Most SDRs try prompt engineering once, get a wall of generic AI text back, and assume it doesn’t work for real outreach.
The difference between “okay” and “pipeline-moving” prompts is simple: real sales data in, clear instructions out, and an automation layer that turns one prompt into multiple steps for AI-powered lead generation.
This guide provides seven prompt templates built around the day-to-day motions of an SDR: researching accounts, qualifying, personalizing, and following up. You’ll also learn how to pair these prompts with PhantomBuster automations so one strong prompt becomes a repeatable, scalable outreach system.
Why prompt engineering matters to SDRs right now
Prompt engineering is a practical skill for turning minimal inputs (prospect name, role, company, recent activity) into specific outputs (connection note, cold outreach message, subject lines, follow-ups) that match your brand voice and desired tone.
Sales reps are being asked to learn prompt engineering as part of their core skill set.
When you guide the AI with clear context and constraints, your message sounds intentional and relevant to the buyer. This way, you can make your offer relevant to their current priorities.
Strong prompts help you:
- Provide clear context about the prospect
- Define who the message speaks to
- Clarify the value the buyer cares about
- Remove ambiguity that leads to vague or flat responses
- Produce copy that fits your tone and workflow
When your prompts are paired with live prospect research and lead qualification signals, you create content that feels timely and relevant, not templated.
How PhantomBuster powers your prompts
PhantomBuster enriches your prompts with real prospect data by sourcing profiles, recent activity, and company signals in one workflow, so the AI writes context-rich messages.
PhantomBuster delivers LinkedIn Search Export results, profile and company details, post engagement cues, and up-to-date qualification signals directly into your prompt template automatically.
With richer inputs, the AI model produces higher-quality content, fewer errors, and more specificity pulled from actual buyer activity. This is the difference between a single prompt that creates a single message and a repeatable system that fuels your entire sales workflow.
How the 7 prompts map to a complete SDR workflow
Think prompt chaining: multiple steps where the second prompt builds on the first prompt’s output. You’ll see how to craft prompts for specific tasks, then connect them to PhantomBuster workflows to automate data collection and safe delivery.
Tip: Keep inputs structured (for example: prospect name, company, role, recent post, and pain points) so your AI assistant and automations can run multiple steps without manual copy-paste.
Prompt 1: Discover new leads that match your ICP
What the large language model (LLM) should deliver: A refined query strategy (titles, industries, company size ranges) and a short checklist of key elements to validate relevance.
Effective prompt (first prompt): “Act as an SDR. Given these ICP attributes (industry, company size, employee range, seniority, pain points), write LinkedIn Sales Navigator filters and a three-step prospecting plan. Output bullet points only.”
PhantomBuster workflow: Run PhantomBuster’s LinkedIn Search Export automation on your Sales Navigator query for automated sales prospecting, then add the Profile Data Extractor to pull titles, company, and public details within the same PhantomBuster workflow. This powers the next prompts with clean, consistent inputs. PhantomBuster enforces conservative daily limits and human-like pacing so you scale safely.
Prompt 2: Summarize each prospect with three hooks you can use
What the LLM should deliver: Three hook angles that reference the role, recent activity, or company context; a friendly tone by default.
Effective prompt: “Summarize this prospect in three bullets using role, company, and a recent post. Surface one likely pain point and one value proposition. Keep a conversational tone. Avoid buzzwords.” Paste the LinkedIn profile of the prospect or use details from the results of PhantomBuster’s data extraction.
PhantomBuster workflow: For richer outputs, add PhantomBuster’s Profile Data Extractor and Post/Activity Extractor in the same automation, then feed that context into your prompt. The AI model cites specific details that sound human and current, producing a better output.
Prompt 3: Write a connection note that references something real
What the LLM should deliver: A 220 to 280-character connection note (LinkedIn allows up to 300) that mentions a post, a shared interest, or a relevant initiative.
Effective prompt: “Write a LinkedIn connection note referencing this specific cue: {recent_post_or_comment}. Friendly tone, first-name basis, no pitch, one line.”
PhantomBuster workflow: Use PhantomBuster’s Post Engagement Extractor to capture the exact post you’ll reference, then schedule connection requests with PhantomBuster’s Network Booster using randomized timing that aligns with LinkedIn’s usage guidelines and conservative daily limits.
Prompt 4: Generate a first message tied to company news
What the LLM should deliver: A short opener that speaks to the buyer’s world (e.g., new hiring, funding, tech stack change), plus one clear call to action.
Effective prompt: “Create a two- to three-sentence cold outreach message for {prospect_name} at {company_name}. Reference {news_or_hiring_signal}, state a specific value proposition in plain language, and end with a clear call to action that asks for a 10-minute chat.”
PhantomBuster workflow: Use PhantomBuster’s Company Page Extractor and Job Posting Extractor in one workflow to surface hiring and news signals for automated lead enrichment that your prompt can cite. Enrich with role-relevant context so the AI’s output reads like a note from a peer, not a generic pitch.
Prompt 5: Produce an objection-aware follow-up
What the LLM should deliver: A follow-up that references the last touch and neutralizes a likely objection (budget, timing, existing tool).
Effective prompt (second prompt in the chain): “Write a two-sentence follow-up that references our last interaction: {previous_message_summary}. Offer one alternative next step (share a 90-second case study or send three bullet points). Keep it concise, friendly tone.”
PhantomBuster workflow: Sync CRM activity and messages into PhantomBuster, then trigger the follow-up prompt inside the same sequence. PhantomBuster helps schedule LinkedIn messages safely and hand off email steps to your mail tool in a multi-touch sequence.
Prompt 6: Craft a bump message that revives cold conversations
What the LLM should deliver: A short, polite nudge with new value, an example, a relevant doc, or a current industry trend.
Effective prompt: “Draft a short bump that acknowledges no reply, adds one relevant proof point or current industry trend, and ends with a simple yes-or-no question. One short paragraph, two to three sentences.”
PhantomBuster workflow: Trigger a bump only if there’s been no reply for X days. Mix channels: after a LinkedIn touch, schedule a cold email using the AI’s output. In our experience, well-paced multi-channel sequences win more replies than single-channel outreach.
Prompt 7: Summarize replies and update CRM status
What the LLM should deliver: A concise classification (interested, referral, not now), a summary of key points, and next steps.
Effective prompt: “Summarize this reply in three bullets. Identify intent: interested, referral, or not now. List any key differentiators they asked about, and propose the next step. Output: bullets + CRM status.”
Capture LinkedIn DMs with PhantomBuster’s Inbox Export automation, classify intent with your prompt, and sync notes and status back to your CRM automatically. This closes the loop so future prompts have accurate background information and the team avoids overlap for better CRM hygiene. This closes the loop so future prompts have accurate background information and the team avoids overlap.
How to write prompts that produce consistent, high-quality outputs
The quality of your prompts determines the quality of AI output. Treat prompts like directing a teammate: clear inputs and plain language get better results.
- Provide context: Include role, industry, current initiative, and known pain points.
- Clarify the desired output: Tell the AI the exact structure, bullets, word/character limits, or a clear call to action.
- Define tone and voice: Label desired tone (friendly, professional) and any brand voice constraints.
- Specify the one or two benefits that matter to this buyer: Name the value propositions relevant to that prospect.
- Avoid vague prompts: Replace “make it better” with “cut to 240 characters, remove technical jargon, keep one value prop.”
- Use prompt chaining: Feed the first prompt’s prospect research into the second prompt that creates the message.
- Leave room for variation: Ask for two variants so you can A/B test subject lines or first lines.
Safety, scale, and what “good automation” looks like
Once you get comfortable with prompt engineering, it’s natural to think about scaling. But scaling isn’t about blasting more messages. It’s about clean inputs, smart prompting, and avoiding the common pitfalls that trigger platform limits.
Teams that scale safely start small and mix actions like profile views, reactions, comments, and messages with real human activity. PhantomBuster’s throttles, delays, and scheduling make this easy to run day to day while keeping you in conservative ranges that avoid account warnings.
- Team operations: Assign account ownership, block duplicates, and centralize routing in your CRM. Use one approved message library with version control so prompts and outputs remain consistent across reps. PhantomBuster pairs with your CRM to keep activity clean and prevent overlap, which is critical once you have multiple seats running the same play.
- Multi-channel reality: LinkedIn plus email wins more often than LinkedIn alone. Use your LinkedIn touches to warm prospects, then follow with email that references those interactions; pace each channel appropriately.
Putting it all together in one simple play
Run this as a single PhantomBuster automation that sources leads, enriches profiles, generates messages, and schedules delivery.
- Discover: Run a Sales Navigator query, export with PhantomBuster, and extract profile details automatically.
- Enrich: Pull company signals and recent posts; clean titles and company names.
- Generate: Use the prompts above to create the connection note, first message, and two follow-ups.
- Deliver: Schedule LinkedIn touches in PhantomBuster with safe pacing, and trigger the email step in your email service provider (ESP) from the same sequence.
- Log: Summarize replies with a prompt; update status and next steps in your CRM automatically.
- Improve: Compare variants, note which prompts and subject lines outperform, and tune instructions to craft prompts that match your brand voice.
Final takeaway
Clear prompts plus live prospect data equals more replies and qualified meetings. Use the seven prompts to guide the AI’s output, and connect them to PhantomBuster to automate the research, generation, and delivery steps safely, at scale, and with a clear call to action in every message.
FAQs
What makes these prompts different from generic templates?
They’re built around prompt engineering techniques (clear instructions, specific tasks, desired tone, structured outputs) and fueled by real prospect research and key pain points via PhantomBuster. That combination improves accuracy, personalization, and reply rates because the AI systems receive better context and constraints.
How do I avoid sounding robotic when using AI prompts?
Provide context (role, company, recent post), keep one value proposition, and ask for two short variants. Specify a conversational tone and character limits. Tie each message to a specific cue so it reads like a note from a peer, not a script.
Can I scale safely without getting flagged on LinkedIn?
Yes, start small, increase gradually, randomize timing, and mix actions. PhantomBuster’s safety controls help enforce conservative limits and human-like pacing so you don’t spike suspicious activity.
Do these prompts work with different AI models and tools?
Yes. These prompts work across GPT-style models as well as Claude or Gemini because they specify inputs, tone, and format. You can run the same prompt generation steps in your preferred AI assistant.
How do I adapt prompts for different segments or industries?
Swap in segment-specific pain points and key differentiators, change the desired tone, and provide industry jargon sparingly. Keep the core structure, but update the background information and call to action by persona.
What if I only have one prompt’s worth of time each day?
If you only have time for one prompt, start with Prompt 2 (prospect summary) or Prompt 4 (first message). These usually deliver the biggest early impact. As you build momentum, add Prompt 7 to keep your CRM clean going forward.
Can I use chain of thought prompting for sales copy?
You can, but keep it lightweight. For copy, simpler prompt chaining, which involves research, message, and follow-up, usually outperforms lengthy reasoning chains. Reserve deeper reasoning for lead qualification, scoring rules, or multi-step planning.
How do I A/B test AI outputs efficiently?
Ask the AI to generate two variants with different first lines or subject lines. Track acceptance and reply rates by variant in your CRM or spreadsheet, and keep only the top performers as your default prompt templates. PhantomBuster can push variant labels to your CRM or Sheets so you can track reply rates by version.