SELECT * FROM leads WHERE source NOT IN ('apollo', 'zoominfo', 'lusha');

Go-to-market,
engineered
end to end.

I find markets in places data vendors don't look and turn them into a repeatable system that closes revenue.

→ see the work

$2M+ ARR closed · $10M+ pipeline · 14+ countries

stack
  • Clay
  • HubSpot
  • Claude
  • Smartlead
  • Instantly
  • GetSales
  • Make
  • Python
  • Google Maps
  • AI Ark

case_studies · 10 rows · open a row for the outcome and the system

#target marketdata sourceoutcome
Outcome
$1.4M ARRclosed in 3 months
18.1%positive reply rate
The system

Problem: B2B databases barely cover psychiatry. Buy a list and half of it is the wrong specialty.

  1. Start from the source of truth. Every licensed US provider is on a public registry. That became the base list, not a data vendor.
  2. Keep only real clinics. AI read each listing and dropped anything that wasn't psychiatry.
  3. Cover the map. Google Maps sweeps across 31 states found the practices the registry missed.
  4. Write to the clinic, not only the doctor. Front-desk inboxes replied as well as named contacts.
Provider registryGoogle MapsClayClaudeSmartlead
Outcome
~10 clientssigned on ~$5k/month retainers
15–20meetings booked every month
The system

Problem: Lead generation is a broad offer. Everyone needs it, so the hard part is knowing who will actually sign.

  1. Learn from the wins. Worked out which kinds of companies signed, and why.
  2. Target four markets. US, UK & Ireland, Australia and Israel, each with its own list.
  3. Test the angle. Offers and copy rotated per market until one pulled ahead.
  4. Route every reply. Straight into HubSpot, so nothing went cold.
ClaySmartleadHubSpotClaude
Outcome
£180kpipeline in 3 months
121positive replies in 6 weeks
The system

Problem: A venue that doesn't sell tickets online will never buy ticketing software, and no database tells you which ones do.

  1. Sweep the country. Venues, comedy clubs and arts centres, city by city across the UK, plus US directories.
  2. Read every website. Detected the CMS and ticketing platform each venue already runs.
  3. Only contact a fit. Venues whose setup matched the product got an email. The rest were never touched.
  4. Add festivals. A second segment across the UK and US.
Maps scraperTech detectionClayClaudeSmartlead
Outcome
~65%of Germany's ~80k clubs mapped and enriched (51,445)
3positive replies in the first 3 days
The system

Problem: Germany has no national club database. Each of the 16 states runs its own, on different software.

  1. One scraper per federation. Around result limits, session tokens and signed forms.
  2. Enrich. Contact details clubs must publish by law.
  3. Qualify and segment. Larger and multi-sport clubs, split into 3 ICPs.
Python scrapersImpressum enrichmentHubSpot
Outcome
2 clientssigned in 2 months, one retained 1+ year
137positive replies
The system
  1. Scrape. Restaurants from Google Maps and TripAdvisor.
  2. Classify. AI sorted thousands of sites into verified e-commerce stores.
  3. Personalise. Every email referenced the store's own Instagram.
Google MapsTripAdvisorAI classification
Outcome
30 meetingswith CMOs in 6 weeks
22.1%reply rate
The system
  1. Reverse-engineer the ICP. Read the agency's best case studies to see where they had proof.
  2. Build the list. Exactly those companies in Bulgaria, and their marketing leaders.
  3. Lead with proof. A creative email built around similar work.
AI ArkEmail
Outcome
58opportunities ahead of the trade show
31.7%reply rate, best campaign
The system
  1. Scrape the show. Exhibitor and supplier directories.
  2. Merge. One deduplicated company list.
  3. Split by role. CEO, Plant, Quality and Facilities each got their own angle.
Outcome
4timed campaigns from lists nobody else had
$0.007to qualify each company
The system
  1. Internship season. Companies likely to run internships, qualified before sending.
  2. Award lists. Companies named Top Employer.
  3. Truck Driver Appreciation Week. Every US trucking company, then their HR people.
  4. Events. Conference attendee lists.
AI web researchEmail waterfall
Outcome
4 meetingsin 1 month
37.5%reply rate on LinkedIn
The system

Problem: "Want to buy property?" has no timing signal and no intent data.

  1. Pick a persona, not a signal. Senior salespeople respond to a direct pitch from a fellow seller.
  2. Run LinkedIn. Connection requests into conversations.
LinkedIn
Outcome
100%of the licensed market mapped
896contacts reached in 3 languages
The system
  1. Use the regulator as the database. Licensed managers are a closed group.
  2. Webinar-led outreach. In German, French and English.
Regulator registersLinkedIn

// clients described by what they do, never by name.

lifetime results
$2M+ARR closed
$10M+pipeline created
250k+orgs scraped & enriched from public sources
14+countries targeted
20+clients worked with
10+industries targeted
systems

The engineering behind the numbers.

Outbound only compounds when it runs on systems. These are the pieces I build and hand over.

sources
scrape
qualify
enrich
send
CRM
01 / tam_map

TAM mapping

The whole market, not a sample. Custom scrapers pull every club, clinic or venue from registries, federations, directories and Google Maps, then fill in the data. Built once, rerun whenever the market moves.

02 / qualify

AI qualification

Every account checked against the ICP before anyone gets an email: tech stack, size, category, intent signals.

03 / send

Sending infrastructure

Multi-domain, multi-mailbox setups with deliverability monitoring and auto-pause on bounces.

04 / hubspot

HubSpot & CRM

Pipelines, properties and automations mapped to how the team actually sells, so every reply lands somewhere useful.

"In my 40 years of experience, this is the fastest service with actual results I've seen ever."CEO, US mental-health services company