AI Lead Generation: What Actually Works (and What's Just Noise)

Every week I see another tool promising "10x your leads with AI." Most of them do the same thing: scrape a huge list, blast the same email to everyone, and hope a few people reply. That isn't lead generation. It's spam with a better logo.

I build AI tools for a living, so I'm not against the technology. I'm against the lazy version of it. After building a handful of these systems for recruitment agencies and small businesses, here's what I think separates the useful from the noise.

Timing beats volume

The biggest lever in outbound isn't the message. It's whether the person is ready to hear it. A cold email to someone with no need is a delete. The same email to someone who has the problem right now gets a reply.

So the first job of AI in lead generation should be finding the moment, not writing the email. For a recruitment agency, that moment is a company posting a job. For a web designer it might be a business that just launched, or moved premises, or has a site that is clearly broken. The signal is public. Someone just has to be watching for it, every day, without getting bored.

That is what machines are good at.

Personal doesn't mean "Hi {first_name}"

Mail-merge personalisation fooled nobody in 2012 and it certainly doesn't now. Real personalisation means the message shows you know something specific about them: the role they're hiring for, the thing they announced last week, the problem their industry is having.

AI is genuinely good at this part, as long as you feed it real information to work from. If you give it nothing, it writes the same generic paragraph everyone else's tool writes. The quality of the output depends on the quality of what you put in.

Follow-ups are where the leads are

Most replies don't come from the first message. They come from the second or third, a few days later, when your note happens to land at a better moment. Almost nobody follows up consistently because it's boring and easy to forget.

This is the least glamorous and most valuable thing to automate. A system that quietly sends a polite nudge after a few days, and stops the moment someone replies, will out-earn a brilliant first email that nobody ever follows up.

Keep a human on the decisions that matter

There's a version of "automation" where nobody is in the loop and the software does whatever it likes. I wouldn't run a business that way. The parts I'd keep human:

• Who you choose to target. No tool knows your ideal client better than you do.
• What you promise. Anything that sounds like a commitment should be yours.
• The conversation once someone replies. That's where trust gets built or lost.

Let the software do the searching, the drafting and the chasing. Keep the judgement and the relationships for yourself.

Sending too much is a real risk

Sending hundreds of near-identical emails an hour is a fast way to wreck your domain's reputation, and once your messages land in spam, they stay there. Good systems pace themselves: a handful of messages spread across the day, with gaps between them, the way a person would send them. It's slower, and it's the reason it keeps working.

A quick test for any AI lead tool

Before you pay for one, ask:

1. Does it find people at the right moment, or just build a big list?
2. Would I be comfortable if the recipient saw exactly how the message was written?
3. Does it follow up, and does it stop when someone replies?
4. Does it pace sending sensibly?
5. Do I stay in control of who gets contacted?

If the answer to most of those is no, it's a spam cannon.

We built DCA Recruit around these ideas for recruitment agencies, and the same thinking applies to almost any service business. If you'd like to talk about what this could look like for yours, get in touch.