I had the opportunity to discuss the use of predictive scoring with Infer’s Nikhil Balaraman, Director of Product Marketing and Sean Zinsmeister, VP of Product Marketing. I asked them to challenge my thinking on the adoption of Predictive tools as well as explain more about how Infer and predictive can help firms improve marketing ROI.
Josh: In the Martech Maturity Model I wrote about, I placed Predictive tools at Stage 6 – the very end of the 24-36 month implementation timeframe for firms to build out martech. Do you agree or disagree, and why?
Waiting until the end of a martech implementation is certainly one approach to adopting predictive tools, however, I’d argue that in most cases there’s no need to wait that long before getting a leg up on your competition. In fact, many of our customers start with predictive fit scoring prior to implementing marketing automation (MAP). Here are a few key use cases for predictive that we’ve seen at early stages of the Martech Maturity Model:
- Stage 0 (Marketing Transformation): Most companies don’t start building their sales and marketing stack by selecting a MAP vendor — their first step is usually to purchase a CRM system like Microsoft Dynamics or Salesforce to store sales data. At this juncture, the business challenge is to filter and prioritize leads so that sales knows which ones to work, which is a great use case for a predictive solution like Infer. As long as a company has captured sales data on at least 100 or so conversions in their CRM system, we can build and deploy a statistically accurate model for them that same day. Additionally, we can build Market Development Models for companies. These models are based solely on lists of their target companies, and helps them more efficiency enter new markets or roll new products out to market. In both scenarios, adoption is usually quite fast, since Infer Scores can be easily integrated into pre-existing CRM workflows, such as lead assignment and routing.
- Stage 1 (Automation): Once a company has started the marketing transformation process and adopted a MAP as system of record, predictive behavior models can accelerate the impact and simplify the rest of the stages by providing a system of intelligence with insights and actionable intelligence for reps and marketers. These predictive models assign an immediate quantitative measure of value to each lead and account based on a machine learned model and trained on historical data; therefore, the score not susceptible to human bias in the same way as rules-based scores. This intelligence should be a considered part of every decision a company makes across their funnel.
- Stage 2 (Lead Quality Management): At this stage, we’ve seen great results from predictive with customers like Nitro. The company had a “champagne problem” of so many leads that they were breaking their marketing automation system. Since their reps could only work a tiny percentage of their leads, Nitro needed to implement predictive scoring immediately so that they weren’t wasting time chasing low quality leads. Infer also helped the company determine which leads to keep in their marketing automation system.
- Stage 3 (Nurturing in Sales Context): Here, companies should use predictive fit scoring to identify which prospects are not a good fit for their business, and won’t convert into revenue. These types of leads can be funneled into low touch nurture tracks. In addition, predictive behavior scoring can help monitor all prospects in these nurture tracks and push highly engaged prospects back into sales reps’ hands.
We don’t believe predictive is a single point solution to only be implemented at the end of a 3-year marketing transformation.
Josh: Interesting. While I agree that predictive can support Nurturing, I’ve found firms in these Stages aren’t ready to consider powerful tools because they are still learning how to use Marketing Automation, Nurturing, and sales-marketing alignment.
Josh: How does Infer think about ABM+Predictive? What’s different about ABM+predictive vs. statistically correlated lead scoring?
Account-based marketing and predictive are highly complementary strategies for go-to-market. ABM is all about targeting key accounts, but the reality is that reps today are typically assigned hundreds of accounts. This begs the question: how do you build those lists in the first place, and how do reps know where to start? At the end of the day, ABM needs to begin with predictive, otherwise you are choosing accounts in the dark.
With predictive fit scoring, you can begin by scoring all the accounts and showing reps those that are most likely to generate revenue. With account-based fit models, you first select a target outcome—such as “Accounts with closed won opportunities”— then a tool like Infer will build a predictive model using thousands of data points such as semantic analysis of the website, annual revenue, or technologies in use at each account.
Even if a company does not have enough data to build a predictive fit model, say for a new vertical or market, that company can still benefit from predictive by using a Market Development Model to identify its top account list. From there, many companies layer on account-based behavior scoring, which uses MAP data to show reps which of their accounts are most engaged with the account-based campaigns that marketing is running. In addition, predictive sales intelligence tools (like Infer Glance) provide sales context directly to reps within Salesforce across the Lead, Account, Contact and Opportunity levels. This saves them valuable time since they don’t need to open multiple tabs to do research on their prospects.
Josh: What are the common issues your customers come up against when implementing ABM/Predictive? What do they wish they knew before they started?
We see a broad range of sales and marketing challenges that our customers are looking to solve with ABM and predictive.
Challenges stem from simply not knowing where to begin. Sometimes, there might be a fear of wasting resources on something so new because the opportunity cost is not known (e.g. pipeline generated by events, email campaigns, virtual events, etc.). Without a good initial sense of how much impact ABM will drive, our customers tend to approach this framework cautiously. To find clarity, we instruct them to begin by answering the following questions:
- How should I score accounts to determine their fit for my business?
- Do I have a way to measure both lead and account engagement to understand which accounts are activated?
- Do I have a way to acquire net-new contacts and accounts that look like my best customers?
- Do my reps have the right information on these accounts, and where are the gaps?
For these reasons, Infer customers like Yesware, Nitro, InsightSquared, Rapid7 and Druva infused predictive intelligence into their ABM programs to accomplish the following goals:
- To find an objective way to identify top accounts in their system, and assign them to reps in a fair and transparent manner
- Identify and target high-potential accounts in new regions and verticals where they did not have enough data for a predictive scoring model
- Align sales and marketing on the top accounts based on an objective measure to ensure everyone agreed on the ranking mechanism
- Measure the impact of their ABM programs and campaigns in real-time by tracking engagement scores to ensure that their dollars are being well spent
We often hear early on how helpful it can be to take a moment to zoom out and understand which issues your business is trying to solve. Many companies get caught up in the buzz of ABM, but there might not always be a “there there.” Once you’ve identified the market problem and honed your focus, it’s important to audit your tech stack to determine whether your backend, operations, and workflows will actually support your ABM strategy. For example, do you have a way to A) automatically match leads to their respective accounts, then B) score those leads?
To get started, we recommend our customers use an ABM framework to guide their approach: start by mapping your current database, adding new leads, and completing your total addressable market (TAM) analysis and target account list. Prioritize and activate your best accounts, then measure success with predictive behavior scoring.
Josh: When vendors discuss ROI of their product, they speak of the vision that could be attained, rather than the time savings involved with, say, automation of lead routing. As a buyer of ABM tools, what would you really want to hear about the ROI of ABM?
We see the most important ROI measures for ABM and predictive as those that have a clear impact on the bottom line. For example, do your strategies result in a measurable increase in deal sizes or ACV? Infer customers such as ZipRecruiter have told us that when they focus their effort on the prospects (or accounts, leads, contacts) with the highest fit scores, they increase deal sizes and conversion rates as a result.
Pipeline velocity is another important metric for sales teams, since increased velocity allows the team to work more accounts and close more deals with the same number of people. With account-based behavior scoring, Infer surfaces the most engaged accounts to the reps, which allows them to reach out to prospects when they look like they’re ready to make a purchasing decision, rather than waiting around for the prospect to call. By working accounts when they’re ready to convert, time to close goes down.
For example, if a rep is able to identify when an account is showing heightened interest before they even engage with an inbound channel, the rep can prospect into the account and proactively begin a conversation. As a result, engagement happens earlier than it typically would in a standard sales process because the rep no longer has to wait for the lead to first ‘raise their hand.’ In this case, the rep is able to close this deal faster than they would otherwise, then quickly move onto the next account in their target list that is showing interest.
Josh: Which types of firms are most ready for predictive tools? Which are not?
We have seen predictive platforms adopted by organizations across the spectrum — from eight-person companies that don’t even have a CRM to multinational companies that run their entire business through products such as Salesforce and Marketo. Even when companies think their process of passing good leads to sales is as efficient as possible, we’ve seen gains from sales intelligence tools that helps reps plan their conversations. Additionally, customers with “dirty” or incomplete data are also a great fit for predictive solutions like Infer Glance and Infer Smart Signals, which can take an input such as email address and identify the company that the prospect works for, then provide the rep with valuable firmographic and technographic signals directly in Salesforce.
For example, we’ve found that basic data such as a company name, website, and email address tends to be fairly complete and high quality across systems. For the most part, data quality tends to decrease around data points that reps are expected to proactively gather and manually enter themselves, including fields like employee count, revenue, or complementary technologies in use. To remedy this pain point, we use matching abilities to further expand the account record. With only a company name, website, or email address, we’re able to determine additional details about a company, such as website technologies in use or social media presence, as well as standard firmographic signals. Then, we push the data back into the customer’s system or record at the Lead, Contact, Account and Opportunity levels where it is continually kept up-to-date with Infer’s data cloud.
The most important trait for a company interested in predictive is a desire to harness advances in technology in order to accelerate growth. This typically requires a corporate culture that rewards those who take intelligent risks, is data-driven, or even is simply trying to do more with less. Businesses that find the most success with predictive tend to have a single pain point that can be alleviated with a predictive use case like filtering leads, prioritization, demand generation, or nurturing. This is because they grasp the value of predictive even before the model is live in their system, and are able to champion the solution across their company.
Josh: Thank you Nikhil and Sean for your time and viewpoint.