performing marketing teams and underperformers.

High Performers Focus on the Customer Journey

High performers were 8.8x more likely than underperformers (65% vs. 7%) to indicate that their company had adopted a customer journey strategy as part of its overall business strategy. For this research, Salesforce defined "customer journey" as all interactions that customers have with a company's brands, products, or services across all touchpoints and channels.

Salesforce also found that high-performing marketers were 9.7x more likely than underperformers to be actively mapping the customer journey, and 7.7x more likely to be leading customer experience initiatives across the business. 

These findings indicate that top-performing marketers are highly focused on delivering outstanding customer experiences across the entire customer lifecycle.

High Performers Leverage Technology

Another significant difference between high performers and underperformers relates to the use of marketing technologies. The Salesforce research found that high-performing marketers were:

• 7.6x more likely than underperformers to be "heavy" adopters of marketing tools and technologies
• 10.7x more likely than underperformers to be extensively using predictive intelligence technologies in their marketing efforts
• 6.7x more likely than underperformers to be extensively using marketing automation
• 2.8x more likely than underperformers to be substantially increasing spending on marketing tools and technologies

There's no longer any doubt that marketing has become deeply entwined with technology. So, it shouldn't be surprising that top-performing marketing teams are leveraging technology tools to enable and support their marketing activities and programs.

The Salesforce report includes several findings regarding the effectiveness of specific marketing channels, and it includes a breakdown of the major survey findings for eight individual countries. If you're a marketer, it will be well worth your time to read the entire report.

Illustration courtesy of Yoggl Innsbruck via Flickr CC.

What Makes a Prospect Attractive for Account-Based Marketing

The basic premise of account-based marketing (ABM) is that a company should focus its marketing and sales efforts on prospects that have a strong likelihood of becoming good customers. So it shouldn't be surprising that account selection is widely regarded as the most critical step in building a successful ABM program.

Most ABM practitioners choose their target accounts by identifying businesses that "look like" their best existing customers. This technique - known as look-alike modeling - is an effective way for most companies to select target accounts in most circumstances. However, like any business tool, look-alike modeling must be used correctly, and in some cases, choosing target accounts based solely on look-alike modeling may not produce optimal results. Therefore, it's important for marketers to understand the underlying factors that make companies good targets for ABM.

The following diagram depicts the factors that make a prospect organization attractive for account-based marketing. At the highest level, attractiveness is a function of value and buying potential. In this context, value simply means that a prospect has the potential to be a large and profitable customer for your company. The best measure of this factor is the estimated lifetime value that the prospect would produce for your company.

Buying potential refers to the likelihood that a prospect will purchase your company's products or services, and as the diagram shows, buying potential is a function of two factors - fit and interest.

Fit is one of those business terms that's hard to define in a precise and formal way. The underlying idea is suitability, and one dimension of fit is whether your company's products or services can effectively address a need, problem, or challenge that the prospect is likely to have. In the diagram, I call this solution fit.

The second dimension of fit is more subtle, but equally important. I call this dimension company fit, and it refers to whether your company can effectively market to, sell to, and serve a particular prospect. Company fit is often a function of geography for small and mid-size companies. For example, if your company is based in Atlanta and primarily serves customers located in the southeastern United States, you may not be able to effectively market or sell to, or serve, a prospect located on the west coast, no matter how well your products or services fit the prospect's needs.

The second component of buying potential is interest, which refers to whether a prospect has shown an inclination to evaluate or purchase the kinds of products or services that your company offers. Interest also has two components - engagement and buying signals. Engagement refers to whether a prospect has had direct interactions with your company. Has anyone affiliated with the prospect visited your website, consumed your marketing content, or met with one of your sales reps? Has the prospect bought from your company in the past?

The other dimension of interest is buying signals, and this refers to prospect behaviors (other than direct interactions with your company) that indicate the prospect may be interested in the kinds of products or services your company provides. Today, most accessible buying signals consist of online behaviors such as website visits and content consumption behaviors. These behaviors are represented as intent data, which is collected and sold by B2B publishers. Some providers of predictive analytics acquire access to this data and incorporate it into their PA solutions. Therefore, as a practical matter, you will only have access to this type of intent data if you are using a predictive analytics solution to support your marketing efforts.

Earlier, I noted that look-alike modeling is an effective way for most companies to select target accounts for their ABM program in most circumstances. However, in some cases, choosing target accounts based solely on look-alike modeling won't produce optimal results. In those circumstances, you'll need to step back and use the factors described in this post. In a future post, I'll describe some of the circumstances that require more than pure look-alike modeling.

Two Key Promises of Predictive Marketing

As a B2B marketer, imagine how much more effective your marketing efforts would be if you had the following insights:

• What if you could identify businesses that are likely to have a strong interest in your company's products or services before you market to those businesses?
• What if you could reliably identify which of your current prospects have a strong propensity to buy your company's products or services and thus are ready to have a meaningful conversation with one of your sales reps?

These are two of the most significant promises of predictive marketing solutions. During 2015, predictive marketing was one of the hot technologies in B2B marketing, and it appears that the demand for predictive marketing solutions is poised to grow rapidly. 

Last fall, Everstring published the results of a survey of marketers regarding the use of various marketing technologies. Twenty-five percent of the survey respondents said they were currently using some predictive tools, and another 47% said they were aware of predictive marketing and were investigating how to use it. Two studies by Forrester Consulting - available here and here - reported even higher usage rates of predictive marketing analytics among B2B companies.

Predictive marketing solutions have the potential to dramatically improve the productivity of B2B demand generation by enabling companies to target their marketing and sales activities more precisely. Predictive analytics can be used to address a wide range of business issues, but the two uses that are receiving most of the attention in the B2B marketing world are new prospect acquisition and prospect/lead scoring.

Most predictive marketing solutions employ the same basic approach for both of these use cases. They take data regarding your company's existing customers from your CRM and marketing automation systems and combine that information with external data about those customers - from around the web, social media, and other third-party data sources - to construct a customer data model that describes the attributes and behaviors of organizations that are likely to have a strong interest in your company's products or services.

When predictive marketing is used to identify new prospects, the solution provider will run your customer data model against its (the solution provider's) database of businesses. The result is a list of prospects that resemble - to a greater or lesser extent - your existing customers. The inference is that prospects that closely resemble your existing customers  are likely to be interested in your company's products or services. With this insight, you can target your marketing programs more precisely and use your marketing resources where they are more likely to be effective.

When predictive marketing is used for prospect/lead scoring, the solution provider applies your customer data model to the prospects already in your marketing database and generates a score for each prospect based on how closely the prospect resembles your existing customers. This enables you to qualify prospects or leads using much more data than is typically available in traditional lead scoring systems. In theory, therefore, a predictive marketing solution qualifies prospects and leads more accurately, and it can potentially identify buying signals that are almost impossible to find using traditional lead scoring techniques.

The early indications are that predictive marketing solutions will drive significant business benefits. For example, in a 2015 study by Forrester Consulting, 72% of respondents whose companies were using predictive marketing grew revenues by 10% or more during 2014. Only 33% of non-users achieved the same rate of revenue growth.

While its clear that predictive marketing solutions can provide significant benefits in the right circumstances, there are a few caveats that marketers should keep in mind. For example:

• These solutions rely heavily on data from a company's CRM and marketing automation systems to construct the customer data model. So, if your company is a fairly mature user of CRM and marketing automation, and if your systems contain a significant amount of usable data, predictive marketing could be a sound investment. On the other hand, if you don't have enough reliable CRM/marketing automation data to work with, the value of predictive marketing will be more problematic.
• It's also important to recognize that you need a reasonable number of existing customers to create a customer data model that is reliable and predictive. Put simply, your customer data model will be richer and more reliable if it is based on 500 customers rather than on 50 customers.
• Predictive marketing solutions are not outrageously expensive, but they can require a significant investment. The cost of predictive marketing solutions varies greatly depending on the features of the solution and a variety of other factors. Pricing can always change, of course, but at present, it appears that the starting price for most predictive marketing solutions ranges from around $15,000 per year to over $100,000 per year.

Illustration courtesy of Louise McLaren via Flickr CC.