When it comes to sales prospecting, AI can help with 3 main use cases:

• Structuring unstructured data
• Summarizing text and creating snippet lines
• Writing an entire cold email

It can do the first two with minimal review as long as the prompts are correct. Writing a whole email is still a little out of its reach.

Here are some example prompts to help you write better cold emails:

Find the mission of the company

If you want to begin personalizing your outreach at scale, try finding the mission of the company to get your feet wet. For this, we will be using Clay.com to put all of the relevant data we need as well as the OpenAI API in one place.

The input that we will use for this prompt will be the description of the company from their LinkedIn page. Here is the prompt that I came up with.

What is the mission of the company using the input? The input: “{{LinkedIn Company Description}}” Be specific and use keywords in the description not normally found in other companies. Keep the output under 6 words and make the output conversational/casual. Complete this prompt: "I was on your LinkedIn company page and it looks you're focused on"

Here is a screenshot of the outputs.

We could clean this up by making sure that every line starts with the intended prefix and taking out the quotes, but otherwise, I would say these are all accurate and relevant to the accounts. This prompt is ready to scale.

Discover who the company sells to

This prompt is one that could be very useful to people to either segment accounts or use the output in your outreach. We will be using the LinkedIn company description again to try to get an output about who the company sells to. Specifically, the type of persona. Here’s the prompt I have found to work pretty well for this output.

Here is the prompt: What is the job title that this company usually sells to using the input as a guide for what they do. The input is this: {{LinkedIn Company Description}} Who gets most value out of the product and what is their usual job title? Give me up to three job titles. Do not include any numbers or extra information. Just a comma separated list of titles.

Check out the results here!

Find the pricing terms of the company

In order to find this data automatically, we will automate a Google search to find the pricing page and bring this into the snippet that Google supplies for our data.

The search we used is below.

site:domain.com (”pricing” AND “$”)

The reason we include the dollar sign is so that the snippet that shows up from Google will most likely include a price.

Then we put this into Open AI to structure the data. Here’s the prompt that I have found works very well.

How much is this company's highest pricing per month using the input. Be as specific and short as possible. Also tell me if it's monthly or annual pricing. The input:Google Snippet

Find the focus of a job based on someone’s job title

In order to do this, we will need the description of the person’s job based on their LinkedIn job description. You can find this data in the “Enrich Person from LinkedIn profile” integration.

The prompt we will use is below.

Tell me what is the focus of this person's role based on the title name and the input. Be specific and casual. The title name is this: {{LinkedIn Job Title}}" The input is this: "{{LinkedIn Summary}}" Complete this prompt with the tasks they are responsible for in under 6 words. "As the {{title}}, I'd imagine you focus on”

The sample outputs are below.

Note: after some testing, this works on just a title level as well. The LinkedIn job description is not necessary.

You can also use this and train the model on what to say depending on who you sell into. If you sell to revenue leaders and want to talk about different challenges they face, you can change the prompt to include something like “If there’s marketing in their role, they are currently facing an increase in cost per clicks. If there’s sales in the title, they are currently facing an issue with email deliverability.”

Summarize a news article

In order to summarize a news article, we can find the news article either through a Google search or a Predict Leads search. Most of the time, we can just use the title of the news article as our input.

This is a slightly longer prompt making up for some of the ways headlines are written.

Using the input, complete my prompt using keywords specific to the news article. Compliment them on the findings of the article. Keep it very short and casual. Make sure to leave out corporate jargon. Shorten company names when applicable. Don't make any reference to the time of the article being published. The input is: {{Headline}} Complete this prompt: I saw the recent news article about

Here are some of the results!

Note: that the empty rows there of course didn’t output a useful AI line.

Summarize a open job title and turn it into an inferred problem

Looking at the types of people a company is looking to hire is an amazing way to look into what the company’s main problem is today and what their goals are for the next 6 months.

Simply put, most of the time hiring for a certain position like a copywriter means that they need to fulfill some type of copywriting need in the company.

We can use Open AI to take an input of the role title and make a guess about what kind of problem they are trying to solve at the company.

The first thing we will need to start with is the job title or description that a company is looking to hire. We can get that data with either the Google Jobs Listing data or the Predict Leads data.

Then we will use the open job title in our prompt to infer a problem and make a snippet for our emails.

The prompt we used is below.

Tell me what problem this company is trying to solve based on the open job that is listed in the input. Keep it short and be specific based on what that job title is known to be tasked with in a company. The input is this: {{job title}} Complete this prompt: "I saw your company was hiring for a {{job title}}. In my experience this means you're trying to improve the company's"

Here’s a screenshot of some of the results!

Infer if a company is likely B2B or B2C

If you are selling a solution to help brands reach consumers on TikTok, it could be troublesome to find if they have a B2B offering or a B2C offering. Most databases do not have this data and it’s tough to infer without manually looking at the site.

With Open AI, we can use the company’s LinkedIn page to make the inference for us based on their description!

The prompt we will use for this is below.

A B2B company is a company that sells to other companies. A B2C company is a company that sells to consumers. Using the input, tell me if the company is likely a B2B company or a B2C company. The input is this: {{company description}} The only acceptable output is either "B2B" or "B2C" do not answer in any other way.

Summarize someone’s LinkedIn post

You can use AI to look at someone’s last LinkedIn post and summarize it down to the main idea. Use the Clay “Find Recent LinkedIn posts” to find the text of the posts. This works on company profiles and personal profiles!

The prompt we used for this is below!

Use the input to complete my prompt in under 8 words. Keep the output short and use specific keywords from the post. These are social media posts by other people and I would just like to know the main idea of each post. The input is this: "LinkedIn Post" Complete this prompt: "I just wanted to reach out because I saw your post about"

Clean titles of contacts

Often, when you are reaching out to people, you don’t want to include a custom variable that says their title when they say something like “Founder: We’re Hiring.” It is a dead give away that you scraped their data.

Take a list of job titles and use this prompt.

Using the input, clean the job titles from unimportant information and keep the most important part of the job title. Often times people list two job titles, say things like "we're hiring", or they put seniorities in weird places. Clean the input to only inlcude the main job title. The input: "{{Job titles}}"

Here is an example of the outputs.

Categorize whether a company works on SaaS

Often times, SaaS company’s can be difficult to distinguish because they will list their industry as the industry that they provide a solution to. You can use their LinkedIn company description to infer if they are a SaaS company or not. Here’s the prompt that we used for this.

Is the company in the input a software as a service company?  A software as a sercive company is a company that offers a software usually for a monthly or annual description to multiple users providing them with a service. The input is this: "{{Company description}}" Only return a result as "true" if it is a software as a service company or "false" if it is not a software as a service company.

Here’s an example of the outputs!

Infer a company’s Glassdoor reviews

To start, we need to find the company’s Glassdoor reviews. We can easily do this by automating a Google search in Clay.

The Google search we will use is below.

site:glassdoor.com "{{Company Name}}" "has an overall rating"

The AI prompt we will use is below as well.

Using the input, tell me what is the company review of the company. Only return a numerical value, no words. the input is: "{{Google snippet}}"

Check out the results below!

‍