How to Get Generative AI to Score Your Leads
Want to speed up lead scoring? Find out how to make generative AI do it for you.
When sales teams get lists of leads, scoring them from bad to good is a constant challenge, especially if they run into the hundreds and thousands of possible leads.
Fortunately, with the right prompt, you can tell generative AI how to do it for you, which is exactly what AiSDR does.
Here’s a closer look at how you can have generative AI shoulder the heavy load.
TLDR:
- The goal: Score large lead lists quickly and efficiently
- The tactic: Train generative AI on how to score leads
- The result: Lead lists are quickly scored and processed with little effort
Step 1: Clean the data using basic prompts
A lot of lead lists are a bit on the messy side.
Luckily, you can have generative AI clean your CSV files.
Here’s part of a sample prompt.
Format this text into a proper CSV. Include the fields: company name, first name, last name, email. Use semicolon as a separator. Try to guess the value of the mentioned fields from other available fields.
You’ll need to do a bit of prompt debugging if your prompt doesn’t work on the first try.
Step 2: Specify which fields are the most valuable
The fields that are most relevant and valuable for lead scoring come down to you and your lead scoring criteria.
Usually, you’ll at least want to include these fields:
- First name
- Last name
- Company name
- Experience
- Description
- Keywords
You can tell AI the importance of other fields like revenue and budget, and it will score leads accordingly.
Step 3: Define the buckets or scale for classifying leads
Whether you choose buckets or a scale is primarily up to you.
Common buckets for lead scoring include:
- Qualified – Leads have shown interest and meet your criteria outlined by your ideal customer profile.
- Unqualified – Leads don’t meet your criteria, whether it’s due to lack of interest, budget, or goal.
- Hot – Leads are close to making a purchase decision and should be prioritized.
- Warm – Leads have shown interest but may need extra nurturing before being sent to sales.
- Cold – Leads have shown little to no interest and need a lot of nurturing.
These are far from the only options for buckets. You can also use clustering like “Definitely content – industry sales” or “Should consider – food & beverage”. These evaluate a lead’s quality while assigning it a specific customer segment.
Likert scales are frequently used for scoring. However, it’s better to use a 7-point or 9-point scale since they show greater nuance. In my case, I like to use 10-point scales since people are very familiar with 1 to 10.
Step 4: Tell the AI to start scoring
For lack of better phrasing, once you’ve told AI how to score leads, all that’s left is to tell it to start scoring.
You’ll need to test your prompt a few times and make any adjustments as necessary.
Bonus tip: Ask AI to explain its methodology.
There are two benefits to telling AI to explain how it came to its conclusions:
- It makes debugging simpler as you can easily spot issues in its logic and methods.
- It refreshes data recency, improving the accuracy of results.
The Result
Delegating lead scoring to generative AI unlocks these benefits:
- Time savings – When generative AI handles lead scoring for you, your time spent on scoring drops to near zero.
- Scalability – Huge lead lists are a struggle to score quickly. Automating this allows you to scale your scoring, process leads faster, and pursue more sales opportunities.
- Better accuracy – Automation eliminates many of the typos and other small errors that can plague data entry and analysis. Generative AI can also go much deeper and quicker when scoring each lead.
