Sponsored by
How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads
The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.
A prompt is not a magic spell. It is an instruction. And like any instruction, the quality of what you get back depends almost entirely on the quality of what you put in.
I have seen business owners spend weeks setting up automation platforms and then fill the AI step of the workflow with a prompt that says "write a follow-up email." Then they wonder why the output is generic.
The prompts in today's edition are the actual prompts we use inside real automated workflows. Not cleaned up for presentation. Not theoretical examples. The ones that are running in production and producing results worth writing about.
Each one includes what it does, where it lives in a workflow, and any variables you need to swap in for your own business. Take them, test them, improve them.
Prompt 1: The Lead Intake Qualifier
Where it lives: The first AI step after a new lead form submission.
System: You are a lead qualification specialist for [Business Name]. Your job is to assess new leads and categorize them as Hot, Warm, or Cold based on the following criteria: Hot = clear budget signal, decision-maker, specific problem match. Warm = interested but unclear timeline or authority. Cold = mismatched industry, no budget signal, or unclear need.User: Here is the lead information:Name: {{lead_name}}Company: {{company}}Industry: {{industry}}Their message: {{lead_message}}Respond with: (1) Category (Hot/Warm/Cold), (2) One sentence explaining why, (3) The best first question to ask them.
Why it works: It gives the model explicit criteria instead of asking it to guess. The three-part output structure means you can parse the response programmatically and route leads in your workflow based on the category.
Prompt 2: The Personalized First Response
Where it lives: Immediately after lead qualification, for Hot and Warm leads.
System: You write first-response emails for [Business Name]. The tone is direct, warm, and confident. No corporate speak. No "I hope this email finds you well." Get to the point fast. One paragraph max for the opener. The goal is to make them feel heard and create curiosity about a next step.User: Write a personalized first response email to:Name: {{lead_name}}Their inquiry: {{lead_message}}Why we can help: {{relevant_capability}}Next step we want: {{desired_next_step}}Output the subject line and email body separately.
Why it works: The system prompt does the tone calibration work so you do not have to include tone instructions in every user message. The explicit output format (subject line + body separately) makes it easy to feed the output directly into your email sending step.
Prompt 3: The Meeting Summary and Action Item Extractor
Where it lives: After a meeting transcript lands from your recording tool.
System: You extract structured summaries from meeting transcripts. Format your output exactly as follows:SUMMARY (2-3 sentences max)KEY DECISIONS (bullet list)ACTION ITEMS (format: [Owner] - [Task] - [Due Date if mentioned])FOLLOW-UP QUESTIONS (things left unresolved)Be concise. Cut anything that is not actionable or decision-relevant.User: Here is the transcript: {{meeting_transcript}}
Why it works: The rigid output format is the key. When you are piping AI output into a downstream system, you need predictable structure. This prompt, paired with a tool like Fathom for automatic transcript delivery, creates a full meeting-to-action pipeline that runs without anyone touching it.
Prompt 4: The Content Repurposing Engine
Where it lives: After a newsletter edition is written and approved.
System: You repurpose long-form business content into platform-native short-form formats. You understand that each platform has its own conventions. LinkedIn posts lead with a hook and use short paragraphs. Tweets are punchy and end with a question or strong statement. Do not just summarize. Pull out a single insight and make it self-contained.User: Here is the source content: {{newsletter_excerpt}}Create:1. Three LinkedIn post options (each under 200 words, each with a different angle)2. One Twitter/X thread (5 tweets, each numbered)3. One short-form video script (60 seconds, spoken word style)
Why it works: Giving the model three LinkedIn options instead of one means you are choosing from variations instead of editing a single mediocre draft. Options create selection pressure that produces better final output.
Prompt 5: The Invoice Follow-Up Writer
Where it lives: Triggered when an invoice passes a payment due date by a set number of days.
System: You write professional but firm payment follow-up messages for [Business Name]. The tone shifts based on how overdue the invoice is. At 3 days: friendly reminder, assume oversight. At 14 days: direct, reference previous message. At 30 days: formal and factual, reference next steps. Never threatening, always professional.User: Write a follow-up message for:Client: {{client_name}}Invoice number: {{invoice_number}}Amount: {{invoice_amount}}Days overdue: {{days_overdue}}Previous contact made: {{yes_or_no}}
Why it works: The tone calibration based on days overdue is built into the system prompt, so the model adjusts appropriately without you having to manage multiple separate prompts for different stages. One prompt, intelligent escalation.
Prompt 6: The Customer Feedback Analyzer
Where it lives: Triggered weekly to process reviews or survey responses collected during the week.
System: You analyze customer feedback and extract actionable business intelligence. Identify patterns, not just individual comments. Prioritize issues that appear across multiple responses. Format your output as: TOP PRAISE (what customers love), TOP FRICTION (what is causing problems), TREND ALERTS (anything increasing in frequency), and ONE RECOMMENDED ACTION (the single most impactful thing to address this week).User: Here are this week's customer responses: {{feedback_batch}}
Why it works: Most feedback review is reactive and unfocused. This prompt forces pattern recognition and ends with a single recommended action, which is the part that actually leads to change instead of a long list of observations nobody acts on.
Prompt 7: The Weekly Operations Brief
Where it lives: Runs every Monday morning, pulling from your key metrics dashboards.
System: You write concise weekly operations briefs for a business owner. The format is: WHAT HAPPENED (key metrics vs. last week), WHAT TO WATCH (two or three things that need attention), and THIS WEEK'S PRIORITY (the single most important focus). Write in plain language, not corporate jargon. Be direct about problems.User: Here are this week's metrics:Revenue: {{revenue_this_week}} vs {{revenue_last_week}}Leads: {{leads_this_week}} vs {{leads_last_week}}Open support tickets: {{open_tickets}}Deals in pipeline: {{pipeline_value}}Notable events: {{notable_events}}
Why it works: This prompt turns raw data into a narrative briefing automatically. Paired with a data integration step that pulls from your CRM and analytics tools, it means you start every week with a clear picture of the business without spending an hour pulling reports.
How to Use These in Your Stack
Each of these prompts slots into a workflow step inside Make.com or a similar platform. The curly brace variables ({{like_this}}) get replaced by dynamic data from earlier steps in the workflow, which is how the personalization works at scale. You build the workflow once, connect your data sources, and the prompts do their job every time the trigger fires.
Start with one. The lead qualifier or the meeting summary extractor tends to deliver the fastest visible ROI. Build it, test it for a week, then add the next one. Inside a month, you will have a meaningful portion of your most repetitive cognitive work handled automatically.
That is not replacement. That is leverage.
Want the full prompt library plus workflow templates? Reply ACCELERATOR to get the AI Business Accelerator ($97) which includes 40+ production-tested prompts, workflow blueprints, and a 90-day implementation roadmap.