The New Era of Lead Scoring

Since the dawn of time, sales teams have been trying to properly implement lead scoring systems to figure out who to prioritize in their ICP. This has taken many forms, several of which have required either bad data or a lot of manual work.

However, the bronze age is now over. Lead scoring can now be done with the best data, using any scoring system you want, automatically using the power of Clay (and a few other tools).

This edition of the GTM Cookbook walks through the two ways we suggest you try lead scoring so you can have an automated, sleek and thoughtful way to prioritize your TAM.

Choosing Your Lead Scoring Method

There are two core approaches to lead scoring that I suggest you try when setting up a system, both of which lie on opposite ends of the spectrum. The first one is creating a completely custom lead scoring system using Clay, and the second is letting an AI scoring tool like MadKudu do the heavy lifting.

The latter obviously sounds better at first, but there are pros and cons of each:

Clay 

Pros
- Far more customizable
- More data points to score by
- Can be easily edited after testing once
Cons
- Difficult to set up

MadKudu
Pros
- Super quick to set up
- Refined AI model specifically for scoring
- Easy to integrate engagements into scoring
Cons
- Far less customizability

So, how do you choose which to go with? My quick answer is, it depends on how well you know your ICP and how much you’re willing to spend. Any company with a super well-defined ICP, as well as knowledge on what data points constitute a perfect prospect (as well as which data points are most important) should go in the Clay direction. If you have less budget to spend on scoring and lead routing, also go with Clay. If you’re willing to spend a bit more and would rather have AI do the heavy lifting when creating a scoring model, then get MadKudu.

If you’re already a Clay user and decide to get MadKudu as well, they actually integrate with each other too! The combination of both makes it absurdly easy to score leads as it takes literally one column. I did a webinar with both companies a while ago, here’s the link to watch it. In the following tutorial though, we’re going to be focusing more on doing it in Clay.

Creating Your Scoring Model

Regardless of which option you decide on, creating a scoring model is necessary to give both systems direction. Here’s what you need to determine:

• What data points do I want to score by? (ex: revenue, engagement on LinkedIn, visited website, company size, etc)

• What degree of importance does each data point serve? (the weight of each data point)

• What system do I want to use to score? Do I want to do more binary scoring (ex: if x>30, then score is 1 and if x<30, score is 0), or do I want more incremental scoring (for every 5 employees a company has, add 1 point of score

• How do I intend to find this data point to feed into the scoring model?

The scoring model essentially looks like this, if you’re more of a math person:

(Score x Weight 1) + (Score x Weight 2) + (Score x Weight 3) + …. = Cumulative Score

A big suggestion I have here is, don’t think too deep into the weights when creating V1. It’s better to figure out the important data points, feel out weights, and iterate after a few rounds of tests.

Very Important: Every data point should have the same scoring range. To make it easy, just have every data point score from 0-10.

Automating a Scoring System in Clay

Each data point you want to score by should have these components:

• An enrichment that helps find the data point in question (ex: enrich company integration to find company size)

• A weight column where you can easily assign importance to the score output

• A scoring criteria column

Here’s an example of what that looks like:

Once you have this for every data point, you just have to create a final AI integration in Clay to create the final score. Here’s an example prompt:

Input Data: Receive a set of criteria, each with an associated score and weight.

Multiply Scores by Weights: For each criterion, multiply the score (S) by its corresponding weight (W).

Sum Weighted Scores: Add up all the weighted score values.

Normalize by Total Weight: Divide the sum of weighted scores by the total sum of weights.

Scale to 100: Multiply the normalized value by 100 to obtain the final cumulative score.

The final score should be on a scale of 0 to 100. Output only the score between 1 and 100, nothing else. 

Here is the data:
{{data}}

Then, once built, you should have your v1 scoring system!

Testing the System

Testing is quite simple- find a few companies manually that (in your mind) perfectly match your ICP, and put them into the model. If they output a 100 (or close to it), you’re likely set!

Conversely, do the same with some companies that are definitely NOT a fit, and maybe even some that could be on the fringes of your ICP, just to see what it yields. This will help you adjust your prompting.

_____________________________________________________________________

Once your system is built perfectly, you can now import leads into Clay for automatic scoring! You can also webhook data directly from your marketing initiatives for real time scoring, then send it back to your CRM or to your sales team directly. If you want more content on that, just let me know.

Of course, you could also use the pre-built template we have in Clay to do this exact thing, built out for you to edit → https://app.clay.com/shared-table/share_PG9a8AMGQgox?via=b8a689

and here’s a video walking you through it → https://www.linkedin.com/posts/patrickspychalski_we-just-built-a-table-that-makes-account-activity-7297993179906146305-5zFh?utm_source=share&utm_medium=member_desktop&rcm=ACoAACm96PwBL6gGF2QqD5ak4jITm4uafP1ysy8

Happy scoring, and best of luck!

-Patrick Spychalski, The Kiln