When does dedicated SMTP actually make sense? A volume-based decision guide

We get asked “should I switch to a dedicated SMTP server” about five times a week. The honest answer, most of the time, is: it depends, and there is a decision framework that works. The decision shifted meaningfully during 2024-2025 because Gmail/Yahoo bulk sender enforcement (February 2024 progressive rollout) and Microsoft’s May 5, 2025 enforcement change (550 5.7.515 Access denied for non-compliant 5,000+/day senders) raised the floor for what shared SMTP providers can accept. What used to be “shared works fine until 250K/month” in 2022 is now closer to “shared works fine until 100-150K/month for many operators” in 2026.

This post is the decision framework we use across discovery calls. The authentication baseline that must be in place regardless of which path you choose is covered in our authentication 2026 guide; the broader self-hosted vs managed economics applicable above 1M monthly volume are in our self-hosted MTA vs managed ESP analysis.

The decision flow — visual architecture

Before the per-section detail, the diagram below shows the full decision flow from initial volume/use case assessment through to the four viable infrastructure paths, with the diagnostic gate that determines whether migration is appropriate at all.

The diagram makes explicit the most-misunderstood aspect of the dedicated SMTP decision: the diagnostic gate is not optional, it is the single highest-leverage step. Operators who skip diagnosis and migrate to dedicated to “fix deliverability” produce 30 days of apparent improvement (new IP unrecognised by spam filters), then 60 days of the same underlying problem at the new infrastructure, then conclude “dedicated did not help” while still not knowing the actual root cause. The 30-day diagnostic investment routinely saves 90+ days of misdirected migration effort.

Start with what shared relays actually cost you

Before you worry about dedicated, it is worth being clear about what you are trading away. Shared relays (SendGrid, Mailgun, Postmark, Amazon SES in its shared-IP form) give you a very good deal for low to mid volume: the provider maintains the reputation of a large IP pool, you benefit from that reputation on day one, and you pay about €1 per thousand emails at typical volume. Their deliverability is consistently excellent because they kick off anyone who drags the pool reputation down.

The flip side is that you benefit from, and are exposed to, the pool. If a coincident sender runs a bad campaign on your IP range, your inbox placement can drop by 10-20% overnight for reasons you did not cause and cannot control. And if your own patterns trigger the provider’s risk algorithm — sudden volume increase, any hint of unfamiliar content, a spike in hard bounces — the provider suspends you and asks questions later. For many operators, those two risks are the actual reason to move.

The volume threshold, roughly

The rule of thumb most practitioners agree on: below 50,000 sends per month, dedicated SMTP is almost always the wrong choice. You will spend more on infrastructure than on email, the reputation build takes months relative to the volume you have, and the shared pool gives you better deliverability than a cold dedicated IP would. Stick with a good shared relay and optimise your content.

Between 50,000 and 250,000 per month, dedicated SMTP starts to make economic and deliverability sense. At this volume, a single dedicated IP warms up in a reasonable window (30-60 days) and you gain two things: reputation isolation, so another sender’s mistakes do not hurt you, and throughput predictability, so your send is not throttled by a shared provider’s other customers.

Above 250,000 per month, dedicated is close to mandatory. At 500k+ you want multiple dedicated IPs to spread the load across warmer curves. Above 1 million you are into the territory where running PowerMTA or a full managed MTA stack is cheaper per email than any shared provider’s enterprise tier.

Cost crossover by volume — the visualisation

The chart below shows monthly cost of equivalent capacity across four infrastructure paths at three representative volumes. The shape matters because it shows that the crossover point where dedicated economics dominate is not a single threshold — it varies by which dedicated path you choose, and by 1M monthly volume the cost differential becomes dramatic.

Monthly cost across infrastructure paths and volume tiers, 2026 retail pricing

Categoría	Shared SaaS (SendGrid Pro tier)	Cloud dedicated IP add-on (+SaaS base)	Managed dedicated (Blue Spirit, EUR)	Self-deploy fully-loaded (incl. ops time)
50K/month	90	120	94	1050
250K/month	275	305	158	1050
1M/month	900	930	264	1100

Cost figures based on 2026 published pricing. SendGrid Pro tier: $89.95/month base for 100K, scaling with overage to $389/month at 1M; dedicated IP add-on $30/month per IP. Managed dedicated Blue Spirit: Starter €89/mo handles up to 250K, Growth €149/mo up to 500K, Scale €249/mo up to 750K — converted to USD at approximately 1.06 EUR/USD. Self-deploy assumes Hetzner CCX13 €15/mo + 10-15 hours/month operations time at €75/hour internal SRE rate (€750-1,125/month operations cost dominating infrastructure cost). Cost includes per-email overage at SendGrid; dedicated paths have no per-email overage. Pattern visible: managed dedicated economics break even with shared SaaS at approximately 100-150K/month and become dramatically cheaper above 250K — by 1M monthly volume, managed dedicated is 70-75% cheaper than equivalent shared SaaS while delivering reputation isolation and content policy freedom that shared cannot provide. Self-deploy fully-loaded cost is roughly flat across volume tiers because operations time dominates infrastructure scaling — only competitive with managed dedicated at very high volumes (5M+/month) where the per-email cost differential becomes meaningful.

Use case matters more than volume

Volume is a rough guide, but the actual question is what kind of mail you are sending. The thresholds above are for marketing and transactional email aimed at broadly consumer Gmail / Outlook addresses. Adjust as follows.

Cold email is a special case. Cold senders are persona non grata on most shared relays because the pattern (high volume of messages to previously-uncontacted recipients with low engagement rates) looks exactly like spam to the provider’s risk model. Dedicated SMTP is needed earlier and at lower volume, typically from 10-20k sends per month, combined with dedicated pre-warmed mailboxes on the sending side. See our guide on cold email infrastructure.

Transactional-only senders with high engagement can stay on shared longer than marketing senders because their signals look wonderful to any risk algorithm.

Adult industry senders often have to use dedicated because most shared relays explicitly refuse the content regardless of how clean the sending pattern is.

Regulated industries with audit requirements often need dedicated for traceability reasons, regardless of volume.

Use case fit matrix — which path matches which sender profile

The table below maps the most common sender profiles to the appropriate infrastructure path. The use case column dominates volume considerations for several profiles (cold email, restricted verticals, regulated industries) where the volume math alone produces the wrong answer.

Sender profile	Volume range	Recommended path	Why	Avoid
B2B SaaS transactional only	Any	Shared SMTP (SendGrid/Postmark/SES)	High engagement, predictable patterns favour pool reputation	Dedicated until 1M+/mo (overhead does not pay off)
E-commerce mixed transactional + marketing	50K-500K/mo	Managed dedicated with pool segmentation	Reputation isolation + transactional/marketing IP separation matters at scale	Single-IP dedicated (transactional reputation contaminated by marketing)
Newsletter publisher	50K-1M/mo	Managed dedicated	Reputation isolation for predictable content patterns	Self-deploy without engagement monitoring tools
B2B cold email outreach	10K-500K/mo	Managed dedicated with mailbox infrastructure	Shared SMTP refuses cold; volume threshold lower than marketing	Trying to send cold via SendGrid/Mailgun (suspension imminent)
Adult industry / cannabis	Any	Managed dedicated (offshore-friendly jurisdiction)	Shared SMTP AUP refuses content regardless of compliance posture	SaaS providers (account closure risk constant)
ESP / SaaS reseller	1M+/mo	Multi-server architecture (PowerMTA cluster)	Per-customer reputation isolation requires custom architecture	Single-server managed (architectural ceiling at this scale)
Regulated finance / healthcare	Any	Managed dedicated with audit logging	Audit trail and infrastructure traceability for compliance	Shared SMTP (audit posture impossible without infrastructure control)
Pre-PMF startup testing email	Under 20K/mo	Free tier shared (Brevo/Resend/Mailchimp)	Validate channel before infrastructure investment	Dedicated infrastructure (operational discipline before product-market fit is wasted)

What “dedicated SMTP” actually means, specifically

When you move to dedicated, you are buying three things bundled together that you often conflate. Separate them in your head so you can buy the right one.

The SMTP server process itself — PowerMTA, Postfix, OpenSMTPD, Haraka. This handles the actual protocol delivery. PowerMTA is the industry standard because its throttling, queue management and bounce classification are better than the alternatives at scale.

The sending IP addresses — dedicated IPv4 (and ideally IPv6) that you own the reputation of. The IPs need to be clean, properly configured with rDNS and SPF, and ideally sourced from a provider who did not just hand you a recycled dirty IP.

The authentication and DNS setup — SPF, DKIM with proper selectors, DMARC with a reporting endpoint you actually monitor, and in 2026 you should also have MTA-STS and BIMI sorted out for any Gmail-heavy list.

All three have to be right. A dedicated SMTP server with bad IPs is useless. Good IPs without proper auth are wasted. Authentication without a capable MTA cannot throttle or retry intelligently. The reason managed products exist is that bundling these three well is harder than it looks — and the failure modes of each affect the others.

The decision checklist

Run through this before you buy dedicated. One “yes” answer does not settle it; look at the whole picture.

• Are you sending more than 50k/month of marketing or more than 10k/month of cold outreach?
• Have you been suspended or throttled by a shared provider in the last 12 months for legitimate sending?
• Is your deliverability actually bad (inbox rate under 80%), and do you know that it is a reputation problem rather than content or list hygiene?
• Are you sending content your shared provider does not officially support (adult, cannabis-adjacent, political)?
• Do you have (or can you afford) an operator who understands DNS, IP reputation, and email authentication, or are you prepared to pay someone to manage it for you?

Three or more “yes” answers and dedicated is probably the right move. Two or fewer and you are likely better off fixing the fundamentals on your current relay first.

The honest breakeven math — where the numbers actually flip

Operators frequently ask me to put numbers on the volume thresholds I cited above. The honest math, calibrated against current pricing and what we charge for managed dedicated infrastructure, looks like this.

At 50,000 sends/month, SendGrid Pro tier costs around $90/month all-in. Equivalent dedicated setup — single IP, basic PowerMTA or Postfix MTA, decent VPS — runs about €15-30/month for the VPS plus €40-80/month managed (or 8-12 hours of operator time/month if you self-manage at $75/hour = $600-900 of unbilled cost). The dedicated path is cost-competitive on raw infrastructure but loses the comparison once you account for operator time. At this volume, the case for dedicated is not cost; it is reputation isolation and avoiding suspension risk.

At 250,000 sends/month, SendGrid scales to roughly $250-300/month at this tier, Mailgun similar, Amazon SES around $25 (SES is artificially cheap because it is just the sending infrastructure with no relationship management). Equivalent dedicated runs about €89-149/month managed for a single-IP Starter tier. Now the math flips: managed dedicated is cheaper than equivalent SaaS by 30-50% per month, with all the reputation isolation benefits as a free bonus. This is where most operators should be making the move.

At 1 million sends/month, SendGrid hits roughly $800-1,000/month enterprise pricing, Mailgun similar. Managed dedicated PowerMTA at this volume runs €169/month Growth tier — a 75-85% cost reduction with significantly better reputation isolation. Above this volume the math gets dramatic; we have clients sending 50M-100M/month who would pay $15,000-30,000/month at SendGrid enterprise but pay €279/month + per-server costs on managed dedicated. The savings fund the operational sophistication.

The breakeven is real. The trap operators fall into is comparing “shared SMTP price” to “dedicated server price” without including operator time on the dedicated side. If you self-manage and your engineering team does not have spare bandwidth, the apparent savings from self-hosted dedicated frequently disappear into operational overhead. Managed dedicated solves this by making the operator-time cost predictable as a line item.

Volume tier breakeven matrix — when each path becomes economically dominant

The table below summarises where each infrastructure path is economically dominant, at what volume it becomes appropriate, and the typical migration trigger that signals it is time to evaluate the next tier.

Volume tier	Dominant economic path	Typical monthly cost	Migration trigger to next tier
Under 20K/mo	Free-tier shared (Brevo/Resend free)	€0-€20	Crossing 20K sustained for 60+ days
20K-100K/mo	Shared SaaS Pro tier	€70-€150	Suspension events, reputation pool incidents, or growth to 100K+ sustained
100K-250K/mo	Managed dedicated Starter	€89-€129	Volume crosses 250K or pool reputation incidents accelerate
250K-1M/mo	Managed dedicated Growth	€149-€249	Volume crosses 1M or compliance requirements demand custom architecture
1M-5M/mo	Managed dedicated Scale or self-deploy	€249-€500	Volume crosses 5M or multi-tenant requirements emerge
Over 5M/mo	Multi-server architecture	€1,000-€50,000+	Volume ceiling per architectural pattern (typically 10-50M per managed cluster)

The pattern visible in this matrix: the 100K-250K transition from shared to managed dedicated is the highest-leverage migration decision for most operators. Below 100K, shared SMTP economics work fine. Above 250K, managed dedicated economics dominate dramatically. The 100-250K zone is where pain points (suspension history, reputation pool incidents, content policy conflicts) tip the decision rather than pure volume math.

Vendor landscape — what your dedicated SMTP options actually look like

Once you have decided dedicated SMTP is the right path, the next question is how to acquire it. Four distinct paths, each with different trade-offs:

Self-deploy on cloud VPS is the lowest-cost path on paper. Spin up a Hetzner CX31 or Contabo VPS (€15-30/month), install PowerMTA or Postfix yourself, configure DNS records, register FBLs, monitor metrics. The total infrastructure cost is genuinely low; the operational cost is high. Realistic time investment: 40-80 hours initial setup, 10-15 hours/month ongoing operations. This path makes sense if you have an in-house Linux engineer who genuinely wants to operate email infrastructure as part of their role. It does not make sense if email is a peripheral concern for your team.

Cloud-provider dedicated IPs on shared MTAs is the path SendGrid, Mailgun, and Postmark sell explicitly: pay an extra $79-200/month and get an IP allocated specifically to your sending. You still use the provider’s MTA infrastructure, but the IP is yours. The operational simplicity is excellent — the provider handles the MTA, the IP warmup, the FBL processing — and you get reputation isolation. The catch: you are still subject to the provider’s content policies, suspension procedures, and AUP enforcement. SendGrid will still kick you off for cold outreach regardless of whether you have a dedicated IP.

Managed dedicated infrastructure is what we and similar providers (SmtpHero, Cloud Server for Email, Postal-as-a-Service shops) sell: a complete dedicated stack — VPS or dedicated server, configured PowerMTA, IPs we maintain reputation on, FBL processing, monitoring, support — at a flat monthly fee. Pricing typically €89-€279/month at standard tiers. The operational burden moves to us; you operate the email-marketing application (MailWizz, Acelle, Mautic, your own) on top. This is the path most operators actually want when they say “I want dedicated SMTP” — they do not actually want to operate Linux servers, they want the deliverability outcome.

Enterprise-tier infrastructure with custom architecture is what the largest operators do: multi-server PowerMTA clusters across multiple datacenters, custom IP allocations from upstream providers, custom routing and load balancing, dedicated infrastructure team. Pricing scales with complexity, typically $5,000-50,000/month all-in for serious operators. This path makes sense above 50M emails/month or in regulated industries where infrastructure architecture is part of compliance posture.

Five mistakes operators make moving to dedicated SMTP

After scoping dedicated SMTP migrations for several hundred operators over the past few years, the same mistakes recur often enough to be predictable.

Mistake 1 — moving to dedicated to fix a problem that is actually content or list quality. If your shared SMTP deliverability is bad because your subject lines look like spam, your list has 30% inactive subscribers from years ago, or your content triggers spam filters, moving to dedicated will not fix it. The new IP will hit the same content filters, generate the same complaints from the same disengaged subscribers, and produce the same bad metrics. The first 30 days will look better because the new IP is unrecognized; the next 60 days will produce the same problem at the new infrastructure. Diagnose the actual cause before changing infrastructure. We refuse engagements where the operator has not diagnosed root cause because we know how the engagement ends.

Mistake 2 — buying a single dedicated IP for a workload that needs reputation segmentation. Operators sending mixed traffic — transactional plus marketing plus occasional bulk announcements — frequently buy one dedicated IP for everything. Six months later their bulk announcements are damaging the IP reputation that their transactional sends depend on, and they cannot recover the reputation without separating the traffic. The fix: from day one, separate transactional and marketing onto different IPs (or different IP pools at higher volume). Reputation segmentation is the single most important architectural decision in dedicated SMTP, and it is the one operators most frequently get wrong. See our IP separation guide for the architectural patterns.

Mistake 3 — skipping the warmup discipline because “it is dedicated, so it is fine”. Dedicated IPs are cold IPs on day one regardless of MTA quality. Operators who send their full volume on day one get blacklisted within 72 hours, lose 4-8 weeks recovering reputation, and frequently conclude “dedicated does not work for us” when actually warmup discipline failed. The fix: 14-30 day gradual ramp, segment by engagement, monitor bounce/complaint rates daily during warmup. The discipline is not optional; the math of warmup is independent of infrastructure choice. Per our IP warmup 2026 guide for the proper warmup curves.

Mistake 4 — running dedicated without monitoring the metrics that matter. Shared SMTP providers do this for you (you see your metrics in their dashboard, they alert you when something goes wrong). Dedicated infrastructure puts the monitoring burden on you: you need to track IP reputation (Talos, Sender Score, ReturnPath), inbox placement (seed list testing), bounce rate trending per ISP, complaint rate from FBL signals, blacklist status across major DNSBLs. Operators who skip this run blind and discover problems through customer complaints rather than through alerts. Managed dedicated handles this; self-deploy operators routinely skip it because “the system is working”.

Mistake 5 — treating dedicated as a one-time setup rather than ongoing operations. Email infrastructure is not “set and forget”. ISP rules change (Gmail-Yahoo bulk sender requirements October 2024, Microsoft RBL changes 2025, evolving DMARC enforcement), authentication standards evolve (BIMI, MTA-STS getting more weight), engagement patterns shift seasonally. Operators who treat the initial dedicated SMTP setup as a one-time project rather than ongoing operational discipline find their deliverability degrading 6-12 months later and cannot pinpoint why. The fix: budget ongoing operational time (or pay for managed) and monitor industry-standard signals continuously.

What changes in 2026 that affects this decision

Three specific 2026 dynamics shift the dedicated-vs-shared decision in ways operators evaluating today should be aware of.

First, the Gmail/Yahoo/Microsoft bulk sender requirements that went into full enforcement during 2024-2025 raised the floor for what shared SMTP providers can accept. Senders sending more than 5,000/day to Gmail addresses must have DMARC alignment, one-click unsubscribe, and complaint rates below 0.3%. Microsoft’s May 5, 2025 enforcement specifically returns 550 5.7.515 Access denied for non-compliant senders at this threshold per our Microsoft SNDS guide. Shared providers are increasingly aggressive about suspending senders who fail these checks because the provider’s pool reputation depends on it. The result: senders who would have stayed on shared comfortably in 2022 are increasingly being pushed toward dedicated because the shared provider cannot afford the risk. If you are reading deliverability articles from 2020-2022 and concluding “shared is fine for my volume”, reread the analysis with current ISP requirements in mind.

Second, dedicated IP economics continue to favor self-hosted at lower volumes because hosting prices have been roughly flat or declining (Hetzner, Contabo, OVH all stable) while shared SMTP pricing has been increasing 8-15% annually as providers chase margins. The breakeven volume where dedicated becomes cheaper than shared has been moving downward — what was a 250K threshold in 2020 is now closer to 100-150K for many operators. If you crunched the numbers two years ago and concluded “dedicated is too expensive”, the calculus may have shifted.

Third, the rise of self-hosted MTA alternatives (KumoMTA, Haraka 3.x, Postal v3) has made the dedicated path more accessible for operators with engineering capability. KumoMTA in particular ships an open-source license + Lua-based configuration + Prometheus-native metrics + Kubernetes-native deployment patterns that are operationally cleaner than PowerMTA’s accounting-log model for many operators. Self-deploy dedicated SMTP on KumoMTA + Hetzner VPS + decent monitoring is a realistic option for operators with strong DevOps capability who would have been forced toward managed dedicated 2-3 years ago. We cover the comparison in detail in PowerMTA vs KumoMTA.

EU compliance considerations — GDPR and dedicated infrastructure

Dedicated SMTP infrastructure in EU jurisdictions has compliance dimensions that shared SaaS abstract away. The 2024-2025 regulatory environment (DORA effective January 17 2025, ongoing GDPR enforcement, NIS2 transposition across member states) shifted the compliance posture for dedicated email infrastructure substantially.

GDPR Article 28 — processor agreements: dedicated SMTP infrastructure means you (as data controller) are using a specific processor (the managed dedicated provider) for personal data processing. Standard processor agreements (DPA) must be in place; the agreement should explicitly cover sub-processor relationships (datacenter provider, IP provider, monitoring service providers). Shared SaaS abstracts this — the SaaS provider handles sub-processor management opaquely; dedicated infrastructure makes it explicit, which is operationally more work but produces stronger compliance posture.

GDPR Article 32 — security of processing: dedicated infrastructure requires documenting “appropriate technical and organisational measures” for the specific deployment. Encryption at rest, encryption in transit (TLS 1.3 minimum for SMTP transport), access controls, audit logging, incident response procedures. Shared SaaS provides this through standard service descriptions; dedicated infrastructure requires explicit documentation of your specific deployment’s security posture. Cost-positive when compliance audits happen; cost-negative for operators without sophisticated security capacity.

DORA (effective January 17, 2025) — operational resilience for financial entities: financial sector operators using dedicated email infrastructure must include the infrastructure provider in their ICT third-party register, maintain exit strategies with documented migration paths, conduct annual third-party risk assessments. The DORA register specifically requires identifying the provider, contracted services, and the criticality classification of the service. Dedicated email infrastructure for transactional financial communications is typically classified as supporting (rather than critical) ICT service, but the documentation requirements apply regardless.

Schrems II and EU data residency: dedicated infrastructure enables explicit EU data residency commitment that shared SaaS often cannot provide. Datacenter location can be specified (Frankfurt, Amsterdam, Helsinki, Bucharest are common EU options); the data flow stays within EU jurisdiction; sub-processor agreements can be EU-based. For controllers with sovereignty requirements, dedicated infrastructure is structurally easier than shared SaaS that may use US-based MTAs even when “EU customer” tier is selected.

Country-specific considerations:

• Germany: BSI Grundschutz and ISO 27001 alignment for dedicated infrastructure is increasingly expected for B2B communications with German enterprises; managed dedicated providers should be able to articulate alignment posture.
• France: SecNumCloud qualification (ANSSI cloud security certification) is becoming relevant for dedicated infrastructure serving French government or regulated industry; not yet mandatory but increasingly preferred for sensitive data flows.
• Spain: ENS (Esquema Nacional de Seguridad) compliance for dedicated infrastructure serving Spanish public sector clients; private sector typically not required but increasingly preferred.
• Netherlands: BIO (Baseline Information Security) requirements for dedicated infrastructure serving Dutch government; broadly aligned with ISO 27001 baseline.

Compliance overhead operational reality: for EU-heavy dedicated SMTP operators, the compliance layer adds €500-€2,000/month in operational cost (legal review, DPA management, sub-processor audit, audit-log retention) that shared SaaS abstracts. This is real cost but typically materially less than the cost of regulatory enforcement action against non-compliant infrastructure.

When dedicated is genuinely the wrong answer

To balance the implicit pro-dedicated framing of this article: there are real scenarios where dedicated SMTP is the wrong choice and shared infrastructure is structurally better.

You are testing whether email marketing fits your business model at all. If you do not yet know whether email is core to your strategy, do not commission dedicated infrastructure. The operational discipline required for dedicated does not pay off until you know you are committed to email as a long-term channel. Use Brevo’s free tier, Mailchimp’s free tier, or Resend’s free tier for validation; commit to dedicated only after the channel is proven for your specific business model.

Your volume is genuinely tiny and likely to stay that way. Below 50K/month, dedicated economics rarely work even with managed providers because the operational floor is the dominant cost. If your business model produces 5K-20K emails/month forever (small newsletter, low-volume transactional needs), shared SMTP is the right answer regardless of how much articles like this push toward dedicated.

Your team has zero capacity for operational involvement. Even managed dedicated requires you to make occasional operational decisions — reviewing deliverability reports, deciding when to add IPs, responding to ISP-specific issues, coordinating with your application layer. If your team genuinely has zero bandwidth for any operational involvement, you are going to be unhappy with dedicated regardless of how managed it is. Stay on shared SMTP where the provider makes all the operational decisions.

You operate in a context where SaaS lock-in actually helps you. Some procurement contexts (enterprise vendor consolidation, security review burden, compliance frameworks that prefer enumerated SaaS providers) make dedicated infrastructure structurally harder than shared. If your business operates in that environment, the dedicated path adds friction without delivering value. Stay on whatever shared SaaS your procurement process accepts and optimize within those constraints.

The TL;DR I give operators in 90 seconds: dedicated SMTP is the right answer when you have volume above 50-100K/month, you understand the operational discipline required, and you either have engineering capability or will pay for managed infrastructure. Below that volume or without that capability, fix your shared SMTP setup first and revisit dedicated when the structural case is clearer.

Choose your infrastructure path — the decision tool

The dedicated SMTP decision involves five interacting factors: monthly volume, primary use case, current pain points, engineering capacity, and growth trajectory. Use the tool below to get a calibrated migration recommendation; the math reflects 200+ migration scoping engagements through 2024-2026 and accounts for the post-2025 enforcement environment.

The tool’s logic, in summary:

• Stay on shared SMTP — volume and pain points do not yet justify dedicated; focus on optimising current setup.
• Cloud-provider dedicated IP — reputation isolation without operational burden; middle path for 20-200K/mo with growth trajectory.
• Managed dedicated infrastructure — full reputation isolation with bundled operations; economic dominance at 100-1M/mo for most operators.
• Self-deploy on cloud VPS — strong DevOps capacity makes self-host economically optimal; KumoMTA + Hetzner is the realistic stack.
• Enterprise multi-server architecture — over 5M/mo justifies custom infrastructure with multi-server, multi-datacenter deployment.
• Wait and diagnose — undiagnosed pain is most likely content/list quality (60% + 25% of cases); migration without diagnosis fails predictably.

---

Need help scoping your infrastructure? Talk to sales — we will tell you honestly whether dedicated is the right call for you.