Hosting for Cold Email Agencies
infrastructure built for the operational complexity agencies actually run.

Why mainstream ESPs do not work for cold email

Cold email agencies who tried to operate on SendGrid, Mailgun, Postmark, Mailjet, or similar mainstream ESPs encounter a predictable pattern. The account opens. Initial campaigns begin sending. Complaints arrive from a small percentage of recipients. The ESP's abuse system detects the complaint rate. The account is reviewed against terms of service. Cold email is found to be prohibited. The account is terminated. Sometimes a refund is offered for unused service; sometimes not.

The cycle is structural rather than personal. ESPs operate under bank and processor relationships that punish abuse aggressively. Cold email naturally produces higher complaint rates than double-opt-in marketing email because the recipients did not request the messages. Higher complaint rates impose costs on the ESP across all customers (reputation damage on shared IP pools, processor chargeback exposure, regulatory attention). The ESP responds by prohibiting the category at the terms level, allowing them to terminate any account that exhibits the pattern.

The economic logic is simple. An ESP serving a $50/month customer cannot afford the abuse risk that customer's cold email might generate. The risk is asymmetric: the customer's revenue caps at $50/month, while a single high-profile complaint can cost the ESP thousands of dollars in shared-IP reputation damage spread across all customers on that pool. The terms-of-service prohibition is the rational response to the risk asymmetry.

Agencies thus need infrastructure outside the mainstream ESP segment. The options narrow to: self-hosted SMTP on generic VPS, dedicated SMTP infrastructure from offshore providers, mailbox-based sending through Google Workspace and Microsoft 365 (subject to their own terms but operationally tolerated up to certain volumes), or hybrid approaches combining the above.

The infrastructure architecture that actually works

A working cold email agency infrastructure has several layers that interact in specific ways.

Layer 1: Sending domains

Each client needs multiple sending domains to spread campaign risk. The standard pattern is 5-20 sending domains per active client, each with its own SPF, DKIM, and DMARC records. The domains are typically variations on the brand name or themes adjacent to the campaign topic. Domain purchase happens through bulk registration at registrars that tolerate this pattern (some refuse it).

Layer 2: Mailboxes or SMTP infrastructure

Two operational approaches dominate. Mailbox-based: each sending domain has 1-5 mailboxes (typically Google Workspace at $6/mailbox/month or Microsoft 365 at $6/mailbox/month). Sending happens through the standard Gmail/Microsoft SMTP relays. SMTP-based: dedicated SMTP servers with dedicated IPs, custom rDNS per IP, and direct delivery to recipient mailbox providers. Some agencies use both: mailbox-based for very small campaigns where inbox placement is critical, SMTP-based for volume campaigns.

Layer 3: Campaign management software

The actual campaign sending happens through specialized tools: Instantly, Smartlead, Reply.io, Lemlist, Apollo, or custom-built platforms. The campaign tool handles list management, message personalization, sending cadence, reply detection, and reporting. The campaign tool connects to the mailbox/SMTP layer via standard protocols.

Layer 4: Warmup infrastructure

New mailboxes and new IPs need warmup before being used at production volume. Mailbox warmup typically uses engagement networks where multiple agencies' mailboxes interact with each other, building positive engagement signals at receiving providers. SMTP warmup uses engaged seed traffic to build IP reputation at receivers.

Layer 5: Suppression and complaint management

Recipients who unsubscribe, bounce, or complain must be excluded from future campaigns. The suppression list is operationally critical. Inadequate suppression management produces compounding complaint rates that eventually trigger termination at mailbox providers or upstream networks.

Layer 6: Monitoring and reporting

The agency needs visibility into campaign performance per client: bounce rates, complaint rates, open rates (where measurable), reply rates, blacklist status. The monitoring layer typically aggregates data from multiple sources (SMTP logs, Postmaster Tools, SNDS, FBL feedback) and presents it in client-facing dashboards.

The mailbox-based vs SMTP-based architecture decision

Most cold email agencies in 2026 operate primarily mailbox-based architectures rather than direct SMTP. The reasons are operational.

Property	Mailbox-based	SMTP-based
Deliverability	Inbox-to-inbox reputation	IP reputation + authentication
Volume per sender	50-100 daily (sustained)	10K-100K+ daily
Setup complexity	Lower (mailbox provisioning)	Higher (infrastructure required)
Cost per message	Higher ($6/mailbox base)	Lower (fixed infrastructure)
Warmup complexity	Engagement networks handle	Custom warmup required
Risk profile	Mailbox termination	IP blacklisting
Compliance	Subject to Google/MS terms	Subject to receiver acceptance
Audience tolerance	Strict (Google enforces)	Looser (your responsibility)
Best for	High-touch B2B campaigns	Volume B2C/B2B mixed

The mailbox-based architecture has gained dominance because Google Workspace and Microsoft 365 mailboxes effectively serve as deliverability proxies. A message from a Gmail-hosted account to a Gmail recipient benefits from intra-network reputation that an external SMTP server cannot match. The same logic applies to Microsoft 365 mailboxes sending to Outlook/Hotmail recipients. The cost premium per message is offset by deliverability advantages that direct SMTP cannot replicate at small per-mailbox volumes.

The SMTP-based architecture remains operationally important for two specific scenarios. First, when sending volume per mailbox exceeds what mailbox-based architectures can sustain (~50-100 daily per mailbox before deliverability degrades). Second, when the agency or client has specific requirements that mailbox terms prohibit (certain content categories, certain audience types, mass-scale campaigns).

Where dedicated SMTP infrastructure fits in the agency stack

Our role as infrastructure provider for cold email agencies is specifically the SMTP layer when agencies need it. We do not replace mailbox-based architectures; we complement them.

Agencies typically use our infrastructure for: SMTP-based campaigns at volume that mailbox-based cannot sustain, fallback SMTP when mailbox accounts get terminated mid-campaign, specific clients with content or audience profiles incompatible with mailbox terms, geographic-targeted campaigns where regional SMTP improves deliverability, and operational testing of new IP and domain combinations.

The pricing model fits the agency segment. Our €149/month Scale tier includes dedicated IPs, managed warmup, FBL processing, and the operational tooling. Additional IPs are €8/month each, which is operationally cheap compared to per-message ESP pricing. A typical 10-client agency running mixed mailbox + SMTP architectures runs €400-€800/month total on our infrastructure for the SMTP portion, with the mailbox infrastructure handled separately through Google/Microsoft.

The 14-check pre-flight on every IP

One of the operational properties that distinguishes purpose-built email infrastructure from generic VPS hosting is IP reputation hygiene at assignment time.

Our IP acquisition process includes a 14-check pre-flight on every IP before customer assignment. The checks include: Spamhaus SBL listing, Spamhaus XBL listing, Spamhaus PBL listing, Spamhaus DBL (domain check on associated reverse DNS), SURBL listing, Microsoft SNDS reputation score, Google Postmaster Tools historical data, SpamCop listing, UCEPROTECT listing, Barracuda Reputation Block List, Senderbase reputation, Composite Block List, recent reverse DNS history, and historical PTR record patterns.

An IP that fails any of these checks is not assigned. The check is rerun before every new customer assignment because reputation moves over time. The process means agency customers receive IPs that start with clean reputation and can be ramped through normal warmup rather than starting from a damaged baseline.

This is different from how generic offshore VPS hosts assign IPs. The generic host typically assigns from a pool with no pre-flight check, so customers may receive IPs with historical damage from previous tenants. The damage shows up empirically when sending begins: mail goes to spam at specific receivers, blacklist notifications arrive, complaint rates climb without cause. The customer experiences this as "deliverability problems" without realizing the underlying IP reputation issue.

How managed warmup actually works for agencies

IP warmup for cold email is operationally different from warmup for opt-in marketing email. Understanding the difference helps agencies plan timelines and expectations.

Opt-in marketing email warmup ramps a new IP from cold to full volume against the actual subscriber list. Initial sending is to most-engaged subscribers (people who have opened recently, clicked, replied). The receiving providers see positive engagement signals (opens, clicks, low complaints) and build positive IP reputation. Volume grows by 1.5-2x per day from a small starting volume.

Cold email warmup cannot use this pattern because cold lists by definition do not have engagement history. Sending cold from a cold IP produces compounding negative signals: high bounces because the list is unverified, complaints because recipients did not request mail, low engagement because the recipients are not pre-qualified. The receiving providers build negative IP reputation from the very first sends.

The working pattern for cold email warmup uses engaged seed traffic to build positive reputation first, then introduces production cold sends gradually. The seed traffic is messages to mailboxes that will reliably engage positively (open, sometimes reply, never complain). These mailboxes belong to the warmup system rather than to actual prospects. Once the IP has positive reputation established from the seed traffic, cold sends are introduced at small percentages and ramped up as the IP tolerates them.

Our managed warmup operates this pattern across customer IPs. The seed pool is maintained centrally. Customers do not need to operate their own warmup infrastructure. The warmup typically runs 30 days for B2B cold outreach use cases, longer for B2C newsletter use cases where receiver tolerance is stricter.

What we ask new agency customers about

The onboarding process for agency customers includes specific operational questions that determine the infrastructure setup.

Sending volume per client

How many messages per day per client? This determines whether dedicated IPs per client are warranted (typically yes above 5K/day per client) or shared SMTP infrastructure suffices for smaller clients.

Audience and content profile

B2B or B2C? Cold outreach to business addresses or marketing to consumer addresses? What is the content profile (services, products, recruiting, partnerships, fundraising)? Different profiles produce different complaint rates and require different infrastructure approaches.

Existing sending infrastructure

Is the agency currently sending through mailboxes (Google/Microsoft) and adding SMTP, or migrating away from another SMTP provider? The starting state affects warmup approach and migration timeline.

Authentication setup

Do client domains have SPF, DKIM, and DMARC properly configured? Authentication is mandatory in 2026 per Gmail/Yahoo/Microsoft enforcement. Customers with proper authentication can start warmup immediately. Customers with broken authentication need DNS work first.

Compliance position

How does the agency handle suppressions, opt-outs, and complaint signals? The infrastructure can support whatever the agency operates, but reasonable practices reduce abuse complaints and improve sustained deliverability.

Campaign management tooling

What software does the agency use to actually run campaigns? Different tools have different integration patterns with our SMTP infrastructure. Most major tools support custom SMTP and integrate cleanly.

The economics of agency infrastructure

Healthy agency economics require infrastructure costs to be a manageable fraction of client revenue. Understanding the cost structure helps agencies build sustainable pricing.

Typical client revenue ranges

Small agencies: $1K-$3K per client per month. Mid-size agencies: $3K-$10K per client per month. Enterprise-focused agencies: $10K-$50K+ per client per month. Revenue determines what infrastructure investment is sustainable per client.

Infrastructure cost per client

Mailbox-based: 5-20 mailboxes per client at $6/each = $30-$120/month in mailboxes. Domains: 5-20 sending domains at $1-$2 per domain per year amortized = $1-$5/month. SMTP-based add: dedicated IPs at €8/IP/month + base infrastructure share = €40-€160/month per client typically. Total infrastructure per client: $70-$280/month depending on architecture.

Healthy infrastructure ratio

Infrastructure costs should be 10-20% of client revenue for healthy agency economics. $3K client revenue → $300-$600 infrastructure budget. $10K client revenue → $1K-$2K infrastructure budget. Agencies operating outside this ratio either price too low (losing money on infrastructure) or price too high (clients leaving for cheaper alternatives).

Volume scaling efficiency

Fixed infrastructure costs (base SMTP infrastructure, FBL relationships, warmup systems) spread across more clients improves agency margins. Adding the 11th client to existing infrastructure adds incremental cost (IPs, domains, mailboxes) but not the full base infrastructure. Agencies typically reach 80%+ infrastructure utilization at 15-20 clients on a base infrastructure deployment.

The categories of customers we serve in this segment

Established mid-size agencies (5-20 clients)

The core customer profile. Established agencies with operational processes and stable client base. Need reliable infrastructure that does not become an operational problem. Scale tier infrastructure or higher fits this segment.

Solo operators with high-value clients

Individual cold email consultants serving 1-3 enterprise clients at $5K-$20K each. Need infrastructure that can support enterprise-grade campaign requirements. Pro tier infrastructure typically sufficient.

Agencies migrating from mainstream ESPs

Agencies who tried SendGrid, Mailgun, or similar and got terminated. Need rapid replacement infrastructure with similar capabilities. Migration support during onboarding is operationally meaningful.

Agencies running SMTP fallback

Primarily mailbox-based agencies who need SMTP fallback for specific clients or campaigns. Smaller infrastructure deployment focused on specific use cases rather than full SMTP infrastructure.

Specialty agencies in restricted segments

Agencies serving content categories or audience types that mailbox terms prohibit. Crypto-related campaigns, adult industry outreach, certain political content, certain B2C segments where mailboxes get terminated routinely. SMTP-only infrastructure is operationally necessary for these segments.

Related operational reading

• SMTP relay service — pricing, tiers, signup
• Offshore SMTP hosting — what email-grade infrastructure requires
• IP warming service — managed warmup details
• Dedicated PowerMTA servers — for very high-volume agencies
• PowerMTA (wiki) — VMTA configuration glossary
• DKIM, DMARC, SPF — authentication protocols
• Feedback Loop (wiki) — FBL mechanics
• Agency case study — actual deployment example