Email infrastructure for agencies running outbound on behalf of dozens of clients

Two patterns. In the first, the client keeps their existing platform (Mailchimp, Klaviyo, etc.) and we sit underneath as the SMTP relay — the platform sends through us instead of through the platform's built-in shared infrastructure. The platform UI stays the same; what changes is which IPs and which domain the mail actually leaves on. This is the most common pattern when an existing client doesn't want to rebuild their campaign workflows.

In the second pattern, the agency builds their own thin layer on top of us (often using MailWizz, Mautic, or a custom React admin) and the client logs into the agency's branded platform instead of Mailchimp. Pricing favours this pattern past a certain scale because the platform fees compound, but it's a bigger build.

30 days of monitoring at p=none with aggregate reports collected and parsed (we run the parser, you get the readable summary) to identify every legitimate sending source for that client. Most clients have 2-4 sources that no one in the agency knew about: a forgotten Mailchimp instance from a previous agency, a CRM doing transactional sends, a HubSpot landing-page confirmation flow, the office 365 tenant. All of those need DKIM-aligned authentication before the policy can move forward.

Then 30 days at p=quarantine to confirm nothing legitimate is getting caught. Then p=reject. Total time: ~90 days for a clean migration. Faster if the client's infrastructure was already simple; sometimes 4-6 months if there are vendors who need to be replaced because they can't authenticate properly.

Domain reputation moves with the domain (it's a property of the sending domain, not the infrastructure). IP reputation does not — moving to new IPs means going through the 14-28 day warmup. We typically run dual-send during the warmup: the existing platform handles maybe 80% of traffic while the new IPs ramp from 10% on day one to full capacity by day 21. By week 4, traffic is fully on our infrastructure and the old contract can be cancelled.

If the client's previous deliverability was poor (DMARC was at p=none with lots of failures, or they were on a shared IP pool with bad neighbours), the move often improves reputation rather than maintaining it. We've seen first-month inbox placement gains of 8-15 percentage points for clients who were on the worst of the shared-pool platforms.

This is the conversation our reports are built to support. Per-client send logs, complaint rates by recipient domain, DMARC failure reports, FBL data, and DNSBL listings are all attributed back to the originating client. When a client claims "your deliverability is bad," you can show them their own complaint rate climbed past 0.30% three weeks ago because they uploaded a list from a conference badge scan. The data is yours; we just make sure it's collected and presentable.

For agencies, this is often the value that's hardest to communicate up front but most valuable in retention. The clients who were going to fire you because deliverability dropped will stay when they see the data showing it was their actions, not your competence.

Stockholm phone number, a Slack channel for the agency's deliverability lead, and a CC on the operator inbox. On critical incidents (multiple clients affected, Microsoft block, /24 listing), the on-call rotation responds within an hour during EU business hours; off-hours response is typically 2-3 hours for critical, longer for non-urgent. For agencies on the Managed Deliverability retainer (+€1,200/mo), the response and the relationship are tighter — there's a named deliverability engineer who's on your account.

Standard chain: the client is the controller, you (the agency) are the processor, we're the sub-processor. Our DPA is built for this pattern and includes the standard contractual clauses for any data movement (which is rare anyway — Stockholm primary, Frankfurt secondary, no transfer outside EU by default). For French and Spanish agencies, the chain works under the equivalent RGPD framing. For German agencies, the AVV per Art. 28 DSGVO is the canonical form.

Yes. The sending hostnames, HELO, Received headers, FBL configuration, and rDNS PTRs all use your agency's branded hostname. The dashboard can be served on a subdomain you own (e.g. mail.your-agency.com) with your branding. The only places we appear are inside the legal documents at the bottom of the chain — typically not visible to the end client. Most agencies present us internally as "our deliverability infrastructure partner" and never name us to clients.