The Campaign Guide, for resultful campaigns!
A well-crafted prompt instructions is the key to successful SBL Campaigns. The clearer and more precise your instructions, the better the AI can deliver results that meet your expectations.
This guide will help you master prompt writing by covering essential elements like campaign setup, communication style, and user flow for high-performance SBL campaigns.
A. Overview of Campaign Creation Fields (Variable Inputs)
Before designing a successful campaign, you need to understand the key variables involved. These inputs define the structure, communication style, and flow of your campaign.
Campaign Name
Campaign Type
Override Company
Company Name
About Company
Override Persona
Persona Name
Persona Job Role
Communication Style
Target Users
Title
Description
Campaign Objective
Campaign Flow
Product Links
Initial Message
Followup Message
B. Writing Effective Instructions for Each Variable
Now that you understand the core variables, let's break down how to create effective prompts for each. The goal is to guide the AI in the best possible way, so it can deliver optimal performance in your campaign.
1. Campaign Name
The Campaign Name field doesn't directly affect the Chat. However, it should be descriptive enough to easily identify the campaign's purpose or target audience for internal use.
Tips:
Use numbered and lettered headings to organize the campaign structure or funnel.
Include a reference to the target contacts for the campaign, as this enhances its value for future use cases and analytics.
Example:
"1. User survey for the New Batch 05 - Hubspot 20k List (till 20th September)"
"2.A. Pre webinar promotion - Website Inbound Leads"
"2.B. Post webinar sales - Attendees (28th September)"
2. Campaign Type
The Campaign Type determines the approach and tone. Choose between Sales, Survey, or Support.
“Survey”: For user interviews, lead qualifications, product-related feedback, etc.
“Sales”: For sales pitch about an offering, booking meetings, etc.
“Support”: For handling all inbound customer query messages, guiding users through their product journey, etc
3. Override Company
This feature is used when you need to run campaigns for a specific sub-product within your larger organization. By default, your company name is passed to the model during chat. When you override the name and description, the model will treat that information while chatting, only for the particular campaign.
Company Name: The exact name of the creator or sub-brand.
Best Practice:
Use the full name of the company or sub-product to ensure clarity.
Use relevant words, such as 'from' or 'by', to show the connection between the sub-brand/sub-product and the main company, .
Example:
"Diet Coke by The Coca-Cola Company"
“Here from HDFC ERGO”
About Company: A brief, compelling description of the company or product.
Best Practice:
Focus on the unique value proposition.
Keep it concise but informative.
Example:
“Helping business leaders manage stress, deal with procrastination, and unlock their potential.”
Pro Tip: Write the description in a way that resonates with the campaign's target audience.
4. Override Persona
The Persona represents the identity of the AI during interactions.
Persona Name: Select a relatable and neutral name that fits the role.
Best Practice:
Choose a name that is easy to pronounce and remember.
Example:
"Shruthi"
Persona Job Role: This reflects the role AI is playing in the campaign.
Best Practice:
Align the role with the campaign objective and the audience's expectations.
Example:
“Community Manager,” “Product Consultant,” or “Career Coach.”
Pro Tip: The persona should feel human and approachable. Avoid job titles that sound too robotic or technical.
5. Communication Style
This field sets the tone of how the AI interacts with the user. You can select one or multiple communication styles, depending on the campaign.
Best Practice:
Use styles that align with your audience and campaign objective.
Avoid conflicting styles (e.g., formal and casual in the same campaign).
Try to use just One or Two style, max.
Example:
"Formal, Friendly"
"Professional"
Pro Tip: Test different combinations of communication styles to see which resonates best with your audience.
6. Target Users
This variable defines who the campaign is addressing.
Title: Identify the category of users the campaign is targeting.
Best Practice:
Keep it broad enough to cover the majority of your audience but specific enough to be meaningful.
Use a single, singular noun to describe the target.
Example:
"Professionals"
"Students"
Description: Explain how these users were identified or segmented.
Best Practice:
Be specific about how you obtained the user information.
Example:
“who all joined a Paid Webinar last week”
“who are the Existing customers of Antern”
“who had put their requests to know more about the S05 Offering”
7. Campaign Objective
The Campaign Objective sets the direction for the conversation between AI and users. This is where you define what the AI should accomplish in the interaction.
Best Practice:
Be clear and specific in what you want the AI to achieve.
Use action words like "engage," "identify," "solve," or "pitch."
Avoid transactional words like "sell" or "buy."
Avoid sharing any Links in the Objective.
Example:
“Engage with users feeling overwhelmed, identify their challenges in personal development, and pitch the 100-day mentorship program.”
“Reaching out folks who’ve shared their Product Challenge in a paid webinar “Unlocking Product Management” held over Zoom last week on 20th September, to ask if they want to join a 1-1 Call with our mentor for a personalized guidance.”
Pro Tip: Small adjustments in wording can lead to significantly different outcomes (e.g., “pitch” vs. “sell”, “get” vs “understand”). Choose words that align with the tone of the campaign.
8. Campaign Flow
This is the outline of how the conversation will unfold between AI and the user.
Best Practice:
Start with a warm, non-intrusive greeting.
Gradually guide the conversation by asking relevant questions that help the AI understand the user’s needs and then come to the main objective.
Example Flow 1:
Firstly thank the user for responding.
Ask if they are working or studying.
If they’re working, Ask where are you working currently?
If they're studying, understand what they are pursuing.
Inquire about any challenges they are facing.
Pitch a relevant service based on their challenges.
Offer a product link if they show interest. If not, end the conversation politely.
Example Flow 2:
First, share with excitement: "I'm thrilled to inform you that you've been selected for the Scholarship Program for our upcoming Batch!"
Inform them that the new Batch starts next week on September 27th. Ask: "Are you excited about it?"
Empathize with the user's response accordingly.
Provide them with the GForm link and ask them to submit their interest via the Form to receive Material Access by email by tonight.
Inquire if the user has any further questions. Assist as needed and END the chat if the user has no more queries.
Pro Tip:
Capitalize key action words like THANK, ASK, or PITCH to highlight important actions for the AI.
Double quote a phrase “ “ to strictly set up your pre-defined wordings.
Use END THE CHAT keyword to explicitly let the model conclude a chat accordingly.
9. Product Links
This is where you provide product or service links for the AI to share with users.
Link Description: Add a brief, descriptive title for the link.
Best Practice:
Keep it short and relevant to the Campaign Flow instruction.
If there is more than one link, try to differentiate each one properly.
Example:
“Batch 05 Pre Booking Link”
“Pro Membership Link”
“1-1 Call Link”
Link: Add a shortened link to be shared to the user.
Best Practice:
Keep it short to take lesser space in chat message.
Always include UTM parameters to track conversions.
Add price or discount info after the link.
Example:
https://bit.ly/proplan05 [Price is $999, use "10OFF" for a 10% discount]
Pro Tip: Keep product title and links short and clear. Include details about pricing and discounts to encourage conversions.
10. Initial Message
The Initial Message is the first touchpoint the AI has with the user.
Best Practice:
Make it brief, personalized, and context-driven.
Start with a greeting, introduce the persona, and ask a question relevant to the campaign’s objective.
Use variable {{Name}} to insert the name of the contact.
Example:
Hi {{Name}}, this is Riya, from Team Antern. Thanks for joining our “Top 1% ML” webinar last day!! 👍 We hope you found it valuable. As your next step, I might have something exclusive for you… 🤩 Can we chat for 2 mins?
Hey {{Name}}, we have Miami university's representative coming to our Winny Ahmedabad office today at 2:30 PM. It's an amazing opportunity to ask about degrees, admissions, and scholarships - would you be able to come by today?
11. Followup Message
The Follow-up Message is manually sent to contacts who have received the Initial Message but haven't responded.
Best Practice:
Keep it brief, personalized, and aligned with the Initial Message's context.
Be direct but avoid spamming the user.
Follow-up does not include the {{Name}} variable.
Example:
"Hi there! I'm following up in case you missed my previous message."
"Hi, hope you had a fantastic weekend. I was wondering if you're free for a quick chat now?"
C. Key Tips for Writing Flows
Use Second Person POV: Instruct the AI addressing as "you" and "your," as if speaking directly to it. Be clear but conversational, using phrases you'd naturally use with your end users.
Capitalize Action Words: Emphasize key actions by capitalizing words like ASK, THANK, or SHARE. This helps the AI prioritize important steps.
Use Quotation Marks for Specific Phrases: When you want the AI to use exact wording, enclose it in quotation marks. This clearly indicates the precise language to be used.
Example*: THANK the user for replying. ASK if they have attended any “Java and Spring Boot development” interviews recently.*
Avoid covering all edge cases in the Flow. With an 8-step limit, focus on precision. The shorter and clearer the instruction, the better it performs.
Address more edge cases by utilizing the Second Brain Manager and adding relevant, high-quality memories to the model.
By default, if the model cannot handle a query, it will ask the end user to email the support address.
Pro Tip: Testing is essential! Iterate on the prompt wording to ensure the AI effectively engages users and achieves the desired outcomes.