How I Built a Prospecting Machine That Finds, Qualifies, and Outreaches to Clients — Automatically
If you run a digital marketing agency, you already know the problem. You spend half your time doing the work for clients, and the other half trying to find the next client. It's a grind that never ends — and the moment you get busy with delivery, your pipeline dries up.
I've been building and running OUNTERNET for over six years across three markets — UK, USA, and Canada. For most of that time, lead generation was manual. Research a company, find the right contact, write a message, send it, follow up, repeat. Hours of work for every prospect. Most of them never replied.
Last year I decided to fix this properly. Not with a VA. Not with a simple email sequence tool. With a real, end-to-end automated system that does the prospecting, the research, the qualification, and the outreach — while I focus on actually running the business.
Here's how it works.
The goal wasn't to send more emails. It was to send fewer, better ones — to exactly the right people, at exactly the right time.
Step 1 — Finding the Right Prospects with Apollo.io
Every good outreach campaign starts with a clean list of the right people. That sounds obvious, but most agencies skip this step and just spray messages at anyone with a LinkedIn profile.
I use Apollo.io to build targeted lists based on specific criteria — company size, industry, tech stack they're using, location, and job title of the decision-maker. For agency clients, I'm typically targeting e-commerce founders and marketing directors at companies with 10–200 employees who are already spending on paid advertising.
Apollo gives me verified contact data including email addresses and LinkedIn profiles. But raw contact data is just the starting point. Knowing someone's name and email tells you almost nothing about whether they're a good fit, what their current challenges are, or how to talk to them.
That's where the next step comes in.
Step 2 — Deep Enrichment with Clay
Once the prospect list is built, every contact flows automatically into Clay — one of the most powerful enrichment tools available right now. Clay pulls in data from dozens of sources simultaneously.
For each prospect, I'm enriching with:
Their LinkedIn activity — recent posts, what they're talking about, any signals of pain or intent
Company growth signals — are they hiring? Expanding into new markets? Recently funded?
Tech stack — what tools are they already using? Are they on Shopify? Running Google Ads?
Ad spend signals — are they actively advertising, or have their campaigns gone quiet?
Recent news about the company — product launches, partnerships, leadership changes
This enrichment step is what makes everything downstream actually work. Instead of a generic message, you now have a full picture of the person you're reaching out to. You know what they care about. You know what's changing in their business. You know what they might be struggling with.
Why this matters: Clay replaces what used to take 20–30 minutes of manual research per prospect. The system does it in seconds, for hundreds of prospects simultaneously.
Step 3 — AI Qualification and Message Writing with Claude
This is where the system gets genuinely intelligent. After enrichment, each prospect's data flows into an AI agent I built on the Claude API.
The agent does two things. First, it qualifies the lead. Based on everything Clay has gathered, it scores the prospect against our ideal client profile — do they have the budget signals? Are they in a market we serve? Is the timing right based on their growth signals? Leads that don't meet the threshold get filtered out automatically. No human reviews them.
Second, for leads that qualify, the agent writes a personalised outreach message. Not a template with a first name swapped in. A genuinely specific message that references something real about their business — a recent campaign they ran, a market they're expanding into, a problem that businesses in their stage typically face.
The AI doesn't just fill in a template. It reads the context and writes something that sounds like it was crafted by someone who actually did the homework.
I spent considerable time engineering the prompts to get this right. The messages are kept short — three to four sentences maximum. They reference one specific, real detail about the prospect. They ask one simple question rather than pitching a full service. And they sound like a human wrote them, because in a sense one did — I designed the thinking process, the AI executes it at scale.
Step 4 — Delivery and Follow-Up via GoHighLevel
Qualified leads with their personalised messages land automatically in GoHighLevel — my CRM and outreach platform. GHL handles the actual sending, the follow-up sequence, and the tracking.
The sequence is simple and deliberately not aggressive:
Day 1: The personalised AI-written message goes out
Day 4: A short follow-up if no reply — adds one new piece of value
Day 10: Final message — keeps the door open, no pressure
When someone replies, GHL notifies me immediately and pauses the automation so I can take over personally. The human conversation starts exactly where the AI left off.
Step 5 — n8n as the Backbone
Everything above is connected and orchestrated by n8n, which I run self-hosted. n8n handles all the plumbing — triggering the Clay enrichment when new Apollo data arrives, passing enriched data to the Claude API, pushing qualified leads to GHL, handling errors, retrying failed steps, and sending me a Slack notification whenever a prospect replies.
The reason I use n8n rather than a simpler tool is control. I can see exactly what's happening at every step, debug anything that breaks, and modify the logic without being locked into another platform's limitations.
The result: A system that runs 24 hours a day, seven days a week. I check Slack for replies. That's it.
What This Actually Changed
Before building this, finding new clients was something I had to actively work on every week or let slip when delivery got busy. Now the pipeline fills itself.
The quality of conversations has improved too. Because the outreach is genuinely personalised and well-timed, the reply rates are meaningfully higher than anything I achieved with manual outreach. And because the AI pre-qualifies leads, I'm only having conversations with people who are actually a good fit.
The cost per acquired client dropped significantly — from manual outreach that was costing £150–250 per client in time and effort, to a system that scales without adding to that cost.
If you're running an agency and still prospecting manually, this is the kind of system worth building. It takes time to set up properly, but once it runs, it keeps running.
About the author
Omid Akrami is the founder of OUNTERNET Digital Solutions, an automation and growth consultancy serving e-commerce and B2B brands across the UK, USA, and Canada. He builds marketing automation systems, AI agents, and Amazon growth strategies for clients who want results without the manual work.
→ ounternet.agency · linkedin.com/in/omid-akrami · omidakrami.com