Scalable Cold Email Infrastructure for Agencies

For an agency, scalable cold email infrastructure does not mean more emails from the same mailboxes. It means you can add a client, pause a client, or lose a client with little or no effect on the others. The way to get there is isolation: each client gets its own dedicated domains, its own official Google Workspace mailboxes, and its own admin boundary, all sized from that client's daily target.

This guide covers how to structure that, how to size it, and the routines that keep it healthy as the client list grows.

Most agencies start with one shared pool. A handful of domains, a few dozen mailboxes, every client's campaigns rotating through them. It works until one of these happens:

• One client uploads a dirty list and bounces spike across the pool.
• One client's offer draws spam complaints and every domain takes the hit.
• A client leaves and wants their domains, but those domains also sent for three other clients.
• A Workspace account is suspended and the outage spans every campaign at once.

The common thread is shared reputation. When clients share domains, they share consequences. Why Cold Email Domains Burn and How to Slow It explains the mechanics.

Give every client a self contained pool:

1. Dedicated sending domains close to the client's brand, never the client's primary domain. See Primary Domain Isolation for Cold Email.
2. Official Google Workspace mailboxes on those domains, two or three per domain for most teams.
3. DNS on every domain: SPF, DKIM, DMARC, and MX live before warmup.
4. Its own admin boundary, so access, 2FA, and offboarding for that client never touch another client.
5. Its own sequencer workspace or campaign group, with mailboxes tagged to that client only.

Scaling then becomes repeating the same unit. A new client is a new pool. A paused client is a paused pool. A client who leaves takes their pool, or it is retired, and the other pools carry on as before.

Work backwards from the client's daily target.

• New contacts per day. Say 300.
• Total daily sends. A three step sequence that is running sends roughly two to three times the new contact count. Plan for about 750.
• Per mailbox cap. Many operators keep cold sends in the range of 20 to 40 per mailbox per day once warmed. At 25, that is 30 mailboxes.
• Domains. At three mailboxes per domain, that is 10 domains.

Add a small buffer of warmed mailboxes so you can rotate one out without missing volume. The full formula lives in How Many Mailboxes Per Domain for Cold Email, and the per mailbox side is in How Many Cold Emails Per Mailbox Is Safe.

Resist the pitch to hit the same number with five mailboxes. The cap per mailbox is what keeps the pool alive, and it is the first thing that breaks when an agency tries to save on seats.

At ten clients, memory stops working. Use a convention and a sheet.

Naming. Domains that read naturally for the client's brand. Mailbox names that match real people or real roles on the client side. A client code tag in your sequencer for every mailbox.

Tracking sheet per client. Domain, mailbox, warmup start date, current daily cap, status, and last change. Review it weekly with the campaign operator.

1. Agree the daily target and the sizing above.
2. Choose domains: the client brings their own, or you buy new ones, with registration in a name both sides agree on.
3. Provision Google Workspace mailboxes and publish DNS.
4. Start warmup and wait two to three weeks before campaign volume.
5. Connect mailboxes to the client's sequencer workspace using the methods that tool supports.
6. Start at a low cap and raise it gradually while watching replies and bounces.

1. Pause campaigns on that client's pool.
2. Decide with the client whether the domains transfer to them or are retired.
3. Remove mailboxes from your sequencer.
4. Revoke agency admin access or close the Workspace accounts.
5. Update the tracking sheet.

Because the pool was isolated, nothing else on your roster has to change.

• Check bounce and reply rates per domain, not just per campaign.
• Look for one domain behaving differently from the rest of its pool.
• Confirm DNS still passes after any domain or registrar change.
• Rest or rotate a struggling domain before it drags the pool down.
• Judge results on replies from real people. Open rate is a weak signal; see Why Open Rate Misleads Cold Email Teams.

Instantly and Smartlead both handle campaigns, rotation, and caps well for agencies, and many agencies run one or the other. The sequencer is not where isolation comes from, though. Isolation lives in the domains and mailboxes underneath. Keep the infrastructure layer separate and you can change sequencers later without rebuilding a single client pool. See Mailbox Infrastructure for Smartlead Explained and Mailbox Infrastructure for Instantly Explained.

ColdMail is the infrastructure layer for this model. It provisions official Google Workspace mailboxes on dedicated domains (bring your own or buy through ColdMail), publishes SPF, DKIM, DMARC, and MX automatically, and gives you an official admin panel per domain with 2FA. Multi organization support, for keeping client pools apart, is available on the plans that include it; see coldmail.app for plan details. Provider health indicators cover Gmail, Outlook, and Yahoo, and AI warmup is available as an optional add on (see coldmail.app for current add on pricing).

ColdMail does not run campaigns. Mailboxes are ready to connect in Instantly and Smartlead using the methods those tools support, and work with n8n and other tools.

If you are planning pools for a growing client list, Book a Strategy Call to map domains and mailboxes per client, or get started at coldmail.app.