AI Personalises Your Outreach
Generic cold outreach achieves 1 to 3 percent reply rates. Hyper-personalised outreach — referencing something specific and relevant about the prospect — achieves 10 to 30 percent. AI makes this level of personalisation achievable at scale, not just for your best prospects.
From Surface to Signal-Based
| Personalisation Level | Example | Typical Reply Rate | AI Automation |
|---|---|---|---|
| Name only | Hi [First Name], | 1-2% | No AI needed, basic merge |
| Company mention | Hi [Name], I noticed you work at [Company]… | 2-4% | Simple template, no research |
| Role-based | As a [Job Title], you probably deal with… | 3-6% | Segment-level template |
| Recent activity | I saw your post about [topic] last week… | 8-15% | AI monitors LinkedIn activity |
| Company signal | Congratulations on your recent Series B… | 12-20% | AI monitors trigger events |
| Insight-based | Your [specific thing] suggests you might be dealing with [specific problem]… | 15-30% | Full AI research and synthesis |
How It Works at Scale
Build your prospect research pipeline
For each prospect in your outreach list, a Make.com scenario runs: (1) Apollo.io API pulls the LinkedIn profile URL, recent job changes, and company data. (2) A web scraper retrieves the prospect's 3 most recent LinkedIn posts and their company's recent news. (3) Google News API searches for recent coverage of their company. (4) BuiltWith checks their technology stack for relevant signals (are they using a competitor's tool you displace?). All research data is aggregated into a structured prospect brief within 5 minutes of triggering.
Generate the personalised opening with AI
The prospect brief is passed to Claude: Write a personalised cold outreach opening for [prospect name], [title] at [company]. Research available: [brief]. My context: I am [your name] from [company], reaching out because [specific reason this prospect fits your ICP]. Write: (1) A 2-sentence personalised opener that references one specific, relevant detail from the research — not generic flattery, but a genuine observation that shows real attention. (2) The connection bridge: one sentence that connects their specific situation to the problem you solve. Avoid: vague compliments, making assumptions, and anything that could be perceived as surveillance of their personal life.
Write the value proposition and CTA
After the personalised opener, the message structure is more consistent: 2 to 3 sentences on the specific problem you solve for [their role / industry / situation], one proof point (a specific result from a similar company), and a single, low-friction ask (a 15-minute call, or simply: does this resonate with where you are right now?). AI can generate these sections at scale with role-based and industry-based variants. The personalised opener is the differentiator; the value proposition can be templated by segment.
Quality-check and send
AI generates a quality check on every message before sending: does the personalised opener reference something that is public and professional (not personal)? Is the connection between the opener and the value proposition logical? Does the message make a specific, realistic ask? Is it under 150 words? Messages that fail the quality check are flagged for human review. Messages that pass are approved for sending. Build a daily sending cap into the workflow: personalised outreach at the right volume (20 to 30 per day) produces better results than blasting at maximum volume.
Personalisation Maintained Through the Thread
The initial personalised message gets the reply; the follow-up sequence maintains momentum. AI generates follow-up messages for each prospect that maintain personalisation rather than defaulting to generic bumps: the first follow-up references the specific topic of the initial message and adds a relevant piece of content (a case study for their industry, a short insight from your experience). The second follow-up uses a different angle — a different pain point, a different proof point, or a direct ask that acknowledges the previous messages.
Follow-ups sent from the same email thread maintain context — the prospect sees the full conversation and can respond to the most relevant point. AI generates each follow-up from the original research brief and the previous message content, ensuring the thread reads as a coherent conversation rather than a sequence of disconnected template emails.
How do I avoid coming across as creepy with deep personalisation?
The line between impressive research and creepy surveillance is whether the personalisation references professional, public information or personal information the prospect would not expect a stranger to know. Referencing their LinkedIn post from last week: impressive and relevant. Referencing their personal Twitter or their home city: invasive and off-putting. Stick to professional context: their published work, their company news, their stated professional interests, and their role's typical challenges. Personalisation should make the prospect think you understand their world, not that you are watching them.
At what scale does AI personalisation stop being cost-effective?
AI personalisation is cost-effective at any scale where your outreach is targeted — a list of 1,000 well-researched prospects who match your ICP will outperform a list of 10,000 poorly matched prospects every time, regardless of personalisation level. The limit is list quality, not AI capacity. If you are targeting thousands of prospects per month, consider whether the list is well-targeted enough for personalised outreach or whether a content-driven inbound strategy would be more efficient at that volume.
Want Hyper-Personalised Outreach Automation Built?
SA Solutions builds Make.com outreach systems with AI-powered research, personalised message generation, quality checking, and follow-up sequencing — integrated with GoHighLevel or your existing CRM.
