The system behind the meetings.
Cold email that works isn't luck or volume. It's three systems running together: deliverability that reaches the inbox, data that targets the right buyers, and copy engineered to earn replies. We run all three, end to end.
Deliverability infrastructure
Most campaigns die in the spam folder before the copy ever matters. We build sending infrastructure deep enough to reach the inbox at scale, and we keep it healthy every single day.
- Dedicated inboxes across Google Workspace and Microsoft 365, with SMTP-based environments available where an architecture calls for it — never your primary domain
- Automated warmup, rotation and real-time reputation monitoring
- SPF, DKIM and DMARC configured and watched so you stay out of spam
Data & list building
The best copy sent to the wrong people is wasted. We build ICP-matched lists of the exact decision-makers worth reaching, verified before a single email goes out.
- Precise ICP and title targeting, not scraped bulk lists
- Every contact enriched and email-verified to protect deliverability
- You own the data and lists, always
Copy & campaign strategy
Opens don't pay the bills — replies and meetings do. We write messaging tuned to your exact offer, split-test it relentlessly, and manage every reply until it becomes a booked call.
- Human + AI copywriting engineered for reply rate and positive sentiment
- A/B tested every send and rewritten as the market responds
- Positive replies qualified, routed in minutes, and booked onto your calendar
The Deliverability Flywheel.
Domains don't last forever, so we never bet a campaign on one. We keep a rotation of inboxes constantly warming, sending, and retiring — so your sending reputation never drops and there's zero downtime.
The mechanics, specifically.
Any agency can say inboxes, warmup and A/B tests. These are the operating rules our system actually runs on — written by the people who run it. Updated August 2026.
How a lead becomes campaign-ready
ICP -> COMPANY UNIVERSE -> EVIDENCE CHECK -> CONTACT -> LIVE VERIFICATION -> CAMPAIGN
A campaign-ready lead is manufactured, not bought. Every list goes through the same pipeline:
- Companies are qualified from their own website evidence — what they say they sell, and to whom — not from a vendor's industry code. A company whose site is dead, parked or contradicts the targeting thesis never enters the list.
- Decision-maker titles are matched to how each market actually titles the role, not a generic seniority filter.
- Emails are verified against the live mailbox immediately before the first send, not when the list was built. A verification from three months ago is an opinion, not a fact — it is also why a "verified" list you bought can still bounce.
When vendor data is wrong
Every data vendor misclassifies companies — industry fields are the least reliable field in B2B data. Our rule: no single vendor field ever decides a send.
- A category field can route a company into review; only company-level evidence decides whether it belongs in the campaign.
- When the website and the vendor field disagree, the vendor field loses — the record is reclassified or dropped, never sent on the strength of a database label.
- Failed lookups are failures, not answers: a lead whose enrichment errored is requeued for another pass, so an API timeout never silently becomes a targeting decision.
We read the receiving side too
Most agencies only manage their own sending. We also classify who we are sending into. Before launch, recipient domains are classified from public DNS records by mail environment — Google, Microsoft, or a security gateway sitting in front of the inbox.
- Gateway-protected segments (enterprise, finance, regulated industries) behave differently from Gmail-hosted SMBs — volume, copy and expectations are set per environment, never averaged across them.
- It is the same classification method behind our published research on who actually receives B2B email.
- When MX records are ambiguous, the segment is treated as unknown rather than guessed.
How we test without fooling ourselves
Split testing is easy. Reading results honestly is the hard part.
- One variable moves at a time — ICP, offer, angle, copy, segment or follow-up depth — against a control that keeps running.
- A test is not read until its segment clears the same minimum-contact threshold we apply in our published benchmark research; below that, a "winning" variant is usually noise.
- Losing tests are recorded, not deleted — what did not work in a market is part of the targeting model, and it is why campaign two is better than campaign one.
What happens to a reply
REPLY -> CLASSIFIED -> ANSWERED -> QUALIFIED -> FOLLOWED UP -> BOOKED -> TRACKED TO REVENUE
Our service does not end when a prospect replies — that is where it starts.
- A human classifies every reply — interested, objection, referral, timing, out-of-office, no — and answers in the prospect's context. Hot replies reach you in minutes.
- "No" and out-of-office are data, not trash: referrals get routed, timing objections get scheduled follow-ups, hard no's suppress the contact for good.
- Booked meetings are followed until they show, then tracked through to closed revenue. Our revenue share depends on the far end of this pipeline — which is exactly why the middle of it is staffed by people.
What we do, and don't.
See the method run on your pipeline.
Tell us your offer and ICP. We'll show you exactly how we'd build your campaign.
Apply to work with usThe longest-running example of the method: the CTAP Media Funds programme, 75,000 emails a month since 2011.