We hope you're not sitting there with a calculator in hand, punching out scores for each lead. We tried that, and it's slow.
Tools, systems, and automated data enrichment are a much better way to qualify leads at scale.
First, we'll talk about the data piece: how to kick-start or optimize your lead score by taking a closer look at the attributes that feed it.
Then, we'll discuss tech stack, sharing an overview of the must-have functions in a lead qualification system. You'll see a breakdown of the real tools Clearbit uses for its own scoring, and get a peek into our internal debates about building a system from scratch versus doing it all in our CRM.
In the earliest stages of a company's life, data-driven qualification is very broad. For example, it might use a couple of simple firmographic attributes to quickly eliminate large groups of leads — like people who are in totally incompatible industries, or at too-small companies (employee size), or in a different country (geography).
This simple approach works well until the company grows and needs more focus and precision. At that point, it's time to revisit the data feeding into the lead score to make it more specific.
Whenever you want to do the following things, it may be a good time to redefine your lead score's data points:
To figure out which data points to use, you can do a regression analysis on your past and current customers to understand which firmographic and demographic attributes they share. That's helpful for the first three scenarios above. And for the fourth scenario, which addresses shifting goals, you may want to run the regression analysis on a subset of customers that best reflect the direction you're heading in.
Here's an example of this: when Clearbit switched from analyzing our full customer list to analyzing just a subset that better reflected a new company strategy.
Until Clearbit was about five years old, before 2020, our singular company goal was to grow top-line revenue. Our definition of a successful customer was that it was a customer at all. Our lead scoring matured through a few beginner-to-intermediate phases: we started to gut-check the loose criteria our sales team had been using, and eventually formalized the data points after doing a regression analysis on our closed/won deals. After enriching a list of our closed/won customers with Clearbit data, we found our customers' most common shared characteristics.
From this analysis, we got a shortlist of firmographic and demographic attributes, like technologies used and role, which we put into a point-based lead score. We also added in data from a signup questionnaire, which asked new users about why they need Clearbit. Selecting the use cases we considered priorities was a positive signal that boosted their lead score.
That was our first approach — using all our closed-won deals to create an ICP and score new leads against it. To learn more, check out An inside look at Clearbit's ICP and lead scoring model, written in 2019.
We switched things up in 2020, when our company goal became less about growing "any type of revenue" and more about renewal and expansion revenue. We wanted to attract the types of customers that would get so much value from Clearbit that they'd stay loyal, expand their contracts, and grow with us long-term. By this point in time, we'd had several years of renewals to be able to calculate the lifetime value (LTV) of different cohorts of past customers and find the cohorts with the highest LTV.
So instead of doing a regression analysis on all of Clearbit's closed-won deals, we looked at a subset of customers with high LTV to find the attributes that set them apart.
Interestingly, it turned out that the number one predictor of high LTV in that group was a company's Alexa rank. It was a signal for companies with a lot of lead volume, which tend to get a lot of value from Clearbit.
So we reoriented our ICP definition and lead scoring data points around this concept, giving more weight to companies that had a good Alexa rank. For more information on this approach, see The search for a healthier ideal customer profile, written in 2020.
Since then, we've expanded our parameters again to include more customer types that were successful with Clearbit, but not captured in this ICP. Lead scoring is never one-and-done.
Wherever you are in your lead scoring maturity, you can gut-check what you're doing or find the next evolution of your lead score as your company scales into new markets. Here's a shortlist of popular SaaS lead scoring data points to try.
An automated lead qualification system includes a number of different functions spanning everything from the actual scoring, to the lead routing and engagement, to the storage of data.
You can get these processes either in an all-in-one platform like a MAP or CRM (think: HubSpot, Marketo, Salesforce), or by linking different tools that specialize in each function.
To illustrate, here's Clearbit's scoring stack, which qualifies and routes leads to the right sales rep. It's owned by our RevOps team.
We connected a bunch of different tools together, rather than using an all-in-one solution — we'll talk about the pros and cons of that choice in a moment.
Step 1. Prospecting and generating audiences:
We generate volume, interest, and intent with marketing and sales motions.
Examples include campaigns that use Clearbit Advertising audiences, or lists generated with Clearbit Prospector.
Step 2. Attribution, tracking & storage: People engage with our ads, content, and landing pages, and they hopefully sign up or make contact requests. This behavior is all captured with Segment.
Step 3. Data transmission:
Tray.io automatically sends the data from Segment to Salesforce in real time, where SDRs and BDRs work the leads.
What's not pictured here is another network of tools connected to this one, which does our intent scoring and marketing attribution. It's owned by our growth team. In Step 3, data also goes from Segment into our data warehouse, Redshift, where we do some data transformation with a tool called dbt and sync it back to Salesforce using Census. We also use Census to feed that data into our email system, Customer.io.
Step 4. CRM: Our sales team uses Salesforce day-to-day. Four main tools and processes plug in here as follows.
Step 5. Data enrichment:
Clearbit Enrichment fills in information about any individual in Salesforce — for example, their title and team, as well as their company's data, like location and size.
We use contacts as our core Salesforce objects (individuals), which are linked to accounts (companies).
Step 6. Lead scoring:
MadKudu plugs in to do fit scoring, giving us a score of low, medium, good, or very good for each lead. It uses machine learning to analyze our performance with past customers, then predicts a lead's success based on shared firmographic and demographic attributes. For example, did past leads that look similar to this one convert to an opportunity, did we win the opportunity, and did the customer retain and stay with us?
As mentioned above, we also do intent scoring in a separate set of tools owned by the growth team. For more information on this setup, see Clearbit's scoring story.
Step 7. Lead routing: LeanData lets us build logic rules in order to route leads to the appropriate team quickly.
Step 8. Direct booking:
When a lead starts filling out a contact request form, Chili Piper plugs into Salesforce and Clearbit Enrichment to quickly assess how qualified the lead is. If they're a great lead, it immediately offers meeting times on an AE's calendar so the lead can book directly. It's immediate scoring and routing with an AE round robin.
A typical MAP like HubSpot or Marketo can handle most of the functions listed in this qualification stack. Many companies, especially those of a small to medium size/complexity, could just use an out-of-the-box platform. The other approach is to build a home-grown solution.
The choices are:
Which way to go?
Two main considerations are your data collection needs and your team's resources. You can collect more data in the exact way you want it with a custom solution, but it requires more upkeep—and people to do that work. Many companies will only do a custom solution if they're getting really advanced and their MAP or CRM truly can't do something they're trying to achieve. Otherwise, it's not worth the extra staffing, design work, and maintenance.
So you need to find a balance between your resource constraints and getting the data and functions you need for good qualification to meet your company goals.
Take Clearbit, for instance. In the early days, we simply relied on Salesforce. But now that we're a little older, we've cobbled together many independent tools for a custom system.
It's fair to say that Clearbit is tech-heavy and that we've come in on the high-complexity side of the spectrum. People within our company are a little divided about that approach.
Julie Beynon, our Head of Analytics who sits in our growth team, explains that one-stop-shop solutions are a challenge for Clearbit because the marketing team wants to collect data from many different places, in many different ways. "With an out-of-the-box solution, you limit what you can do with scoring, routing, and collecting data because you can't custom fit to your needs," she says.
For example, we wanted our intent scoring to reflect someone's usage of Clearbit's free products. We looked for a one-stop shop marketing platform that would do behavioral tracking and found one that almost fit all our needs. The problem was that this platform didn't track user activity the way we wanted it to. If we used it, we'd have trouble understanding how a lead's usage of Clearbit's free product influences our sales cycles. Our scoring could only consider 80% of the story. So instead, we built a custom model that let us track engagement with Clearbit's free products exactly how we wanted, providing more signals about when to send a lead to sales.
And then there's the question of people-resources. We are very lucky to have an analyst like Julie who is dedicated to marketing efforts and can speak data architecture, join datasets, and write SQL in her sleep. Not every startup is staffed this way.
Travis Lee, systems consultant at Clearbit, points out that if Clearbit's current experts win the lottery and stop coming to work, we'd need to think hard about who would be able to inherit our custom system. "If you're a small startup, it's unlikely you have the resources for a marketing systems specialist," says Travis. "Maybe you have one RevOps person spread across five systems and three functions, trying to keep the plates spinning. In my role, I think about scalability, architecture, governance, and maintenance in the long term. You need to align your resourcing and future resourcing plans with your systems capabilities and company maturity."
The high-level challenge here is to match your company's staffing with its business maturity (and where it's headed), and with the right systems complexity. It needs to both meet your business goals and be maintainable.
At Clearbit, we've been investing more resources to evolve Clearbit's RevOps and marketing organization so that we're better matched across these three areas. There used to be just a few people overseeing a complex solution. Today, we have a head of RevOps, two business analysts, two Salesforce and systems consultants, and shared responsibilities with folks on the Growth team who own different pieces of the puzzle — like Julie overseeing automation across tools.
And we're hoping to scale the team even more. This might be the perfect time to mention that our RevOps team is hiring.
Engaging stories and exclusive data, designed for our best customers. One useful issue each month.