Contents
Dedicated IP and shared IP represent the most fundamental choice in email infrastructure: who owns the reputation of the IP address your messages come from. Dedicated IP gives one sender exclusive use of a specific IP, with full responsibility for and full benefit from its reputation. Shared IP pools many senders onto the same IP, sharing the reputation across the pool collectively. The decision is not just about cost; it determines warmup requirements, operational discipline, recovery dynamics when problems occur, and the deliverability ceiling achievable. The 2026 volume threshold for the transition is approximately 100,000 monthly emails minimum (Adobe Marketo guidance), with 300,000 monthly producing more stable dedicated outcomes (Postmark recommendation).
This comparison covers the IP-level decision specifically (distinct from the SMTP service-level decision): the operational mechanics of shared IPs including pre-warmed reputation and pool-managed quality control, dedicated IP mechanics including the 4-8 week warmup process and 30-day reputation memory pattern, cost economics across major providers ($24.95-$250 monthly range for managed dedicated IPs), blacklist scenarios and recovery dynamics for each, volume thresholds in detail with provider-specific guidance, and the decision framework based on sending profile, volume consistency, operational capacity, and business context.
The IP-level decision
One IP. One reputation. One sender's responsibility. Or pool reputation shared across many.
The IP-level choice differs from the SMTP service-level choice. SMTP service comparison addresses which provider operates the email infrastructure (Mailgun versus SendGrid versus Postmark, etc.). IP-level comparison addresses how reputation is allocated within whatever service is chosen: dedicated IPs assigned exclusively to one sender, or shared IPs serving multiple senders.
Most email service providers offer both options. SendGrid Pro includes dedicated IP option as add-on; Mailgun Scale plan can include dedicated IPs; AWS SES separates standard sending (shared pool) from dedicated IPs as discrete features; Postmark provides dedicated IPs on higher tiers. The choice is therefore typically made within a specific provider relationship rather than as a primary provider selection.
The decision drivers at the IP level:
- Volume: sufficient to maintain dedicated IP reputation
- Consistency: regular sending pattern rather than sporadic bursts
- Operational discipline: capacity to manage reputation actively
- Business context: compliance requirements, brand sensitivity, deliverability stakes
- Cost tolerance: willingness to pay dedicated IP premium
Sending profile matters more than total volume. A consistent 50,000 monthly sender often outperforms a sporadic 150,000 monthly sender on dedicated infrastructure because the consistency builds and maintains reputation while the sporadic pattern produces volatile reputation. The 100K monthly threshold is a useful baseline but consistency context is essential.
Shared IP mechanics
Shared IP infrastructure pools sending across multiple senders using the same IP addresses. The operational characteristics:
Pre-warmed reputation. Shared IPs have established reputation from existing pool activity. New senders joining the pool benefit from the cumulative engagement history without going through warmup. Sending starts immediately at normal volumes.
Pool-managed quality control. The email service provider operating the shared pool monitors per-sender metrics and removes problematic senders to protect pool reputation. Quality controls vary by provider: top-tier providers (Postmark, Mailgun, AWS SES) maintain strict pool quality; budget providers may tolerate more variance.
Pool segmentation. Many providers operate multiple shared pools segmented by sender quality. High-engagement senders are placed in higher-quality pools; lower-engagement senders are placed in different pools where they cannot damage premium pool reputation. The segmentation is typically transparent to senders but affects their assigned pool quality.
Volume flexibility. Shared IPs tolerate inconsistent sending well. Sending 1,000 emails one day and 100,000 the next does not damage shared IP reputation the way it would damage dedicated IP reputation. The aggregate pool volume smooths over individual sender variations.
Cost included or low. Shared IPs are typically included in standard email service plans without additional charge. The provider distributes infrastructure cost across many customers.
Reputation ceiling. Shared IPs have a deliverability ceiling determined by aggregate pool behaviour. Individual excellent sending cannot exceed the pool's baseline. Pool quality variance affects individual outcomes.
Contamination risk. A pool member with problematic practices can damage pool reputation, affecting all senders sharing the IP. Top providers minimise this through quality controls but cannot eliminate it entirely.
Burst limits. Shared pools impose per-sender burst limits preventing any individual sender from dominating pool resources or rate-limiting other senders. The limits mean large campaigns may take longer to complete on shared than equivalent volume would on dedicated.
Dedicated IP mechanics
Dedicated IP infrastructure assigns exclusive IPs to a single sender. The operational characteristics:
Cold IP requires warmup. New dedicated IPs have no reputation history. Mailbox providers initially treat them as unknown; the warmup process gradually establishes reputation through controlled volume increase.
Complete reputation isolation. Once warmed, dedicated IPs reflect only the operator's sending behaviour. Good engagement signals build the IP's reputation; bad signals damage only the operator's IPs.
Full control over sending characteristics. The operator decides exactly when and how much to send through their IPs. No artificial throttling from pool considerations; no constraints from neighbour behaviour. Large campaigns can complete quickly without external rate limits.
Predictable deliverability ceiling. Excellent sending practices produce excellent deliverability outcomes. The ceiling is the operator's own effort, not aggregate pool behaviour.
Higher cost. Managed dedicated IPs cost $24.95-$250 monthly per IP across major providers. Multiple IPs (typical for serious deployments) multiply the cost. Self-hosted infrastructure trades license/hosting cost for operational time.
Operational responsibility. The operator bears full responsibility for IP reputation. Authentication must be correct, sending practices disciplined, monitoring continuous, recovery timely. The discipline requirement is real.
Sensitivity to volume drops. Dedicated IPs require consistent sending pattern at sufficient volume. Volume drops or irregular patterns degrade reputation quickly. The 30-day reputation memory pattern means substantial inactivity essentially resets the IP.
Reputation atrophy. Without consistent sending, dedicated IP reputation atrophies. A 30-day gap in sending typically requires reinitiating warmup. This contrasts sharply with shared IP behaviour where pool reputation persists regardless of any individual sender's gaps.
The dedicated IP warmup process
Successful dedicated IP deployment requires disciplined warmup. The process establishes reputation with mailbox providers gradually to avoid triggering anti-spam responses from sudden volume.
Warmup mechanics in detail:
Starting volume. New dedicated IPs typically start at very low volumes: 20-50 emails per hour or 500-1,000 emails per day. The starting volume must be low enough that mailbox providers register the IP as new but trusted rather than new and suspicious.
Audience selection. Initial warmup volume should target the most engaged subscribers: those who opened or clicked within the last 15-30 days. The engagement-focused early warmup produces strong reputation signals that build trust quickly.
Volume ramping schedule. Double the volume every 2-3 days during the first week. Increase by 50% every 2-3 days during weeks 2-4. Increase by 25% during weeks 4-6 as the IP approaches target volume. The graduated ramping respects ISP rate limits and engagement signal accumulation.
Audience expansion. Initial 1-2 weeks: only engaged subscribers (last 30 days active). Weeks 3-4: expand to moderately engaged subscribers (last 90 days active). Weeks 5-6: expand to broader audience including less recent engagement. Adding unengaged subscribers too early produces poor metrics and damages warmup.
Per-ISP warmup. Each major ISP (Gmail, Outlook, Yahoo, AOL) effectively maintains separate reputation for the IP. Warmup happens per-ISP based on traffic to each. AWS SES Dedicated IPs (managed) automates per-ISP warmup; manual warmup requires balancing traffic across ISPs.
Monitoring during warmup. Track open rates, click rates, complaint rates, bounce rates daily during warmup. Watch for sudden drops indicating problems. Spam folder placement testing (GlockApps, Litmus, or similar) provides explicit visibility into where messages land at each warmup stage.
Total duration. Typical full warmup takes 4-8 weeks, with 4-6 weeks most common. The duration depends on starting volume, growth rate, ISP-specific responses, and audience engagement quality. Programmes with strong engagement signals can warm faster; programmes with weaker signals require more time.
Common warmup mistakes. Ramping too quickly (doubling daily instead of every 2-3 days); including unengaged subscribers early; ignoring per-ISP variation; failing to monitor metrics during warmup; reverting to broader audience too quickly when initial results look good. Each mistake produces deliverability problems that may take weeks to recover from.
Operators frequently look at warmup progress after 2-3 weeks, see that early metrics look acceptable, and decide to accelerate the schedule or expand audience faster than planned. The premature scaling produces reputation damage that may not be visible immediately: open rates stay reasonable for several days while spam folder placement gradually increases. By the time the operator notices the deliverability problem (typically through reduced click rates or customer complaints about not receiving emails), substantial damage has accumulated. Recovery from premature-scale damage can take 4-12 weeks. The 4-8 week warmup duration is a minimum, not a target; rushing it produces outcomes worse than skipping dedicated entirely.
The 30-day reputation memory pattern
Mailbox providers maintain reputation data for sending IPs over approximately 30 days. The pattern produces specific operational consequences for dedicated IP management.
The memory mechanics:
Reputation systems weight recent sending activity heavily. Sending consistently over the last 30 days produces strong current reputation. Sending sporadically or with gaps produces weak or unstable reputation. After approximately 30 days of inactivity, the IP's accumulated reputation has decayed substantially; further inactivity continues the decay until the IP is effectively treated as cold again.
The consequences for operations:
Consistent volume requirement. Dedicated IPs require regular sending to maintain reputation. Sporadic large campaigns separated by quiet periods produce worse reputation than smaller consistent volumes. The minimum recommended cadence is at least weekly sending; biweekly works for some programmes but is below ideal.
Seasonality challenges. Highly seasonal email programmes (holiday sales, tax season businesses, event-driven sending) often fit shared IP infrastructure better than dedicated because the non-peak periods drop below dedicated IP reputation maintenance thresholds. Adobe Marketo specifically recommends shared IPs for senders making fewer than 2 mailings per month.
Migration and recovery. If a sender pauses dedicated IP sending for 30+ days (planned hiatus, programme pause, infrastructure migration), the IP requires warmup again when resumed. Treating the dormant IP as still-warm produces deliverability problems on resumption.
Backup IP strategy. Programmes maintaining backup IPs for failover scenarios must keep them actively sending to retain their reputation. A backup IP not actively used loses its reputation within 30 days; activation in an emergency produces a cold IP rather than a warm backup.
Multi-IP coordination. Pool deployments with multiple IPs benefit from rotating traffic to ensure all IPs maintain consistent activity. IPs sitting unused while other pool IPs handle traffic atrophy.
The 30-day memory pattern affects shared IP infrastructure differently: shared IPs accumulate reputation from the pool's collective activity rather than any individual sender's pattern, so individual sender gaps do not affect the IP's reputation as long as other pool members maintain consistent volume. This is one of shared IP's structural advantages for senders with irregular volume.
Cost comparison across providers
Dedicated IP costs across major providers in 2026:
| Provider | Dedicated IP cost | Notes |
|---|---|---|
| AWS SES | $24.95/IP/month | No minimum commitment; managed dedicated pool option available with auto-scaling |
| AWS SES Dedicated IP (managed) | Volume-based pricing | Pool auto-scales per ISP; warmup automated |
| Brevo | $251/year ($21/month annualised) | Annual commitment required |
| Customer.io | $50/IP/month | Premium plan includes 3 dedicated IPs free |
| Mailgun | $59/IP/month | Available on Scale plan and above |
| Mailchimp | $30-50/IP/month | Varies by plan; available on higher tiers |
| Mailtrap | Included on Business plan ($85/month) | Auto-warmup included on Business and above |
| Postmark | $50/IP/month | Available on higher tiers; transactional pool stricter |
| SendGrid | $80-89/IP/month | Available on Pro plan and above as add-on |
| Twilio SendGrid Premier | Custom enterprise | Multiple dedicated IPs included in enterprise contracts |
| Klaviyo | $100-200/IP/month | Available on higher tiers; varies by overall plan |
| Infraforge (cold outreach) | $99/IP/month | Includes pre-warmed domains; quarterly billing |
| Instantly Light Speed | $358/month bundled | Includes private servers and dedicated IP blocks with SISR auto-rotation |
| Self-hosted (Hetzner, OVH) | $1-4/IP/month | Plus server infrastructure cost; operator manages everything |
The cost variance reflects different positioning and feature inclusion. Premium dedicated IP services (Instantly Light Speed bundled at $358 monthly) include extensive tooling beyond just the IP allocation: warmup automation, blacklist monitoring, rotation logic, multiple servers. Bare dedicated IP allocations from AWS SES at $24.95 monthly include only the IP itself; operator handles warmup, monitoring, and operations.
For typical commercial deployments, the cost range is $50-100 monthly per dedicated IP from major managed providers. Volume cost calculation:
- Small dedicated (1 IP): $25-100/month, suits 100K-250K monthly volume
- Medium dedicated (2-3 IPs): $100-300/month, suits 250K-1M monthly volume
- Large dedicated (4-8 IPs): $300-800/month, suits 1M-5M monthly volume
- Enterprise dedicated (10+ IPs): $1,000-5,000/month, suits 5M+ monthly volume
Blacklist scenarios and recovery
Blacklist incidents affect shared and dedicated IPs differently. Understanding the recovery mechanics helps with the decision.
Shared IP blacklist scenarios:
Cause attribution unclear. When a shared IP gets blacklisted, the specific cause is typically not visible to individual pool members. The provider's quality control team investigates internally; affected senders may not know exactly which behaviour triggered the listing.
Provider handles resolution. The provider contacts blocklist authorities, demonstrates that problematic senders have been removed, and works through the delisting process. Individual senders cannot directly engage with the blocklist for their pool IP.
Migration to alternative IPs. The provider typically migrates affected traffic to other pool IPs while the listing resolves. Operators may notice changed sending IPs but typically don't need to coordinate the change actively.
Resolution timeline. 24-72 hours for quick resolutions; 1-2 weeks for more complex situations. The timeline depends on the provider's response speed and the blocklist's delisting process.
Persistent risk. Pool members with consistently good practices are not entirely immune to repeated incidents; a pool with chronic quality control issues exposes all members to repeat events. Choosing a higher-quality provider reduces this risk.
Dedicated IP blacklist scenarios:
Cause attribution clear. The operator's own sending caused the listing (or in rare cases, IP allocation came with pre-existing issues). The cause is investigable: review recent sending patterns, complaint rates, list quality, content for spam triggers.
Operator handles resolution. Investigate the cause, correct the underlying problem, contact the blocklist for delisting. Most major blocklists provide self-service delisting if the operator can demonstrate the issue is fixed (Spamhaus offers this for many of their lists; Barracuda has a delisting process; smaller blocklists may require email contact with reputation services).
Reputation rebuild required. Beyond delisting, the IP's reputation with mailbox providers may have suffered. Recovery requires sustained good practices over 4-8+ weeks to rebuild reputation signals.
Resolution timeline. Quick delisting from a single blocklist with clear cause: 1-3 days. Moderate reputation problems: 2-6 weeks. Severe damage: 1-3 months or longer.
No fallback by default. Single dedicated IP deployments have no fallback when problems occur. Multi-IP pool deployments can route traffic to healthy IPs while affected IPs recover, but require deliberate architecture and operator management.
A B2C client we worked with in 2024 had been operating on dedicated IP at AWS SES for 18 months at approximately 400K monthly volume. During a Black Friday promotional period in 2024, they had a sudden complaint surge from a campaign with weaker engagement than typical. The complaint rate hit 0.4% (vs their typical 0.05%). Within 48 hours, the IP was listed on Spamhaus SBL. We paused all sending immediately, investigated the cause (a specific email creative with high promotional intensity), removed the problematic campaign from continued sending, requested SBL delisting (granted within 3 days based on the investigation evidence we provided), and slowly resumed sending with engagement-focused traffic only for the next 3 weeks. Total recovery to normal deliverability: 5 weeks. Lessons reinforced: dedicated IP reputation is fragile; sudden complaint spikes can produce listings within hours; recovery is operator's responsibility but is achievable with disciplined response; the trade-off versus shared IP is real but each side has different failure modes. A similar incident on shared IP would have been distributed across the pool (less individual damage) but harder to attribute and resolve. The operator's preference for control over the recovery process kept them on dedicated despite the incident.
Volume thresholds in detail
The volume thresholds for dedicated IP viability vary by provider and use case. Detailed thresholds across providers:
| Provider/Source | Minimum threshold | Recommended threshold | Notes |
|---|---|---|---|
| Customer.io | 50K emails/week (~200K/month) | Higher for stability | Hard requirement for dedicated IP eligibility |
| Adobe Marketo | 100K emails/month | Plus 2+ mailings/month minimum | Less is "low volume"; shared recommended |
| Postmark | Not explicit minimum | 300K emails/month | For stable dedicated IP performance |
| AWS SES | Not explicit minimum | Variable by use case | Managed dedicated IPs (managed pool) auto-removes IPs if volume too low |
| Suped guidance | 50K emails/week | Up to 250K/month max single IP | Below threshold: shared better |
| Mailflow Authority | 100K emails/month minimum | 500K+ for 2-3 IP pool | 1M+ recommends 4-8 IPs |
| Cold outreach (Smartlead data) | Variable | Most senders shared | 85-90% of cold emailers use shared infrastructure |
The threshold pattern across sources: 100K monthly is the widely cited entry point; 200-300K monthly provides stable dedicated operation; below 50-100K monthly, shared infrastructure typically produces better outcomes.
Beyond raw volume, consistency matters substantially. The thresholds assume regular sending pattern (at least weekly, ideally daily for high-volume programmes). Sporadic large campaigns with quiet periods between produce worse outcomes than smaller consistent volumes even when total monthly volume is similar.
Special cases:
Transactional email. Transactional infrastructure benefits from dedicated IPs at lower volumes than marketing because the engagement signals are stronger (transactional opens approach 60-80% versus marketing 20-30%). Postmark recommends dedicated transactional pools at lower thresholds than marketing equivalents.
Compliance-sensitive sending. Healthcare, financial services, and other regulated industries may require dedicated infrastructure regardless of volume for audit and isolation reasons. The compliance requirement overrides volume-based optimization.
Multi-tenant operators. ESPs, agencies, and platform operators may need dedicated IPs at lower per-tenant volumes for stream separation. Aggregate volume across tenants supports the dedicated infrastructure even when individual tenant volumes are below typical thresholds.
Decision framework
The decision framework for dedicated IP vs shared IP in 2026:
Use shared IP when: monthly volume under 100K (insufficient for dedicated IP reputation); sending pattern is irregular or seasonal with substantial gaps; cost minimisation is the primary priority; the team is new to email infrastructure and lacks operational discipline for dedicated IP management; the use case is testing, staging, or low-stakes communications.
Use dedicated IP when: monthly volume exceeds 100,000-300,000 reliably; sending pattern is consistent (regular weekly or daily cadence); the team has operational discipline for warmup, monitoring, and recovery; deliverability is materially important to business outcomes; compliance or brand requirements favour reputation isolation.
Use multiple dedicated IPs (pool) when: volume exceeds 500,000 monthly (single IP approaches practical limits); stream separation matters (marketing on one set, transactional on another); risk distribution across multiple IPs is operationally valuable; the team has capacity to manage multi-IP coordination including warmup staggering and per-stream policy management.
Choose AWS SES Dedicated IPs (managed) when: dedicated IP benefits are desired but operational simplicity matters; auto-scaling per ISP and adaptive warmup are operationally valuable; managed service is preferred over self-managed IP operations.
Choose self-hosted dedicated infrastructure when: volume justifies the operational investment (typically 5M+ monthly); full control over IP allocation and management matters; cost savings of self-hosted (potentially 50-80% versus managed) justify the operational time; existing in-house operational capacity supports the additional responsibility.
Stay on shared IP indefinitely when: volume remains below dedicated thresholds; the email programme is not central enough to business outcomes to justify dedicated investment; the operational simplicity of managed shared is preferred over the discipline of dedicated management.
The 2026 default progression for typical programmes: start on shared IP at any volume; transition to dedicated IP at approximately 100K-200K monthly volume when operational discipline supports it; expand to multi-IP pools at 500K-1M monthly; consider self-hosted at 5M+ monthly. The progression should be volume-driven and operational-maturity-gated rather than rushed for marketing reasons alone. Premature dedicated IP transitions produce worse outcomes than staying on quality shared infrastructure.