Personalised outreach with AI: cold email and LinkedIn messages that don't feel automated

Generative AI has made outreach cheaper and faster than ever. It has also made it noisier. Decision-makers in New Zealand, the UK and everywhere else now get a steady stream of messages that open with a compliment about a recent post, mention their company name, and pitch something unrelated. People can spot them instantly.

The opportunity isn't to send more messages. It's to use AI to do the research and preparation that makes every message relevant, then let a person shape and send it. This guide shows how, and how to stay on the right side of the rules. If you'd like help setting up an outreach workflow, see our AI automation and outreach service.

Why most AI outreach doesn't work

• Fake personalisation. Inserting a first name, company name or "loved your recent post" isn't personal. People know it's a template.
• No real reason to reach out. The message is about what you sell, not about something happening in their business.
• Generic AI writing. Long, polished, over-enthusiastic messages with the same phrases everyone else's AI uses.
• Volume over fit. Thousands of messages to a loosely defined list, which damages your sender reputation and your brand.
• Wrong facts. AI confidently mentions a job title, achievement or event that's out of date or simply wrong.

What real personalisation looks like

Personalisation is relevance. A good outreach message answers three questions for the reader within a few seconds:

1. Why me? You're clearly the right person, in the right kind of business.
2. Why now? Something has happened that makes this timely: a new role, a launch, hiring, expansion into a new market, a change in their industry.
3. Why should I care? There's a specific, believable benefit or insight for them, not a list of your services.

How to use AI in each step of outreach

1. Define who you're really looking for

Use AI to analyse your best existing clients: industry, size, role of the buyer, the problem they came to you with and what they said in their first conversation. That gives you a sharper ideal customer profile than "small businesses in NZ".

2. Research each prospect

AI can summarise a prospect's website, recent news, job posts, public content and reviews in minutes, and suggest a likely priority or challenge. Always check the facts before using them.

3. Find the trigger

Ask AI to look for timely signals: new leadership, a product launch, a rebrand, hiring for marketing roles, moving into the UK or Australia, or a poor experience customers mention in reviews. The trigger becomes your reason to reach out.

4. Draft a short, specific message

Give the AI your tone of voice, examples of messages that worked and strict limits: under 100 words, one clear point, no hype, no fake compliments. Then edit it so it sounds like you.

5. Offer something useful

Replace "Can I have 15 minutes?" with something valuable: a quick audit finding, a relevant example, a short checklist or a specific idea. Generative AI can help you create a personalised asset quickly, such as a one-page review of their website or AI visibility.

6. Follow up with purpose

Plan two or three follow-ups that each add something new, rather than "just bumping this". Stop as soon as someone replies, books or asks you not to contact them again.

7. Learn from the replies

AI can sort replies by interest, objection or timing, and spot which triggers, subject lines and offers perform best, so each round of outreach gets better.

Cold email vs LinkedIn: what works where

On LinkedIn, warm the relationship before pitching: engage genuinely with someone's posts, connect with a short note that explains why, and only move to a business conversation when there's a natural reason. LinkedIn's User Agreement doesn't allow bots or third-party software that automates activity such as sending connection requests or messages, so use AI to prepare, not to send on autopilot.

The rules in New Zealand and the UK

Outreach rules differ between countries and between email and other channels. This is general guidance, not legal advice.

New Zealand

• Unsolicited Electronic Messages Act 2007. Commercial emails and texts need consent. That can be express consent, consent inferred from an existing relationship, or deemed consent where someone has conspicuously published their work address and your message relates to their role, unless they've said they don't want marketing messages.
• Identify yourself and include an unsubscribe. Every commercial message must say who sent it and include a working, free way to opt out, and opt-outs must be honoured promptly.
• Privacy Act 2020. Collect only the information you need about prospects, keep it secure and be open about how you use it.

United Kingdom

• PECR. Marketing emails to individuals, including sole traders and some partnerships, generally need prior consent or must meet the "soft opt-in" rules. Emails to corporate addresses, such as employees of limited companies, don't need prior consent, but you must identify yourself and offer an easy opt-out.
• UK GDPR. A business email address that identifies a person is still personal data. You need a lawful basis, often legitimate interests, and should tell people where you got their details and how they can object.
• Accurate, honest messages. Don't use misleading subject lines or pretend to know someone you don't.

Whichever country you're in, keep records of where each contact came from, honour opt-outs everywhere, and be transparent if AI helps write your messages when someone asks.

Our approach: AI prepares, people connect

At Terra Nova Digital we use AI to inform, review, automate and execute, with human-centred strategy directing it. For outreach that means AI handles research, triggers, first drafts and reply sorting, while a person decides who to contact, checks every fact and sends messages that sound like you. Fewer, better messages win more conversations than a flood of automated ones.

A simple outreach workflow to start with

1. Pick one well-defined audience, such as UK leadership coaches or Auckland professional services firms.
2. Build a list of 50 well-matched prospects, recording where each contact came from.
3. Use AI to research each one and suggest a trigger and a useful insight.
4. Write short messages with AI help, then edit each one yourself.
5. Send in small batches, follow up twice with something new, and track replies and booked calls.
6. Review what worked after two weeks and refine your audience, triggers and offer.