Context

Cedar Rose provides UBO and KYB verification across the Middle East, Africa and Central Asia. Their buyers are compliance, risk and onboarding leads inside banks, payment companies, logistics groups and professional services firms. These are not people who respond to volume. They are senior, they are busy, and they have seen every template that has ever been sent.

I run the whole outbound programme for them. Not just the copy, and not just the sending. Everything from buying the domains through to the file their sales team opens on a Monday morning.

Where outbound usually breaks

Most cold email programmes do not fail because the copy is bad. They fail for duller reasons, and almost all of them happen before anyone reads a word.

• Sending from the company's main domain, so one bad campaign damages the reputation of every invoice and contract that domain sends.
• Mailboxes pushed to volume before they are warm, which lands the whole account in spam.
• One generic message aimed at three very different buyers, so none of them recognise themselves in it.
• Replies arriving in a mailbox nobody watches, going cold before anyone answers.
• Reporting that shows open rates but never becomes a list of people worth calling.

Goal

Build a sending operation that runs every day without supervision, keeps the client's primary domain untouched, and hands the sales team something they can act on rather than a dashboard they have to interpret.

The sending infrastructure

I manage around 100 mailboxes spread across 10 dedicated sending domains, bought and configured specifically for outbound and kept entirely separate from the client's main domain. If a campaign ever goes wrong, it goes wrong on a domain that exists only for that purpose.

Domains and mailboxes are provisioned in Mailforge with the authentication records set correctly from day one, then handed to Warmforge before they send anything real. Warm-up is not a box to tick. It decides whether the next three months of work reaches an inbox or a spam folder, so mailboxes stay on a conservative daily volume and the load is spread thinly across all of them rather than concentrated in a few.

Data, segments and targeting

The programme runs across three client segments, each treated as its own campaign rather than a filter on one big list. Fintech and payments, trade and logistics, and professional services all buy verification for different reasons, and a message only works if it names the reason that applies to the person reading it.

Lists are built and enriched before anything is written. Each prospect gets a role summary, the pressures that role actually carries, and a personalisation anchor tied to specific evidence rather than a guess. Rows where the evidence is thin get scored down and dropped instead of padded out with filler, because a weak line costs more than a missing one.

The sequence

Each segment runs a six-touch sequence built in Salesforge, mixing LinkedIn messages and email so the name is already familiar by the time the second email lands. The voice is deliberately peer to peer. No pitch language, no superlatives, no manufactured urgency.

Copy passes through review gates before it can go live: a banned-phrase sweep that strips the usual outbound tells, a per-message checklist, and a final pass that renders real sends with real merge fields so nothing ships with an empty variable in it. It is unglamorous work, and it is the difference between a sequence that gets replies and one that gets reported.

• Six touches per prospect, split across LinkedIn and email.
• Angle and trigger scored per row before a single message is written.
• Programmatic banned-phrase check on every variant.
• Rendered test sends reviewed by a human before the import goes live.

What happens when someone replies

A reply is the one moment in outbound that is genuinely time sensitive, so none of it waits on somebody remembering to check a mailbox. Webhooks push replies straight into Slack as they arrive, with enough context in the message that the sales team can answer without opening another tool.

The same automation handles the tidying up behind it. Contacts are enriched, deduplicated and pushed into the CRM, so the record of a conversation exists whether or not anyone remembered to create it.

Reporting that gets used

This is the part I am most pleased with. Engagement data on its own is decoration, so I built a daily report that turns it into work. Every morning it collects the people who engaged with a campaign the previous day, matches them back to the research behind their sequence, and writes a HubSpot-ready file with the pain point, trigger and segment already filled in against each name.

The sales team opens it and starts calling. Nobody cross-references a dashboard against a spreadsheet to work out who is worth a follow-up. Alongside it, a weekly report reads the campaign numbers off the Salesforge dashboard and turns them into a client-facing summary: goal against actual, a channel breakdown for email and LinkedIn, and the rates that matter, all as live formulas rather than pasted values.

Stack

Salesforge for sequences and sending, Mailforge for domains and mailboxes, Warmforge for warm-up, Leadsforge for list building. Around them sit the parts that make it a system rather than a set of tools: webhooks, Slack alerts on replies, contact enrichment, a database of engagement history, and generated spreadsheets the CRM can swallow.