Cold email agency: what we run, and what we refuse to run

What a cold email agency is actually responsible for

The phrase covers four separate jobs, and the difference between agencies is usually which of them they genuinely do.

Audience

Defining the ICP, sourcing companies and people against it, and verifying that the addresses exist before anything is sent to them. Most of the quality of a campaign is decided here, before a word of copy is written.

Infrastructure

Sending domains, mailboxes, authentication records, warmup, and the ongoing monitoring that catches a reputation problem while it is still small. This is the part clients most often do not know they are buying.

Copy

The angle, the offer, and the specific message. Visible, easy to judge, and the part every agency demonstrates in a pitch, which is exactly why it is a poor way to choose one.

Reply handling

Reading what comes back, routing genuine interest to a booked meeting, and suppressing the people who asked not to be contacted again. Unglamorous, and it is where campaigns quietly leak.

One message per campaign, and no follow-up sequences

Most outbound programmes send a sequence: an opener, then several scheduled follow-ups landing in one conversation until someone replies or the steps run out. We do not do that, on any channel, for any client.

Each campaign sends exactly one message. If someone does not reply, they are not chased in that thread. They become eligible for a later campaign built on a genuinely different angle, which reaches them as a new first message rather than as pressure applied to an old one.

One message per campaign is a real constraint and it costs us volume, so it is worth being plain about why we accept it. A second message to someone who has already ignored the first reads as pressure regardless of how it is written, and on LinkedIn it is literally the same conversation, sitting directly under the message they chose not to answer. The reply we want comes from a better first message to a better-chosen person, not from persistence.

Your primary domain is never used for outbound

Cold outbound runs on separate sending domains and mailboxes that we own, warm and monitor. Your main company domain, the one your invoices and your customer support run on, is never used to send campaigns.

This matters because deliverability damage is not reversible on a useful timescale. A domain that acquires a poor sending reputation carries it for months, and if that domain is also the one your existing customers receive mail on, an outbound experiment has quietly become a business continuity problem. Keeping the two apart is not an optimisation, it is the default.

How the engagement is structured

We work on a performance basis where the engagement suits it: you pay for qualified meetings that are booked and attended, against the written definition below, rather than for sends, leads or positive replies. Retainer engagements exist too, usually where the scope is broader than meeting generation.

Which of those fits is a conversation, not a fixed menu, and pricing is agreed per engagement rather than published here. What does not change is the definition of the thing being counted, which is settled in writing before work starts.

What you approve, and what you do not have to

You approve the messaging framework once, up front: the angles, the claims we may make about you, and anything we should never say. After that we ship copy freely inside those guardrails, and only a change to the angle, the offer or the claims comes back to you.

You do not review prospect lists. List review is the single biggest cause of launch delay, and the legitimate reason behind the request is handled structurally instead: send us your existing customers, live opportunities and any accounts you want left alone, and they are loaded as a suppression list before anything sends.

Factual, claim or brand problems get fixed immediately, whenever you flag them. Style preferences are batched into the next revision.