How to Write Effective AI Prompts for Sales Tasks? Best Practices and Proven Examples

Given the hype around ChatGPT and AI in general, one might think of it as a magic wand that can do wonders on your request. And it can do that, at least to some extent.

The thing about generative AI is that the outcomes it creates are only as good as the instructions you give to it. Just like a magic wand, AI requires the right spell to really do the magic ✨

So if you see someone ranting about how useless and overrated all AI tools are, it’s likely that they just don’t know how to effectively use them. Effective AI prompting is the skill any modern sales professional has to learn to stay productive and competitive in the market. 

In this post, we’ll explore the best practices for prompt writing and share some proven AI prompts for common sales tasks to help you establish an effective dialog with AI.

What’s an AI prompt? The basics of prompt engineering

The concept of generative AI isn’t new. The first rule-based systems that could simulate human thinking date back to the 1960s. However, early on, it took some serious tech skills to “communicate” with those AI models, for example, by submitting the data via an API. 

With the development of consumer-grade AI tools, most algorithms are now capable of understanding plain language. As a result, you can ask AI to do something just as would ask a human colleague.

The starting query used to task an AI agent is usually called a prompt. Think of it as what you would say after “Hey, Siri!”

As mentioned earlier, it can be a straightforward ask written in a conversational manner for simple tasks, like translating a text or summing up your call notes. 

However, it gets trickier when dealing with more complex requests, especially those that involve a series of consecutive actions, e.g., analyzing your ideal customer profile and coming up with a few relevant personalization snippets or value propositions. 

This is the case for prompt engineering (or simply prompting) — the process of refining and optimizing your queries in order to get the best output from the AI agent. 

In other words, prompt engineering can help you write prompts that are clear, detailed, context-rich, and effective. Which is a valuable skill that can make you more productive while also allowing you to get the most out of the AI tools at hand.

Let’s talk about how you can improve your prompting skills.

How to write AI prompts to get the desired results? Best practices and pro tips

Any effective AI prompt consists of 2 key elements: Context (background information to give AI a good idea of what you’re trying to accomplish) and instructions (guidelines for the task).

The more context you can provide and the clearer your instructions are — the more effective your prompt will be. It’s as simple as that.

However, the approaches to prompting might slightly differ based on the specific AI tool you use. For example, there are separate best practices to write prompts for Jason AI, our own AI-powered sales assistant. 

Regardless of the AI tool you’re using, there are some of the universal best practices you should follow to write AI prompts that work.

7 best practices for creating an effective AI prompt

• Give clear and specific instructions for the task you want AI to accomplish for you. Let’s say you want AI to rephrase a template for you. Rather than simply asking that, make your query more informative — specify what exactly you want to rephrase, e.g., improve the I/you ratio, readability, structure, or change the tone of voice.
• Structure your prompts well to avoid misreading. While the language model’s perception of written text differs from that of humans, proper spacing and punctuation can still make your prompt a lot clearer. Namely, it’s recommended to separate the instructions and the context, especially when quoting something or providing examples.
• Always give as much context as possible to be 100% sure AI understands your reasoning and purpose based on any relevant background info you might have. For example, a prompt asking simply to write a follow-up email after a call with the prospect won’t be as effective as the one where you include call notes/summary.

Pro tip: There are AI tools or add-ons that allow you to attach different files — PDFs or images — so be sure to use that when trying to add more context to your prompts.

• Other than the contextual info, try to provide an example of what you’re looking for in terms of output format, style, etc. AI tools like ChatGPT can really do a lot of things. A prompt is a way to make them do exactly what is needed and not go overboard by setting certain limitations and defining your expectations. In this case, including an example will give AI a pattern to follow when completing a task.
• Specify the role you want AI to perform using phrases like “Imagine you are” or “Act as.” Just like any other parameters you use, this will help you set the stage for the specific task and get a more precise output. Don’t believe me? Try asking AI to perform the same task using different personas and see how different the results will be. An email created from the point of view of an SDR will differ from the one a CEO would write (let alone other less likely personas like Eminem or Hemingway).
• Iterate with additional requests to fine-tune the AI outcomes. Also known as iterative questioning, this technique suggests rephrasing your initial request or specifying some parts of it if you don’t get the desired outcome on your first try. Sometimes, using a series of shorter, more specific prompts can lead to a more comprehensive result.
• Build on your previous conversations with AI. Tools like ChatGPT save the history of your previous requests and outcomes so if you use the same chat for related tasks, it can use the previous context to ensure more accurate responses for your consequent queries. This works really well for context-rich prompts, like creating templates or developing sequences for each of your buyer personas. 

Pro tip: Train AI to write prompts for you. That might sound meta, but asking ChatGPT to give you the best prompt for a certain task is a pro move that few will actually think of. However, it can be a great shortcut to get the most out of your AI efforts when you’re only getting started.

Aside from these general best practices, there are some prompting techniques you can try. But let’s not dive into details here. Instead, let’s explore some of the proven AI prompts salespeople can use right away.

20+ proven ChatGPT prompts for sales teams

It’s practically impossible to cover all the potential ways you can use tools like ChatGPT for sales development. There are new, creative ways to use AI for sales-related tasks and unique prompts salespeople come up with based on their specific needs and preferences.

Yet most of the prompts used by SDRs will look like this:

• • Write an email/LinkedIn message to {persona} about {purpose}
  • Personalize the email below based on {data points}
  • Generate {5/10/15} variants of subject lines for an email
  • Suggest a P.S. text for the email below in {language}
  • Translate the email below to {language}
  • Improve email to follow the {style/format} of the {example}
  • Rewrite the email below in {style/tone/language}
  • Make the email below shorter than {number} words
  • Write an email follow-up based on the call summary {notes}

And this is perfectly fine! As long as you get what you need and include any important details, as discussed above. 

For example, the ready-to-use prompt we would use to create a targeted cold email would be: “Write a cold email to a prospect in the {industry} industry and with job title {job_title}. Find problems related to this industry and generate solutions based on value proposition {value_proposition}. Keep the email within 100 words. I know prospect data — use it for personalization: company is {company}, first name is {first_name}, last name is {last_name}. Act as salesperson, name – {user_name}, company – {user_company}.“

However, there are situations where you will have to get creative and use more sophisticated frameworks to get the job done with AI. Here are some examples of such prompts from sales and AI practitioners.

Thibaut Souyris, one of our favorite sales AI experts (and invited speaker at our upcoming Sales AI Marathon) shared his set of go-to prompts for prospecting in a recent LinkedIn post.

• Find prospect problems: Act like [prospect name], the [job title] at [company]. List your top 3 challenges for 2023. Bullet point and concise answer only.
• Find problem symptoms: Now list 3 symptoms for each challenge above. Bullet points and short answers only.
• Generate email subject lines: Here are 10 email subject lines that got my attention: [examples] Act like a top-performing SDR and generate 10 new outbound prospecting email subject lines based on the ones above. Be short and creative, and focus on teasing the prospect’s curiosity.
• Generate cold outbound messages: Now I’d like to build a 9-touchpoint outbound sequence in order to get [prospect name] to reply to my prospecting. I use the following framework for each message: [framework] Based on the framework I shared, build a cold outreach message for each symptom listed above. 
• Turn a marketing resource into a prospecting resource: Act like an online marketer and turn the following content into a playbook: {summary}

Another outstanding expert, Heather Murray, shares her creative Chain of Thought prompting technique for ICP research. 

1. Start with a pretty straightforward prompt: “Act as a business consultant. I’m the Director at a lead generation agency, and I’m looking to launch a new high-end lead generation product for the aerospace industry. Let’s start by discussing some initial ideas that would work — ask me any relevant questions to flesh this out into a strong idea and the beginnings of a plan to launch it.”
2. Build on this with the second prompt in the same thread: “Great, now let’s explore the people that would be most interested in this new product. What job titles would they have, and why might they be interested in the product?”
3. Dig a bit deeper: “Tell me more about the pain points of people with these job titles in the aerospace industry. List pain points commonly experienced, and match my solution to their pain.” 

If you need, you can go on like that, building on each previous response in the same chat until you have all the responses you need. This is a perfect approach for optimizing your value proposition and targeting, brainstorming new approaches, or fine-tuning your current tactics. 

Lastly, Jake Dunlap, CEO of Skaled Consulting (and an expert speaker at the Sales AI Summit) shared various prompt scenarios in his recent webinar on LinkedIn. 

He points out that ChatGPT is mostly used for research and writing messages, but this isn’t the only use case sales teams can benefit from. Instead, he suggests incorporating ChatGPT into your SDR onboarding/training strategy. There are a few possible scenarios here.

• Objections handling: “Act as {buyer persona} and give me 5 objections which I will try to counter.” After that, you can also ask AI to tell you how you did and come up with some more objections for you to handle.
• Role-playing: “I want you to impersonate {role} in {industry} and I want to role-play a sales meeting with you. A typical sales meeting looks like this {summary}. Here’s what our product does {value prop}. Let’s go!”
• Scenario planning: “I have a call coming up with {persona} discussing {topic}. I’m trying to think of any reasons they might say no. Help me think through navigating a meeting where I’m talking about {proposal}.” 

Over to you

As you can see, there’s nothing too difficult about AI prompting. Once you know the basics and play around with your AI tool of choice, you can find the magic formula to get the desired output.

If you’re not into experiments or don’t have the time to learn effective AI prompting, you can choose AI tools that offer a more seamless user experience. 

For example, one of the latest product enhancements for Jason AI — our built-in sales email assistant — is a smooth, intuitive way to optimize your prompts so you can get exactly what is needed every time.

The feature is available in Beta at the moment, but you can sign up with Reply anyway to explore the available AI-powered capabilities and be the first to know when it goes public.