Artificial intelligence is incredible technology.
It can massively accelerate your workflow and automate mind-numbing tasks. It can calculate and comprehend staggering amounts of information in fractions of a second.
But one thing it can’t do is read your mind.
There’s a misguided expectation that AI should be able to handle any task we throw at it. But in reality, the output AI generates is only as good as the input provided.
Providing the right input is called prompt engineering.
This article explains the basics of prompt engineering specifically for sales and prospecting. The goal is for you to walk away with the knowledge and confidence to start implementing AI into your day-to-day and begin reaping the rewards this game-changing tech has to offer.
Let’s dig by explaining what a prompt actually is.
What is a prompt?
A prompt is an instruction or question you provide to an AI system to guide its output.
It’s essentially a way to communicate what you want the AI to do. Because most consumer-facing AI systems—think ChatGPT and Gemini (the AI formerly known as Bard)—are large language models, or LLMs, they understand natural language.
Meaning you can talk to it like you would a human and get a (mostly) human-like response.
💡Tip: A large language model, or LLM, is a type of artificial intelligence. LLMs are trained on huge amounts of data and designed to understand and generate human language.
However, just because AI understands you it doesn’t mean it can read your mind.
You need to tell it exactly what you need to get the best output.
Prompting is the process of doing that. As you repeat and iterate that process to get even better responses, you’re now doing a basic form of prompt engineering.
What is prompt engineering?
First thing to understand about prompt engineering is it’s not a new concept. But it is having quite the moment in the sun.
There’s a lot of noise about what prompt engineering is, and what it isn’t. In an attempt to provide some clarity, we’ll provide the simplest definition possible.
Prompt engineering is the act of creating prompts to effectively guide an AI model.
It is a skill that is part art, part science. The goal is to get the best possible outcome, or at least one that most closely resembles your expectations.
What we’re talking about here is prompt engineering 101—in a sales and prospecting context.
Don’t let the big words (engineering) and acronyms (LLMs, GPT-4) intimidate you. A lot of prompting is simply trial and error. Playing around until you get the hang of it.
4 elements of a good prompt
Each new chat or interaction with an AI model is a blank slate.
Expecting it to produce the perfect output without providing any context is like asking a chef to prepare a meal without telling them what ingredients are available or who they are cooking for.
To get the output you want, you need to tell it exactly what you need via an effective prompt.
Good prompts require four things: instruction, context, roles, and rules. Let’s explore each.
Instruction: What you want the AI to do
Example: Write a sales prospecting email.
Let’s start with the basics. What do you want it to do? Describe the task to be done, or the question to be answered, in clear and simple language.
Context: Background information to improve the output
Example: The contact to whom you’re writing is the VP of Engineering at HubSpot.
Context is much broader. It can be any information you feel will produce a better output.
Following the example of writing a prospecting email, good context includes (but isn't limited to) details about the person you’re writing to, their responsibilities and challenges, and the company at which they work.
The clearer and more specific the context you provide, the better.
Roles: How you want the AI to behave
Example: Act as the CEO of a large private company that’s just completed a major acquisition.
Asking a LLM to assume a role can help make the output more authoritative and nuanced. It enhances the relevance, tone, and expertise of the responses.
It’s like getting it to look at a topic through a specific lens. The AI focuses and filters its responses according to the perspective of someone in that role.
Rules: Guidelines for the AI to follow
Example: Use simple and direct language, no fluff words or jargon, and short sentences.
Unless you tell it otherwise, AI will format the output in the way it decides is most appropriate.
To ensure the output it generates aligns with what you have in mind, you need to give it strict rules and guidelines to follow. Again, AI can’t read your mind. You need to spell out exactly what you want it to do.
General tips for writing good sales prompts
The four practices above are the pillars of good prompt writing.
While the following tips aren’t essential, they’re still good practices to keep in mind. And it’ll help you generate more accurate, more compelling, and more creative responses.
- Provide clear instructions. The more descriptive and specific the prompt, the better the output. Clear and well-structured instructions are far more effective at producing better responses than using specific keywords. Include only relevant details, don’t repeat yourself, and be mindful of the overall length. We’ve found with longer prompts things can sometimes get lost in the mix.
- The output is only as good as the input. Another, slightly crude way of saying this is garbage in, garbage out. The AI’s response is only as good as the information you provide it. Yes, it is extremely capable technology. But it’s not magic. You need to give it the ingredients to make the magic happen.
- Create a new chat for each prospect/account. ChatGPT and other LLMs recall information from earlier in the chat to make responses better. Leverage this by teaching it about your prospect and their company as it can recall the information it learns to generate more relevant responses.
- Ask AI to help you write prompts. This is meta. When the right words escape you, ask the AI how it should be prompted. This will give you tips for creating better formatted prompts that in turn generate better responses.
- Provide examples of what you want. If you’re writing a cold email, share some examples of good emails that have worked well in the past, or that you’ve received. This will help the AI create more relevant outputs.
- Don’t expect magic. Build up to the final desired output. This is called prompt chaining. Instead of asking the AI model to do a complex task like write a cold email straight off the bat, break it into subtasks. As each subtask is completed it has more context to draw on, which will make the final output far better.
- Prompting is trial and error. The best way to become a better prompt engineer is to continually test different approaches to see what the AI model is capable of doing. Iterate and tweak your approach, and document what works.
5 AI use cases for outbound sales prospecting—with prompts
Let’s dive into some applications of AI in outbound that you can incorporate into your workflow straight away.
Assessing lead quality
You can use AI for basic qualification.
For example, it can check whether certain keywords are present on a lead’s profile or a target company’s website and qualify or disqualify that account accordingly.
Prompt:
Navigate to this website [target company URL]. Check for keywords related to:
- [Keyword #1]
- [Keyword #2]
- [Keyword #3]
Are these keywords or close matches present on this website? Provide a yes or no answer.
💡Tip: Include up to 12 keywords that are important to your business. If there are more than 12, consider spreading them over multiple prompts.
Prospect research
Research is essential for understanding the people you’re reaching out to.
However, as important as it is, that makes it no less of a drag on your time and energy. AI can greatly speed up the research process, which leaves you more time to create better messaging.
This prompt string will help you gain a good understanding of your contact in a couple minutes.
Prompts:
Learn about my prospect [prospect’s name]. My prospect is the [title] at [company]. Please browse for any information or recent news about [prospect].
Now read [prospect]’s LinkedIn profile and summarize it: [Paste LinkedIn profile data]
💡Tip: ChatGPT can’t access LinkedIn profiles directly, so it’s best to copy information from your prospect’s profile and paste it in.
Give me the top five pain points [prospect] may be having in this role.
Company research
Just as AI can help you research prospects, it can also help you gain a better understanding of the companies they work for.
B2B companies are hard to understand. Their products are complex and technical and their websites are filled with marketing jargon. But you need to understand what they do if you want to successfully prospect into them. These prompts can help you understand what a company does in a couple minutes.
Go to this URL: [Target company home page] Read the page and await my next prompt.
Repeat for the About and Product page.
Now do a search for information about [Company].
Now it has the context it needs.
Based on the above information, provide ~100 word answers in simple language to the following questions:
What does [Company] do?
Who are [Company]’s ideal customers?
How does [Company]'s product work?
What is the one big problem [Company] solves for their customers?
And for bonus points:
Pretend you're meeting the CEO of [Company]. What are some insightful complements I could offer or questions I could ask? Give three examples. Use a conversational tone.
This prompt string gives you a neat summary in plain language about what the company does, and arms you with some questions that’ll impress your contact.
Writing outreach emails
AI can give you a huge head start in writing outreach messages.
Providing the AI model context about your prospect using the prompts above will help you generate far more relevant and personalized messaging. It’s a good idea to run some of the above prompts before asking it to write an email.
Play around with the formatting and structure of the email till you get an output you like.
💡Tip: AI-generated emails won’t be perfect. They still require a human to edit and fact check.
Prompt:
Write a sales prospecting email to [prospect]. Follow these strict rules:
- Output must not be longer than 75 words—make every word count
- Use simple and direct language, no fluff words or jargon
- Use short sentences and paragraphs
- Use a paragraph break between each of the lines below
Follow this structure:
Line 1: Highlight an interesting observation about [prospect] or [prospect’s company].
Line 2: Link this observation to a broader problem or pain point [prospect] is experiencing.
Line 3: Describe how [your company] can help by [insert your value proposition].
Line 4: Include a call to action phrased as an open-ended question. The purpose of this call to action is to prompt [prospect] to respond.
Line 5: Sign off exactly as follows:
Best,
[your name]
Line 6: Add a short P.S. that is personal and relevant to [prospect]. It may or may not be related to the subject of the email above. It could be a past personal or professional achievement, a promotion, or other relevant and interesting detail from their background. Use a casual and conversational tone.
Writing P.S. lines for emails
Adding a P.S. to cold emails can increase responses by 36%. It does more than just show you’ve done your homework; it humanizes your outreach by making it personal to your prospect. Here’s four to try:
Prompts:
- “Write a short P.S. referencing a recent publication, social media post, or accomplishment.”
- “Write a one-line P.S. posed as a question that reframes the prospect’s challenges and gets them to think differently.”
- “Add a short P.S. that references a relevant or interesting insight from my prospect’s background.”
- “Add a friendly P.S. that highlights and compliments a skill or ability relevant to this person.”
How to prompt with OneShot
As AI continues to dominate more and more of B2B, prompting is an invaluable skill that will set you apart.
The good news is, we’re still on the ground floor. You can dive into the world of AI and prompt engineering now, and in six months time be far ahead of the majority of sales professionals out there.
The ability to write great prompts is how you unlock the power of AI.
But you still hit the productivity ceiling of only being able to work one account or prospect at a time.
That’s why we built OneShot.ai.
With OneShot.ai you can run your best prompts on an entire list of contacts at the same time. Instead of researching one prospect at a time, you can research fifty, or a hundred or more, simultaneously. Same with qualifying accounts, or writing outreach messaging.
Plus, you can nest prompts within prompts to get exponentially better responses.
It’s true outbound at scale, powered by AI. Curious to see it in action? Grab a 1-1 demo of OneShot.ai right here.
