01 — Outbound Infrastructure

Domain Architecture for Scalable Outbound

Most outbound teams think about domains as inventory.

How many domains should we buy?

How many mailboxes should sit on each domain?

When should we rotate them?

Those are operational questions.

The architectural question is different:

How should domains be structured so outbound can scale without turning sender identity, authentication, reputation, and observability into one tangled system?

That distinction matters because a domain is not simply the part after the @ symbol.

It can participate in multiple layers of email infrastructure:

  • the visible From domain

  • the DKIM signing domain

  • the SPF-authenticated domain

  • the organizational domain

  • the subdomain structure

  • the relationship between sending streams and IPs

Google and Yahoo both treat authentication and sender reputation as important parts of their delivery systems. Google’s Postmaster Tools, for example, exposes domain and IP reputation based on sending behavior, while Yahoo explicitly recommends segregating different email types by IP or DKIM domain.

So scalable outbound domain architecture starts with a better principle:

Design domains around sending identity and email streams—not around the number of emails you want to send.


What Is Domain Architecture?

Domain architecture is the structure that determines:

  1. Which domains represent your organization

  2. Which domains or subdomains send different types of email

  3. Which domain signs email with DKIM

  4. Which domain is authenticated through SPF

  5. How those identities align with the visible From domain

  6. Where reputation and delivery signals can be observed

  7. How different sending programs are separated

A simplified model looks like this:

                   ORGANIZATION

                         │

              ┌──────────┴──────────┐

              │                     │

        Core Business          Email Programs

           Domain                   │

              │          ┌──────────┼──────────┐

              │          │          │          │

           User Mail  Transactional Marketing  Outbound

                                      │          │

                                      ↓          ↓

                                  Sending Identity

                                      │

                              SPF / DKIM / DMARC

                                      │

                                      ↓

                                  IP / Provider

                                      │

                                      ↓

                                  Reputation


The domain is therefore one part of a larger system.

It should not be designed independently from authentication, sending infrastructure, and email streams.


1. A Domain Is More Than the From Address

Consider an email that appears as:

From: alice@example.com


There are other identities involved in delivery.

For example:

From:

alice@example.com


DKIM:

d=example.com


Return-Path / SPF:

bounce.example.com


These identities do different jobs.

DKIM authenticates a signing domain through the d= value.

SPF evaluates an authenticated sending identity.

DMARC then evaluates whether the authenticated identity aligns with the domain in the visible From address.

The DMARC specification defines identifier alignment between the RFC5322.From domain and a domain validated by SPF or DKIM. In relaxed alignment, the relevant identifiers can share the same organizational domain; strict alignment requires an exact domain match.

This means domain architecture cannot be reduced to:

Which domain is in the mailbox address?

The real question is:

How do all of the authentication identities relate to the domain the recipient sees?


2. Authentication Should Shape the Architecture

Google currently requires all senders to Gmail to use SPF or DKIM authentication.

For senders exceeding 5,000 messages per day to Gmail accounts, Google requires SPF, DKIM, and DMARC, and requires the From domain to align with either the SPF or DKIM domain for direct mail.

Yahoo has similar requirements for bulk senders: SPF and DKIM, a valid DMARC policy, and alignment between the From domain and either SPF or DKIM.

This creates a foundational architectural principle:

Every sending domain needs an intentional authentication relationship.

Adding domains without designing their SPF, DKIM, and DMARC relationships creates infrastructure without control.


3. Start With Email Streams, Not Domains

Before deciding how many domains to use, map the email streams.

For example:

Company

├── Employee / user email

├── Transactional email

├── Marketing email

├── Product notifications

└── Outbound sales


These streams do not necessarily have the same audience, content, sending behavior, or operational requirements.

Yahoo explicitly recommends segregating email types by IP or DKIM domain and advises against sending bulk/marketing email from the same IPs used for user mail, transactional mail, alerts, and similar traffic. Yahoo states that each IP and DKIM domain has a reputation that can affect delivery.

This is one of the strongest pieces of evidence for designing domain architecture around email function.

The principle is not:

“Every outbound program needs its own domain.”

It is:

Different email streams may need deliberate separation when their sending behavior and reputation requirements differ.


4. Core Domain vs Sending Domain

A company may have:

example.com


as its primary corporate domain.

That domain may support:

  • employees

  • website

  • customer communication

  • transactional messages

  • marketing

  • sales

The architectural question is whether all of those functions should share the same sending identity.

There is no universal answer.

Instead, evaluate:

Identity

Does the recipient need to clearly associate the message with the core company?

Stream

Is this employee communication, transactional email, marketing, or outbound?

Reputation

Would one stream's sending behavior create unwanted exposure for another?

Observability

Can you identify which stream is responsible for a delivery problem?

Operations

Can the team maintain authentication, DNS, monitoring, and compliance for every additional identity?

These questions are more useful than simply counting domains.


5. Subdomains Create Logical Separation

A common architecture is:

example.com

├── mail.example.com

├── tx.example.com

├── marketing.example.com

└── sales.example.com


Subdomains can provide logical separation between email programs.

They can also be useful for monitoring.

Google Postmaster Tools allows senders to add subdomains for independent monitoring. Google also states that its Domain Reputation dashboard displays messages sent from the exact domain used for DKIM and SPF authentication.

But there is an important distinction:

Logical separation is not the same thing as complete reputation isolation.

You should not assume that putting a sending program on a subdomain creates a completely independent reputation silo.

DMARC itself recognizes relationships between subdomains and their organizational domain, particularly under relaxed alignment.

So subdomains should be viewed as an architectural tool—not as a guaranteed reputation firewall.


6. Separate Organizational Domains Are a Stronger Boundary

Another model is:

example.com

examplemail.com

example-business.com


These are separate organizational domains rather than subdomains.

That can create a clearer identity boundary.

But it also creates additional operational responsibility.

Every additional domain can require:

  • DNS management

  • SPF configuration

  • DKIM configuration

  • DMARC configuration

  • verification

  • monitoring

  • reputation tracking

  • ownership

  • renewal and security controls

Google explicitly recommends setting up authentication for each sending domain.

Therefore:

A new domain should exist because it solves an architectural problem—not simply because the system has more capacity.


7. Domain Architecture and Reputation

Sender reputation is not a single number attached to an entire company.

Google describes reputation at both domain and IP levels and says reputation is determined by sending behavior. Its Postmaster Tools Domain Reputation dashboard is tied to the exact domain used for DKIM and SPF authentication.

Yahoo similarly says its reputation model considers factors including:

  • IP reputation

  • URL reputation

  • domain reputation

  • sender reputation

  • ASN reputation

  • DKIM

  • DMARC

Yahoo also explicitly states that IP and DKIM domain reputation can affect delivery.

This means domain architecture is partly reputation architecture.

A useful mental model is:

Sending behavior

       ↓

Authentication identity

       ↓

Domain / IP signals

       ↓

Reputation

       ↓

Delivery outcomes


The exact reputation systems used by mailbox providers are proprietary and can involve many signals.

So the architecture should focus on control and observability, not on assuming a simple one-domain/one-reputation formula.


8. Domain Architecture Is Not IP Architecture

These two layers are related but different.

Domain layer

From domain

DKIM domain

SPF domain

Subdomains

Organizational domain


IP layer

Shared IP

Dedicated IP

IP pool

Sending infrastructure


Amazon SES provides a useful illustration of this distinction.

AWS explains that dedicated IPs can isolate sender reputation from other SES customers and can also be separated into dedicated IP pools for different components of an email program. AWS specifically gives marketing and transactional email as examples of streams that can use separate pools.

So:

More domains do not automatically mean more reputation isolation.

Reputation architecture is an interaction between domain identity, IP infrastructure, authentication, sending behavior, and recipient-provider systems.


9. The Case for Separating Email Streams

Consider a company sending:

Transactional

100,000 messages/month


Marketing

500,000 messages/month


Outbound

20,000 messages/month


These streams have very different operating characteristics.

Transactional messages may be triggered by customer actions.

Marketing messages may be subscription-based.

Outbound may involve prospecting.

If all three streams share exactly the same sending infrastructure, a problem in one stream can become harder to identify and manage.

Yahoo explicitly recommends separating different email types by IP or DKIM domain for this reason.

The architectural objective is therefore:

Separate streams when separation improves reputation control, operational control, or observability.

Not:

Separate everything by default.


10. More Domains Also Create More Complexity

Domain separation has a cost.

Suppose an organization operates one sending domain:

1 domain

→ 1 SPF policy

→ DKIM configuration

→ DMARC policy

→ monitoring


Now multiply that across ten domains:

10 domains

→ 10 DNS configurations

→ 10 authentication surfaces

→ 10 reputation histories

→ 10 monitoring surfaces

→ 10 ownership/security dependencies


The infrastructure may become more resilient in some dimensions.

But it also becomes more difficult to operate.

This is why:

Domain count is not a scalability metric.

A scalable architecture is one that can grow while remaining understandable, observable, and controllable.


11. Don't Confuse Domain Rotation With Domain Architecture

These concepts are often mixed together.

Domain architecture

Answers:

Which domains exist, what do they represent, and how are they connected to sending systems?

Domain rotation

Answers:

Which domain sends a particular message at a particular time?

They solve different problems.

Architecture is structural.

Rotation is operational.

A system can have:

3 carefully designed domains


without continuously rotating them.

It can also have:

20 domains


and still have poor architecture.

The existence of multiple domains does not itself establish a sound reputation strategy.

Google describes domain reputation in terms of sending behavior, while AWS emphasizes historical sending patterns and predictable volume for IP reputation.

That suggests a more useful principle:

Optimize for stable, observable sending identities rather than arbitrary identity rotation.


12. Don't Build Architecture Around “Domain Age”

Outbound communities often discuss:

  • aged domains

  • 30-day domain warming

  • 60-day domain warming

  • 90-day domain warming

These concepts should not automatically be treated as mailbox-provider requirements.

Google's published sender requirements focus on authentication, DNS, TLS, spam rates, DMARC alignment, and related requirements—not on a universal domain-age threshold.

AWS does document gradual warm-up for new dedicated IP addresses, because receiving providers evaluate IP reputation and sending history. AWS specifically recommends increasing volume gradually rather than immediately using a new dedicated IP at full capacity.

That is an important distinction:

IP warm-up is not proof of a universal “domain aging” requirement.

A scalable architecture should be built around documented provider requirements and measurable sending behavior—not unsupported age formulas.


13. Authentication Architecture Matters More Than Domain Count

Consider this architecture:

From:

sales@example.com


DKIM:

d=example.com


SPF:

bounce.example.com


DMARC:

p=none


versus:

From:

sales@example.com


DKIM:

d=otherdomain.com


SPF:

bounce.otherdomain.com


The second architecture introduces a different authentication relationship.

DMARC alignment becomes a design consideration.

Under relaxed alignment, the DKIM signing domain and From domain can align when they share the same organizational domain. Under strict alignment, the domains must match exactly. The same principle applies to SPF alignment.

This is why adding an external sending domain is not simply:

“Buy domain → create inbox → send.”

It is:

Define identity → authenticate identity → align identity → monitor identity.


14. A Better Way to Think About Domain Pools

Instead of thinking:

Domain 1

Domain 2

Domain 3

Domain 4


think:

                    OUTBOUND PROGRAM

                           │

             ┌─────────────┼─────────────┐

             ↓             ↓             ↓

          Market A      Market B      Product A

             │             │             │

             ↓             ↓             ↓

        Domain/Subdomain Domain/Subdomain Identity

             │             │             │

             └─────────────┼─────────────┘

                           ↓

                    Sending Systems

                           ↓

                     IP / Provider

                           ↓

                     Observability


Now each domain has a reason to exist.

It belongs to a business or operational boundary.

That makes the system easier to reason about.


15. Observability Should Be Designed Into the Domain Structure

A domain architecture should make it possible to answer:

  • Which identity sent this message?

  • Which DKIM domain signed it?

  • Which IP delivered it?

  • Which email stream generated the traffic?

  • What is the authentication status?

  • What reputation signals are available?

  • Which system owns the identity?

  • What should happen if the stream becomes unhealthy?

Google Postmaster Tools provides domain and IP reputation, authentication-related information, spam-rate information, and delivery-error reporting for qualifying Gmail traffic.

Yahoo's Sender Hub similarly provides domain-level reporting and its Complaint Feedback Loop is based on DKIM-signed mail and the DKIM domain.

This makes DKIM domain design particularly important.

The domain is not merely an authentication record.

It can become an observability boundary.


16. A Scalable Domain Architecture

A practical architecture might look like:

                   COMPANY

                       │

        ┌──────────────┴──────────────┐

        │                             │

   CORE DOMAIN                  OUTBOUND SYSTEM

   example.com                       │

        │                   ┌─────────┴─────────┐

        │                   │                   │

   User / Product      Market A             Market B

   Communication           │                   │

                           ↓                   ↓

                    sales.example.com    sales.example.co

                           │                   │

                           ↓                   ↓

                        DKIM/SPF            DKIM/SPF

                           │                   │

                           └─────────┬─────────┘

                                     ↓

                              Sending Provider

                                     │

                                     ↓

                                  IP Layer

                                     │

                                     ↓

                             Gmail / Yahoo / ISP

                                     │

                                     ↓

                               Monitoring


The exact implementation will vary.

The important part is the logic:

Business function → email stream → identity → authentication → sending infrastructure → observability


17. When Should You Add Another Domain?

A new domain should answer at least one clear architectural requirement.

Add a domain when you need:

Identity separation

A distinct business or sending identity is genuinely required.

Stream separation

Different email programs have materially different sending behavior or operational requirements.

Organizational separation

Different business units or systems need independent ownership and controls.

Observability

You need a meaningful identity boundary for monitoring and diagnosis.

Infrastructure separation

The sending system requires a different authentication or infrastructure relationship.

But do not add a domain simply because:

  • you want more sending volume

  • another mailbox is available

  • a tool recommends a domain-to-inbox ratio

  • you heard that “more domains = better deliverability”

Those are not sufficient architectural reasons.


18. When Should You Not Add Another Domain?

Keep the architecture simpler when:

  • the email stream is small

  • the same organizational identity makes sense

  • authentication can be cleanly implemented

  • monitoring is already sufficient

  • additional separation would only create operational overhead

A simpler architecture is often better than a larger one.

The objective is not maximum fragmentation.

It is appropriate isolation.


19. The Domain Architecture Decision Framework

A useful decision framework is:

Question

If “Yes”

Architectural implication

Is this a different email function?

Yes

Consider separate stream

Does it have materially different sending behavior?

Yes

Consider IP/DKIM separation

Does it need independent ownership?

Yes

Consider separate identity

Would separation improve monitoring?

Yes

Consider subdomain/domain boundary

Does authentication become simpler?

Yes

Separate where appropriate

Is the only reason higher volume?

Yes

Don't automatically add a domain

Is the only reason “domain rotation”?

Yes

Re-evaluate the architecture

Can the team operate another domain reliably?

No

Don't add it

This is the core architectural decision.


20. The Four Layers of Domain Architecture

A scalable outbound architecture can ultimately be understood through four layers.

Layer 1 — Identity

Who is sending?

  • From domain

  • Organizational domain

  • DKIM domain

  • SPF identity

Layer 2 — Stream

What type of email is being sent?

  • Transactional

  • Marketing

  • Outbound

  • Product

  • User communication

Yahoo's guidance is particularly clear that email types may need to be segregated by IP or DKIM domain.

Layer 3 — Reputation

Where does sending behavior accumulate?

  • Domain

  • DKIM identity

  • IP

  • IP pool

  • Provider-specific reputation signals

Google explicitly exposes domain and IP reputation through Postmaster Tools.

Layer 4 — Observability

Can you see what is happening?

  • Authentication

  • Spam complaints

  • Delivery errors

  • Domain reputation

  • IP reputation

  • Sending behavior

This layer is what turns a collection of domains into infrastructure.


21. The Real Scaling Problem

The wrong scaling model is:

More prospects

      ↓

More emails

      ↓

More mailboxes

      ↓

More domains


A better model is:

Business objective

      ↓

Email streams

      ↓

Required capacity

      ↓

Identity architecture

      ↓

Authentication architecture

      ↓

Domain / IP architecture

      ↓

Monitoring

      ↓

Controlled scaling


This changes the role of domains.

Domains are no longer treated as disposable capacity units.

They become part of a controlled infrastructure layer.


Conclusion

Scalable outbound does not require the maximum number of domains.

It requires the right domain architecture.

That means understanding the relationship between:

From identity → DKIM → SPF → DMARC → email stream → IP infrastructure → reputation → observability.

Google's current sender requirements make authentication and alignment foundational for large-scale Gmail sending. Google also exposes domain and IP reputation through Postmaster Tools. Yahoo similarly emphasizes authentication, reputation, and separation of email types by IP or DKIM domain.

The architectural principle is therefore simple:

A domain should exist because it represents a meaningful sending identity or operational boundary—not merely because the outbound system needs more capacity.

The goal is not to own more domains.

The goal is to build a sending system where every identity has a purpose, every authentication relationship is intentional, every email stream is controllable, and every important reputation signal can be observed.

That is domain architecture for scalable outbound.


Engineering Reviews & Logs
0 ENTRIES