Create a better B2B target account list with machine learning

Account-Based Marketing is more than a buzzword echoing through advertising trade groups and your LinkedIn network. Account-Based Marketing­­––or ABM––is a revised approach to attracting customers. The core of the ABM approach is applying a refined specificity to the customers you’re trying to attract. Rather than casting a wide net to pull in as many potential customers as possible, ABM challenges companies and marketers to think carefully about who they most would like to attract and create narrowly tailored messages directly to that audience.

Implementing Account-Based Marketing typically involves a redesign of the entire marketing approach, from creative and messaging to targeting and tracking. While a successful execution of ABM requires all of these to be working in tandem, of critical importance is the Target Account List (TAL)––a modest-sized group of specific clients you are targeting, and marketing to. 

Curating a Target Account List requires deep thought devoted to developing an ideal client profile and composing a list of clients based on that profile. Ideal client profiles, and the resulting TAL that they produce, are multi-faceted, but not always easy to use. One problem that may come up––and which came up for us at BOL––is a profile characteristic that isn’t easy to filter on: whether a company mainly sells to businesses or directly to consumers.

Here’s the approach BOL took to solve it:

Refine your target account list 

For ABM, you rely on both quantitative and qualitative measures to gather a list of accounts that you are hoping to target through an integrated ABM campaign. 

There are plenty of ways to approach this, but your focus should be coming up with account-level criteria based on factors like the number of employees, estimated revenue, number of locations, HQ location, industry, and operational service footprint. 

Crunchbase Pro and DnB Hoovers are tools that allow you to query through their databases of companies and discover your initial Targeted Account Lists. 

Most marketers that do ABM are satisfied with the output and use this list from the database vendors to then serve targeted display ads programmatically through platforms like 6sense and/or manage firmographic campaigns on LinkedIn and similar services.

Get rid of waste to improve efficiency

B2B Marketers who are running ABM campaigns to attract B2B customers often struggle with excluding a portion of the companies that land in their Targeted Account List due to limitations in the criteria available from data providers. We typically find that 20–30% of the companies that make it through particular filtered criteria are exclusively B2C companies. 

The danger of keeping these companies in your ABM Campaign is that you waste money and dilute the effectiveness of your marketing efforts. This is about efficiency and effectiveness at the same time.

The two-sided impact

  1. Waste: By having unwanted companies in your TAL, you are paying to serve valuable impressions to companies that you don’t want undermining effective performance.
  2. Reward: The Ad Platforms within Linkedin & Google reward marketers for their campaign performance by ranking your digital ads higher and/or serving your ads at a more efficient CPM. 

Let’s play this out.


Imagine that your original Target Account List yields 1,000 companies where 800 of them are B2B and 200 of them are exclusively B2C. 

With a $25k ABM budget at an average CPM of $5, this would serve 5 million impressions.

  • Across 1K accounts, this would be 5k impressions per account
  • Across 800 accounts, this would serve 6,250 impressions per account. 

By removing 20% of your waste, you increase your ability to serve 25% more impressions per account. This will not only improve cost efficiency but also enable greater account-by-account penetration to drive greater campaign performance. 


Ad platforms like Google & Linkedin maintain a ranking of quality for your campaign. Google refers to this as ‘Quality Score’ and rates the quality and relevance of your ads which has a direct impact on the CPC you pay. 

Google uses an array of different factors (such as the exact search term of a user, their location, device type, and bounce rate among other metrics) to assess the quality and relevance of your ads against specific audiences in real-time during the auction. 

  • If you can improve your score, Google will reward you with a higher ad ranking, leading to more—and better—ad placements. 
  • As an added benefit, your CPC will decrease as your ad quality improves.
  • In addition, higher Google Ads Quality Scores mean lower costs per conversion

The intrinsic value of your ad ranking higher, coupled with a more cost-efficient cost-per-click and improved cost-per-conversion will help generate more actionable opportunities for you than reducing waste alone. 

This could result in:

  • Serving more impressions for the same budget with better account penetration.
  • Spending more efficiently on your ABM campaign than you originally forecasted. 
  • Generate a lower CPC to justify additional marketing spend to generate a richer pipeline.

Identify B2C companies sneaking into your ABM campaigns

At BOL we launch ABM Campaigns. We love personalization and automation since they create a better customer experience at scale and enable true full-funnel integration. 

We found an opportunity to improve the process by tackling the arduous process of manually removing companies that are exclusively B2C. 

It seems pretty simple, but do you have the time to go through a list of 1,000 companies to determine if those companies are B2B vs B2C? Probably not. I tried it, it was fun for the first batch of 50. But it soon became boring and time-consuming, and introduced the risk of human error.

We contacted three of the top five database vendors that aggregate company details and none of them have an out-of-the-box capability to query on a field related to B2B vs. B2C. We could filter on proxy attributes that would be more indicative of B2B, but it’s not foolproof and certainly not scalable.

Meanwhile, the ABM team is frustrated because all they want is a list of B2B companies from the initial Targeted Account list that they provided to us. 



B2B or B2C






Enter creativity 

To address this issue, we combined our B2B marketing experience with our analytic muscle and constructed a method to procedurally classify prospective clients as B2B or B2C. In the end, we built a multi-faceted approach to inferring a company’s orientation.

Technographic profiling

Have you ever tried to navigate the marketing technology stack landscape? There are over 8K marketing technologies in Scott Brinker’s latest gathering. Understanding this environment is further complicated by the fact that nearly every company has a unique collection of marketing technologies in its stack.

BOL has been serving B2B clients for 20 years. We know from our experience what successful B2B companies have - and don’t have. Whether they are optimally using those marketing technologies is another issue, but we have a pulse on what technologies are exclusively B2B and which aren’t. 

In addition to knowing which technologies our clients should or shouldn’t have, we can lean on our expertise to back out how a company functions by observing what they do and what technology they have in play.

The first step in our approach to classifying the companies in our TAL was to understand their tech stack. We are able to do this with the help of BuiltWith’s API, which trawls websites and collects information about the tools embedded within them. Now, some of these are not very informative––pretty much every website from to has a Content Management System ––but other technologies can send a strong signal about what a company does.

We trained a machine-learning algorithm to identify exactly the kind of technologies that are informative in understanding how a business operates––B2B or B2C. The result is a decision tree that we can feed an account through, which returns that account’s modeled orientation.

Sentiment analysis

The best way one could classify prospective clients as B2B or B2C would be to ask them. There’s almost no margin for error in this approach. Obviously, that’s neither a practical, nor advisable approach to gaining this kind of information. The next best thing would be to pour over the company’s website, teasing out every detail that can be informative in making this determination. Still, this is not scalable. What we’ve opted to do, is to take a concise, readily available description of the company and use machine learning techniques to parse the identifying information from it. 

We use CrunchBase, a company database platform, to enrich our Target Account List with a textual description of their business. This information is processed and analyzed by our sentiment analysis algorithm to return a designation.

Now that we have a list of companies classified as B2B and B2C, our ABM team excludes the irrelevant ones––B2C-oriented companies in our case––from the Target Account List on a digital marketing platform like 6Sense. The ABM platforms will omit serving impressions on B2C accounts, further tailoring the Target Account List.

This process will empower you to achieve double-sided benefits: reduce wasted impressions on non-ideal customers, and improve your ABM Campaign Quality with the major ABM ad platforms.

Connect with our ABM team to start putting machine learning-supported ABM to work for you.