Let me start by saying that AI won't replace sales people, but sales people using AI will replace sales people that don't. Now I did not come up with this phrase, I adjusted the well known phrase about AI not replacing people but people using AI replacing people that don't.
So what does that mean for you? Well that means that you need to start using AI to augment yourself and get more out of your time, yesterday.
The best way to get into using AI to augment yourself is by learning how to prompt AI.
And if you can prompt AI to get the desired result, then you are already half way there. The other half is data, because guess what, AI needs data to understand the context, but that's a post for another day.
So let's get on with it and talk about everything prompting.
Table of contents:
- AI prompts: the new age of personalized outreach
- The anatomy of a good AI prompt for outbound sales
- The context of the sender
- The context of the recipient
- The objective of the sending
- The must-dos and must-nots
- Example of the desired result
- Getting started with AI prompting
AI prompts: the new age of personalized outreach
"There are many prompts out there but this one is mine." - to paraphrase the line from Full Metal Jacket, AI prompts are the new templates.
Remember templates?
Like email templates with placeholders and text that were sent to 100s if not 1000s of people with minor differences from one contact to another. When I say minor, I mean max 10% of the email would change from John to Jack.
Well prompts are the new templates, but they at one meta level higher than templates. They don't result in 100s of emails with minor differences, when done well, they result in 1000s of 100% unique emails to based on the context of the prospect, context of you, and context of the outreach.
In this post, I'm going to unpack all three of these for you and then some more.
NOTE: This will be a living post, meaning I will be regularly updating it with new findings and tactics. Bookmark it. Put it in your RSS feed. Check it from time to time as I will be using this post as my journal for all my prompting experiences.
I will also be synthesizing tactics used by hubsell customers and sharing them here in an anonymous format (and with their permission of course).
The anatomy of a good AI prompt for outbound sales
I say outbound sales because the desired outcome requires the information of the two parties, the sender (you) and the recipient (the prospect).
So here are, as per my experience so far, the 5 components of a good AI prompt for outbound sales.
- The context of the sender
- The context of the recipient
- The objective of the sending
- The must-dos and must-nots
- (Optional) Example of the desired result
I marked the last one optional not because it's not important but rather it will be hard for you to give AI the desired results from the get go. You are better off giving the AI the first four and then take some of the better examples that AI gives you and edit them to perfection and give one those examples as the desired result.
I'm going to explain each of these, for which I will be using the following prompt as an example.
You are an experienced B2B sales person who specializes in writing emails to setup appointments with potential decision makers. As that person, I want you to write a highly targeted, snappy and to the point email to [[firstName]] who is [[title]] of [[companyId.fullName]], which operates in [[companyId.industry]] industry.
You are representing hubsell, as [[senderId.firstName]], who is a Co-founder at hubsell.
To write this email I want you to look at [[firstName]]'s and his company [[companyId.fullName]]'s information to first determine what their ideal customer profile is.
Here is some information about [[firstName]]:
[[firstName]], describes [[#if (eq gender 'Female')]]herself[[else if (eq gender 'Male')]]himself[[/if]] as the following in their LinkedIn about section: """[[[summary]]]"""
[[firstName]] has the following work experiences: """[[#each experiences]] [[[this.title]]] [[[this.company]]] [[[this.summary]]] [[[this.since]]] [[[this.upto]]] [[[this.locationCity]]][[/each]]"""
[[firstName]]'s company, [[companyId.fullName]] says the following about themselves:
Here is [[companyId.fullName]]'s home page summary: """[[companyId.homepageSummary]]"""
Here are [[companyId.fullName]]'s specialities: """[[companyId.specialities]]"""
Here are [[companyId.fullName]]'s about section on their LinkedIn page """[[companyId.shortDescription]]"""
Based on the information above information for [[firstName]] and [[companyId.fullName]] craft an email that introduces to [[firstName]] the idea of hubsell.
Here is proposed structure of the email:
The first line in the email should be one of the following:
"""Would love your take on the following.""" OR """Let me get your take on this.""" OR """How does this land with you?"""
Then the right after that, at the start of the email there should be something specific about [[firstName]] to grab their attention. Start this part with something like Looks like your ideal customers are X, where X is what to determine to be there ideal customer profile. Based on X determine what are some of the pain-points in reaching out to X (not here that the idea is NOT to list X's pain points but rather the difficulties in reaching out to X via outbound sales).
Then once you've described the pain points of reaching out to X, mention that while it has gotten difficult, outbound sales is still one of the most reliable ways to get a foot in the door with your target decision makers, given that the outreach is done with relevant messaging to relevant decision makers.
Then introduce the idea of hubsell with words similar to but not as elaborated as the following:
hubsell uses AI to pull in someone's LinkedIn profile to personalise each email and DM so it's unique. This results in better response rates so you get better returns on your cold email outbound efforts. Essentially a shift from volume to conversion. hubsell enables outbound sales with 100% unique and contextual messages across email and LinkedIn. It is based on data that the solution gathers to be added into AI instructions which results in highly detailed and specific context provided to AI to come up with unique messages. It's how we reduced CAC for Workspace 365 by 33% within 12 weeks of them starting.
End the email with sentences similar to the following:
"""Do you want me to send a 3-min video showing you how this works?""" OR """Do you want some more information on this?""" OR """Wanna know more?"""
Rules to follow when writing the email:
Write the email in British English.
Write the email at Flesch-Kincaid score 80 reading level.
Always write hubsell in lowercase letters.
The email should be between 50-60 words and should have 2-3 paragraphs.
Start the email with on of the following 3 options:
Hello [[firstName]], Hi [[firstName]], Hey [[firstName]],
The email should not start with phrases like """I hope you're well""" or """I hope this email finds you well""" or any form of introducing the sender (Karan)
End the email with: Best,
[[senderId.firstName]]"""
Use very simple terms, and use short words. For example where you can say 'use' don't say 'leverage'
Don't use the following words or any of their derivations in the entire email: remarkable, impressive, distinctive, extensive, considerable, significant, passion, innovative.
Don't mention [[[firstName]]]'s track record and expertise.
Don't use marketing style of writing.
Don't use hype words.
Don't use exclamation marks.
Don't specifically name [[[firstName]]]'s jobs or the companies they worked for.
Don't be sycophantic.
Do not make things up which are not clear from the content shared above.
Do not provide the subject line, we are focusing only on the email body.
Do not add any placeholder text that needs to be replaced. The email should be ready to send without any editing.
The end result should look like the following example:
"""Hi [FirstName],
How does this land with you?
Looks like your ideal customers are [focus] departments and sales teams overwhelmed with [repetitive tasks] requests. Reaching them can be tricky due to the sheer volume of outreach they receive. Outbound sales, when done with tailored messaging, is still a reliable way to connect with the right decision-makers.
hubsell uses AI to pull in someone's LinkedIn profile to personalise each email and DM, resulting in better response rates. It's all about moving from volume to conversion.
Do you want me to send a 3-min video showing you how this works?
Best,
Karan"""
*Please note that the content with [[]] or [[[]]] is replaced by two and three curly brackets respectively. Furthermore, the content inside the [] is replaced in this prompt to keep the content anonymous.
In the example above you've gotten all 5 five parts of a good AI prompt.
Let's start with the first.
1. The context of the sender (you)
Here is the context of the sender of the prompt above, which in this case is me.
You are an experienced B2B sales person who specializes in writing emails to setup appointments with potential decision makers. As that person, I want you to write a highly targeted, snappy and to the point email to [[firstName]] who is [[title]] of [[companyId.fullName]], which operates in [[companyId.industry]] industry.
You are representing hubsell, as [[senderId.firstName]], who is a Co-founder at hubsell.
I explain to the AI who I am, what company I represent. Additionally, I add later in the prompt what my company does.
hubsell uses AI to pull in someone's LinkedIn profile to personalise each email and DM so it's unique. This results in better response rates so you get better returns on your cold email outbound efforts. Essentially a shift from volume to conversion. hubsell enables outbound sales with 100% unique and contextual messages across email and LinkedIn. It is based on data that the solution gathers to be added into AI instructions which results in highly detailed and specific context provided to AI to come up with unique messages. It's how we reduced CAC for Workspace 365 by 33% within 12 weeks of them starting.
Based on this information the AI can have a good understanding of what I and my company does.
I can be as detailed as I want when I describe my context because I know it well and there is only sender vs 100s of recipients.
2. The context of the recipient (the contact)
The context of the recipient is trickier because unless you are targeting a handful of contacts per day, the amount of research you'll need to do personally per contact will be a huge time and productivity drain. Good news is that AI can help with that.
You can add all the data you have to the AI to understand the contact better. In the example below, I share with the AI my contacts about section on their LinkedIn profile, their full work history with the description of each job.
To write this email I want you to look at [[firstName]]'s and his company [[companyId.fullName]]'s information to first determine what their ideal customer profile is.
Here is some information about [[firstName]]:
[[firstName]], describes [[#if (eq gender 'Female')]]herself[[else if (eq gender 'Male')]]himself[[/if]] as the following in their LinkedIn about section: """[[[summary]]]"""
[[firstName]] has the following work experiences: """[[#each experiences]] [[[this.title]]] [[[this.company]]] [[[this.summary]]] [[[this.since]]] [[[this.upto]]] [[[this.locationCity]]][[/each]]"""
[[firstName]]'s company, [[companyId.fullName]] says the following about themselves:
Here is [[companyId.fullName]]'s home page summary: """[[companyId.homepageSummary]]"""
Here are [[companyId.fullName]]'s specialities: """[[companyId.specialities]]"""
Here are [[companyId.fullName]]'s about section on their LinkedIn page """[[companyId.shortDescription]]"""
Based on the information above information for [[firstName]] and [[companyId.fullName]] craft an email that introduces to [[firstName]] the idea of hubsell.
I can of course go further than this and add open positions, funding rounds, news and even parts of their 10-K filing for the AI to get a much deeper understanding of the target lead and account.
However, it is key to not overload the AI with information because that will make the end result more unpredictable. The idea is to give the AI the right information to write the message copy for you.
3. The objective of the sending
Here is where things get interesting. So far we have been looking at relatively static information. The context of the sender does not change from sequence to sequence, and while each recipient's context is different from one another, it is relatively static over time. The objective of the sending on the other hand can change for each sequence or even email or step in the sequence.
For example, the first email will have a different objective (to introduce) than the second email (follow up). The objective of the sending helps the AI further narrow down what it should write.
Based on the information above information for [[firstName]] and [[companyId.fullName]] craft an email that introduces to [[firstName]] the idea of hubsell.
Here is proposed structure of the email:
The first line in the email should be one of the following:
"""Would love your take on the following.""" OR """Let me get your take on this.""" OR """How does this land with you?"""
Then the right after that, at the start of the email there should be something specific about [[firstName]] to grab their attention. Start this part with something like Looks like your ideal customers are X, where X is what to determine to be there ideal customer profile. Based on X determine what are some of the pain-points in reaching out to X (not here that the idea is NOT to list X's pain points but rather the difficulties in reaching out to X via outbound sales).
Then once you've described the pain points of reaching out to X, mention that while it has gotten difficult, outbound sales is still one of the most reliable ways to get a foot in the door with your target decision makers, given that the outreach is done with relevant messaging to relevant decision makers.
Then introduce the idea of hubsell with words similar to but not as elaborated as the following:
hubsell uses AI to pull in someone's LinkedIn profile to personalise each email and DM so it's unique. This results in better response rates so you get better returns on your cold email outbound efforts. Essentially a shift from volume to conversion. hubsell enables outbound sales with 100% unique and contextual messages across email and LinkedIn. It is based on data that the solution gathers to be added into AI instructions which results in highly detailed and specific context provided to AI to come up with unique messages. It's how we reduced CAC for Workspace 365 by 33% within 12 weeks of them starting.
End the email with sentences similar to the following:
"""Do you want me to send a 3-min video showing you how this works?""" OR """Do you want some more information on this?""" OR """Wanna know more?"""
Moving on to the dos and don't for the AI.
4. The must-dos and must-nots
The must-dos and the must-nots are AI's guardrails. Further they enable you to instruct AI to imitate you more closely.
In the text below you see that I am telling the AI to write the email in British English, think here things such as spelling favour with a U instead of the US spelling without the U. The Z vs I in words with zation ending, personalisation vs personlization. But furthermore, different words are used to describe the same thing in British vs US English, and since this prompt is used in a sequence with UK based decision makers, the appropriate language is British English.
AI also understands models and concepts quite well and has a host of information on them so in order to get the AI to write in simple English, I tell it to write the email at "Flesch-Kincaid score 80" level.
I also give the AI some hardcoded choices to lower the variation in some parts of the email. For example I give it 3 options to start the email and 1 option to end. Stronger even, I tell the AI to avoid certain phrases which typically cause emails to be not opened or ignored after being opened.
By providing the AI with rules to follow, you can lower the variation without limiting the creativity.
Rules to follow when writing the email:
Write the email in British English.
Write the email at Flesch-Kincaid score 80 reading level.
Always write hubsell in lowercase letters.
The email should be between 50-60 words and should have 2-3 paragraphs.
Start the email with on of the following 3 options:
Hello [[firstName]], Hi [[firstName]], Hey [[firstName]],
The email should not start with phrases like """I hope you're well""" or """I hope this email finds you well""" or any form of introducing the sender (Karan)
End the email with: Best,
[[senderId.firstName]]"""
Use very simple terms, and use short words. For example where you can say 'use' don't say 'leverage'
Don't use the following words or any of their derivations in the entire email: remarkable, impressive, distinctive, extensive, considerable, significant, passion, innovative.
Don't mention [[[firstName]]]'s track record and expertise.
Don't use marketing style of writing.
Don't use hype words.
Don't use exclamation marks.
Don't specifically name [[[firstName]]]'s jobs or the companies they worked for.
Don't be sycophantic.
Do not make things up which are not clear from the content shared above.
Do not provide the subject line, we are focusing only on the email body.
Do not add any placeholder text that needs to be replaced. The email should be ready to send without any editing.
5. Example of the desired result
The desired result gives the AI an idea of what the end result should look like. It will lower the variation of the result significantly so I do recommend trying the prompt without it and generate several different options and then select the best ones to give as example(s) to the AI.
The AI won't copy the end result ditto, it will follow it closely enough that the emails look similar but they won't be the same (which is a good thing).
The end result should look like the following example:
"""Hello [FirstName],
How does this land with you?
Looks like your ideal customers are [focus] departments and sales teams overwhelmed with [repetitive tasks] requests. Reaching them can be tricky due to the sheer volume of outreach they receive. Outbound sales, when done with tailored messaging, is still a reliable way to connect with the right decision-makers.
hubsell uses AI to pull in someone's LinkedIn profile to personalise each email and DM, resulting in better response rates. It's all about moving from volume to conversion.
Do you want me to send a 3-min video showing you how this works?
Best,
Karan"""
Getting started with AI prompting
The prompt that I shared here is just an example so I encourage you to follow the structure to create your unique prompts. However if you anything like 90% of our users, you are thinking that prompt engineering is a daunting task.
Take my word for it, it's not.
As someone who as create email templates for the last 12 years and has done more than dabble in prompt creation, I can tell you that prompt engineering is a lot simpler and less mentally exhausting than creating email templates.
The process of creating, testing and iterating with prompts is actually fun, in contrast to the process of creating email templates, which is tiresome.
Start by taking the above prompt and make changes to it to fit your contexts.
Get in touch with us if you have any questions or need any help.
Hint: Bookmark this page as there will be a lot of more content added to this post over the next weeks and months.
.png)