Own SMTP Server or Email Sending Provider: Costs and Risks Compared
What you really buy when you run your own SMTP
Own SMTP server vs a sending provider - this is not a question about pricing. It is a question about who picks up the phone when a major filter stops accepting your messages. You can set up a VPS with Postfix in fifteen minutes. And that is the start, not the end of the work.
What do you get? Control. Over the queue, the IP address, the logs. You see every response code from the recipient’s server and can change any transport setting. But along with control you take on the whole operational burden: warmup, bounce handling, updates, on-call duty.
A provider sells something completely different. Ready-made IP reputation, established relationships with filters, and feedback loops (spam complaint feedback loops) that tell you about recipient complaints. You pay so you don’t have to start from zero.
Let’s separate two worlds right away, because people mix them up all the time. An SMTP relay handles transactional mail: order confirmations, password resets, invoices. A marketing platform handles lists, consents, unsubscribes, segmentation and reporting. A technical relay won’t replace consent management. And a marketing platform is rarely the best transport for system notifications from an application.
The cost calculation: what the invoice shows and what it doesn’t
The invoice for your own infrastructure looks modest. A VPS, a dedicated IP, backups, monitoring, certificates. The total fits in the marketing budget without a conversation with the board. And here is the catch: the problem lies outside the invoice.
The hidden cost is engineering time. Warming up a new IP, bounce analysis, delisting requests, package updates, responding to an incident on a Friday evening. Add your own code too: automatic unsubscribes, hard bounce handling, list cleaning. Without it you send to addresses that stopped existing long ago and hurt your own results.
A provider charges differently: a fee per plan or per thousand messages, plus throughput limits and overage charges. It’s worth comparing your own numbers with the sending plan pricing page, because the break-even point is set by volume and your team’s hourly rate. Not by comparing price lists.
- Warmup - weeks of gradually increasing volume before a full campaign goes out.
- Bounce handling - parsing codes, separating hard from soft bounces, automatically disabling addresses.
- Feedback loops - registering with major mailbox providers and processing spam complaints.
- Reputation monitoring - blocklists, volume, queue, delays.
- On-call duty - someone has to answer an alert outside working hours.
Deliverability risk on your side
A new IP has no history. To a major filter you are nobody, and an unknown sender gets restrictions: slower acceptance, temporary rejections, the spam folder. History is built by sending mail that recipients open and reply to. There is no other way.
Warmup takes weeks. You start with small batches to your most engaged recipients, then raise the volume step by step. One weekend? Forget it.
The minimum setup, without which there is no point in starting:
- SPF - a record covering the actual sending address.
- DKIM - a signature on your domain, with a key of adequate length.
- DMARC - a policy plus a reporting address that someone actually reads.
- Return-Path - aligned with the sending domain, required for DKIM and SPF alignment.
- PTR and rDNS - matching the name given in HELO.
You can recognize a block at a major provider by the text of the bounce code. Next comes an unblock request and a break in sending, and nobody will guarantee how long it lasts. The most common cause of trouble is not technology but list quality. The filter judges recipient behavior, not what the sender claims. Nobody gets around filters - you can only send in a way that gives them no reason to block you.
Risk on the provider’s side: shared IPs and lock-in
On a shared IP pool you share reputation with strangers. Someone else’s failed campaign drags down your results, even if your list is spotless. Providers police this very differently, and it’s one of the first questions to ask before signing a contract.
A dedicated IP at the provider solves the problem, but it only makes sense with steady, regular volume. An address used once a month goes back to being an unknown sender every time.
The second risk: account suspension. When complaints spike, it can be immediate. No notice period, no negotiation. Sending stops in the middle of a campaign.
Migration isn’t free either. Exporting the list is a formality: you click and it’s done. Moving reputation means starting over: a new IP, a new warmup, filters learning about you again. The only thing you take with you without loss is your domain.
Tip: send from your own domain or subdomain, never from an address on the platform’s domain. Sign DKIM with a key on your domain. Then switching providers changes the transport, not the sender’s identity.
Law and data: GDPR, the Electronic Communications Law and B2B consent
Consent to commercial communication applies to businesses too. A personal address in the format firstname.lastname@company.com is personal data, so GDPR applies directly. The belief that B2B exempts you from consent costs more than a license for any tool. A lot more.
Article 398 of the Polish Electronic Communications Law requires the recipient’s prior consent to receive commercial information. Prior - meaning before the first message, not after it.
The responsibility model differs between the two options. With your own server you combine roles: you are the data controller and you also process the data technically yourself. Documentation, logs, procedures, responding to data subject requests - all of it is on you. With a provider you add a data processing agreement and questions about server location and any transfers outside the EEA.
But the most important thing is still the consent register and proof of how consent was collected: when, where, and with what clause wording. The sending technology comes second. Also set retention periods for logs and recipient data before a regulator or the recipient asks about them.
How we choose in practice
Running our own fleet of sending servers also means alerts in the middle of the night. We do it because sending is our product. For a company where email is a channel and not the product, the same math comes out completely differently.
In practice, a mixed model works best: transactional mail through your own relay, marketing through the MailCraft email marketing platform. Separating the streams protects your reputation - a complaint about a newsletter won’t block a password reset.
Questions to ask before you decide
- What is your monthly volume, and is it steady or does it spike during campaigns?
- Does anyone on the team administer servers and have time for it during working hours?
- How much does a day without sending cost you?
- Who will handle bounces and unsubscribes if a ready-made tool doesn’t?
- Do you have a consent register that will hold up in an audit?
To be honest: some of the features described above as industry standard are on our roadmap, not in the dashboard yet. We’d rather say so plainly than promise a feature you won’t see after logging in. You can check the current status in the platform feature overview.
Summary: a decision about responsibility, not price
Your own SMTP gives you control and moves the risk onto you. A provider gives you reputation and takes away some control. But neither option will beat a dirty list - technology won’t fix missing consent or addresses bought in a database.
The order is always the same: domain authentication and list hygiene first, infrastructure choice second. Doing it the other way round ends with a delisting request.
Tip: before you calculate the cost of a server, calculate the cost of a week without sending. That number usually settles the question faster than comparing subscription plans.
