Advanced Email Personalization: Injecting AI Variables into Proven Copy

Advanced Email Personalization: Injecting AI Variables into Proven Copy

There is a misconception that using AI in cold email means letting ChatGPT write your entire message.

This is a mistake.

The best way to use AI for email copy isn't to have it write the email—it's to have it write the variables.

You want to create the copy lines yourself to ensure the tone and offer are perfect, and then use AI to find specific data points to insert into that copy. This gives you control over the output while making every email look 100% researched and personalized to the prospect.

Here is how we execute this strategy using Clay.

The "Variable Injection" Strategy

Instead of a generic prompt like "Write me a cold email for a bank," you write the email template yourself and leave placeholders for specific data.

Generic Approach:

"We help companies like yours grow deposits..."

Variable Injection Approach:

"We help [Competitor A] and [Competitor B] grow deposits..."

The difference is subtle but powerful. It proves you know who they are and who they compete with.

Example 1: The "E-commerce" Shift

We used to send emails for a client saying:

"We can reach out to 2,000 potential customers for you."

People replied asking, "Who are my customers?" It wasn't compelling.

We changed the variable to be specific to their target market:

"We can reach out to 2,000 e-commerce brands for you."

Suddenly, the prospect thinks, "Okay, these guys actually know who we target." It looks less generic and significantly more personalized.

Example 2: The "Competitor Drop"

Let's say you are reaching out to a bank, like Northfield Bank.

1. Company Name Normalization

First, never use the raw company name from a database. "Northfield Bank Inc." or "Revenue Flow LLC" screams automation. Use a simple AI prompt in Clay to "Normalize this company name."

• Bad: "Want to grow Revenue Flow LLC's deposits?"
• Good: "Want to grow Revenue Flow's deposits?"

2. Relevant Competitors

This is the highest-ROI play. Ask your client for a list of all the banks they currently serve. Then, for every prospect you scrape, use Clay/AI to:

1. Look at the prospect's region and size.
2. Compare it against your client's customer list.
3. Pick the top 3 biggest competitors from your list that the prospect definitely knows.

The Resulting Copy:

"We've generated $25B in deposits for hundreds of institutions, including [Competitor A], [Competitor B], and [Competitor C]."

If you mention three banks in their region that they compete with daily, you instantly build authority.

How to Execute This in Clay

You don't need to overcomplicate this. You don't need a 50-step waterfall to get started.

1. Create your Columns: In Clay, set up columns for the variables you need (e.g., Normalized Name, Competitor 1, Competitor 2).
2. The Prompt: Use an AI column (like Claude or GPT-4) to find the information.

Here is a list of my client's customers: [List]. 

Here is the prospect I am emailing: [Prospect Name]. 

Find the 3 companies from my list that are the most direct competitors to this prospect.

3. Iterate: You will likely need to run this prompt 5-10 times to refine it. Make sure it doesn't hallucinate or pick irrelevant companies.
4. Insert: Map these output columns directly into your email sender variables.

For a more advanced way to handle data processing in Clay (especially if the data structure is complex), check out our guide on how to find decision makers in Clay which covers handling list outputs.

The Low-Hanging Fruit

You don't need to get too fancy yet. Start with these low-lift variables:

• Normalized Company Name: Essential for sounding human.
• Specific Customer Type: "E-commerce brands" vs "customers".
• Relevant Competitors/Partners: Mentioning names they recognize.

The offer is still the most important part of your email. But adding these specific variables makes a strong offer feel like it was written just for them.

If you want to see how we scale this approach across hundreds of prospects, read our guide on scaling personalization with Clay.