Our relay layer exposes an API compatible with Mailgun v3, SendGrid Web API v3, and Postmark — same endpoints, same tag structure, same parameter names. Switch your base URL and API key. Your existing application code, SDK calls, and webhook handlers continue working without modification. See full API compatibility details →
Complete IP warming guide with week-by-week volume schedules for Gmail, Outlook, and Yahoo. Prerequisites checklist, per-ISP monitoring, segmentation strategy, and recovery protocols.
IP warming is the structured process of gradually increasing sending volume from a new dedicated IP address to build sender reputation with major mailbox providers. Without it, ISP filtering systems treat your new IP as a potential spam source — because statistically, most new IPs used to send bulk email are spam sources. IP warming proves through demonstrated behavior that you're not.
This hub covers the complete IP warming process: the pre-warm-up prerequisites, week-by-week volume schedules, per-ISP monitoring, and how to handle setbacks when they occur.
IP warming that begins before infrastructure is properly configured wastes time and reputation capital. These must be verified before the first warm-up email:
dig -x [IP] and dig [hostname]. Without this, Microsoft and Yahoo will reject mail regardless of reputation.The schedule below is calibrated for a standard permission-based marketing program with a clean, engaged list. Adjust based on your program's specific characteristics — higher-quality lists with recent engagement can warm faster; programs with older or less-engaged lists need to be more conservative.
| Week | Daily Volume | Audience | Gmail Cap | Outlook Cap | Yahoo Cap |
|---|---|---|---|---|---|
| Week 1 | 200–500 | Last 30-day openers | 100–200 | 50–100 | 50–150 |
| Week 2 | 500–2,000 | Last 30-day openers | 300–800 | 150–500 | 200–600 |
| Week 3 | 2,000–8,000 | Last 60-day openers | 800–3,000 | 500–2,000 | 600–2,500 |
| Week 4 | 8,000–25,000 | Last 60-day openers | 3,000–10,000 | 2,000–8,000 | 2,500–8,000 |
| Week 5 | 25,000–75,000 | Last 90-day openers | 10,000–30,000 | 8,000–20,000 | 8,000–25,000 |
| Week 6 | 75,000–200,000 | Last 90-day openers | 30,000–80,000 | 20,000–60,000 | 25,000–70,000 |
| Week 7 | 200,000–500,000 | 180-day openers | 80,000–200,000 | 60,000–150,000 | 70,000–180,000 |
| Week 8 | 500K–1M | All engaged | 200K–500K | 150K–400K | 180K–400K |
| Weeks 9–12 | Ramp to target | Full list | Scale to target | Scale to target | Scale to target |
Important caveats: These numbers assume complaint rates below 0.05% and hard bounce rates below 1.5% throughout. If either metric exceeds these thresholds, pause and investigate before continuing the ramp. Per-ISP volumes are daily totals — if your Gmail distribution is 40% of your list, 60% of total volume goes to Gmail addresses.
Audience quality during warm-up determines how fast reputation builds. Tier your list by recency of engagement:
ISP filtering is not uniform — Gmail's algorithms are independent of Outlook's, which are independent of Yahoo's. An IP can build strong reputation at Gmail while facing throttling at Outlook. Monitor each ISP separately and address ISP-specific problems independently.
Check Gmail Postmaster Tools daily. The domain reputation indicator (BAD/LOW/MEDIUM/HIGH) is your primary signal. During warm-up, expect to see LOW for the first 2–3 weeks as Gmail collects data. Movement to MEDIUM typically happens by Week 4–5 of clean, engaged sending. Movement to HIGH typically follows Week 6–8.
If domain reputation doesn't improve after 3 weeks of clean sending, investigate: are authentication records correct? Is complaint rate truly under 0.08%? Are you sending to engaged segments only? Is there any spam trap activity visible in DMARC reports (unexpected failures from IPs you don't control could indicate spoofing of your domain).
Register at SNDS (Smart Network Data Services) — sendersupport.olc.protection.outlook.com. SNDS shows a color-coded status per IP: green (acceptable), yellow (Microsoft is watching), red (filtering or blocking applied). Check weekly, and immediately when delivery failures to Outlook addresses are reported in your bounce logs.
Microsoft's systems are more sensitive to volume spikes than Gmail. If you see yellow SNDS status, reduce your Outlook-bound volume by 50% for 3–5 days before attempting to scale again.
Process Yahoo FBL complaint data daily. Yahoo typically responds more quickly to positive reputation signals than Microsoft, making it a useful leading indicator of overall program health. If Yahoo inbox placement is improving but Microsoft is lagging, it's often a volume pacing issue — slow down Microsoft-bound sends while maintaining or slightly increasing Yahoo sends.
Three scenarios require immediate action:
Scenario 1: Complaint rate spike above 0.10%. Pause all sending from the affected IP immediately. Identify which list segment generated the complaints — pull FBL data and correlate with your send schedule. Permanently suppress any segment that generated complaints at above 0.15%. Do not resume until complaint source is fully suppressed and removed from future sends. Return to Tier 1 audiences only for 2 weeks after resuming.
Scenario 2: DNSBL listing during warm-up. A listing on any major DNSBL during warm-up is a significant setback — the IP has minimal reputation buffer, so any listing has more impact than it would at full production. Pause sending from that IP immediately. Investigate the root cause (spam trap hits? complaint spike? compromised infrastructure?). Request delisting only after root cause is addressed. If Spamhaus SBL or CBL is involved, expect 5–7 days for the investigation and removal process. During this period, route traffic to a backup IP if available.
Scenario 3: Gmail domain reputation drops to BAD. This is the most severe warm-up failure. Stop all sending immediately. Check DMARC reports for any unauthorized sending from your domain — a phishing campaign using your domain can damage domain reputation even if your own sending is clean. Verify all authentication records are correct. If authentication is clean and the problem is behavioral, it indicates complaint rates were significantly higher than monitoring showed. Conduct a full list audit before resuming. Recovery from BAD typically requires 8–12 weeks of low-volume, high-engagement sending.
Cold email warm-up operates by different standards than permission-based email. The volumes are smaller, the ramp is slower, and the per-domain limits are stricter:
IP warm-up is functionally complete when: Gmail Postmaster Tools shows HIGH domain reputation stable for 2+ consecutive weeks, Microsoft SNDS shows green status, Yahoo FBL complaint rates are consistently below 0.05%, and sending at full target volume produces no ISP throttle responses (no 4xx deferred messages above normal levels).
Warm-up completion is not a license to stop monitoring. The practices that build good reputation during warm-up — engagement segmentation, complaint rate monitoring, FBL processing, bounce suppression — are permanent operational requirements. Reputation earned during warm-up can be damaged quickly by a single poorly-managed campaign send. The investment in warm-up is only valuable if the operational discipline continues after it.
The generic "start with 200, double each day" warm-up schedule is a starting point, not a fixed rule. The right schedule depends on your target daily volume, the quality of your subscriber list, and the level of engagement you're able to generate during warm-up. Below are three calibrated schedules based on target program size.
Use this if you're new to dedicated IP sending, have experienced deliverability issues in the past, or are migrating from a shared IP pool to dedicated infrastructure for the first time.
| Week | Daily Volume | Cumulative Segment Depth | Focus |
|---|---|---|---|
| Week 1 | 100–300/day | Engaged last 30 days | Build initial reputation; monitor every send |
| Week 2 | 400–800/day | Engaged last 30 days | Confirm consistent acceptance at all major ISPs |
| Week 3 | 1,000–2,500/day | Engaged last 60 days | Expand slowly; add second ISP monitoring dashboard |
| Week 4 | 3,000–5,000/day | Engaged last 60 days | Approach target volume; prepare for full deployment |
| Week 5–6 | 5,000–7,000/day | Engaged last 90 days | Final calibration; confirm stable metrics before cutover |
Use this for established email programs migrating from shared IP pools or launching a new dedicated IP for an existing program with good historical engagement data.
| Week | Daily Volume | Cumulative Segment Depth | Focus |
|---|---|---|---|
| Week 1 | 500–1,000/day | Engaged last 30 days | Establish authentication signals; verify PTR/FCrDNS |
| Week 2 | 2,000–5,000/day | Engaged last 30 days | Scale up; monitor per-ISP acceptance rates daily |
| Week 3 | 7,000–15,000/day | Engaged last 60 days | Move toward operational volume; check Postmaster Tools spam rate |
| Week 4 | 20,000–30,000/day | Engaged last 60–90 days | Near-production volume; confirm SNDS green for Outlook |
| Week 5 | 33,000+/day | Full active list | Full deployment; maintain monitoring cadence |
Only appropriate for programs with an established, positive sending history migrating from previous dedicated infrastructure where engagement rates are demonstrably high (above 35% opens, below 0.05% complaints). Mailgun recommends starting at 100 emails on day one regardless of volume — but for established senders with verifiable engagement data, the aggressive schedule can compress the warm-up timeline to 3–4 weeks.
Key requirement: you must have 30+ days of engagement data from a previous sending program showing consistent positive signals. An aggressive warm-up applied to a new program with no engagement history will result in ISP blocking within days.
ISPs maintain reputation data with a sliding 30-day window. Most ISPs weight recent sending behavior more heavily than older behavior, and after 30+ days without sending activity, reputation data becomes stale or expires entirely. This creates the re-warming requirement: any dedicated IP that goes inactive for 30 days or more must be re-warmed before resuming production volume.
This rule has significant operational implications for seasonal senders — businesses that have high email volume periods (holiday season, annual conferences) separated by long quiet periods. An IP that sent 50,000 emails/day in December and goes inactive in January will need re-warming before the next high-volume campaign in spring.
The re-warming schedule doesn't need to start from zero if the previous sending history was positive, but it shouldn't assume full prior volume is immediately available. A reasonable approach: restart at 10–20% of peak volume and ramp back to full volume over 2–3 weeks while monitoring acceptance rates.
For programs with irregular cadences, consider maintaining a "keep-warm" sending schedule: a small volume of sends (5–10% of peak daily volume) sent consistently throughout quiet periods to prevent reputation decay. This is more efficient than repeated re-warming cycles.
Programs requiring multiple sending IPs — either for stream isolation (transactional vs. marketing) or for volume reasons — face the question of whether to warm all IPs simultaneously or sequentially.
Sequential warming: Warm one IP completely before adding the next. Simpler monitoring, clear isolation of reputation per IP, easier to diagnose problems. Disadvantage: extends total warm-up timeline proportionally (3 IPs = 3× the time).
Parallel warming: Warm all IPs simultaneously, distributing volume across the pool from day one. Faster timeline, but requires careful monitoring of each IP individually. If one IP in the pool encounters problems (blocklisting, complaint spike), you need to identify which IP it is and pause only that IP while the others continue.
For pools up to 5 IPs, parallel warming is typically manageable. For larger pools (10+ IPs), sequential warming in batches of 3–4 IPs avoids the monitoring complexity of tracking 10 independent reputation histories simultaneously.
PowerMTA and similar commercial MTAs support per-VMTA accounting that makes parallel pool monitoring practical — each IP generates its own delivery statistics, making it easy to isolate which IP in a pool is generating 4xx deferrals or 5xx rejections. With Postfix, parallel pool monitoring requires more custom log analysis tooling.
IP warming focuses on the IP address's reputation, but domain reputation is an equally important factor in inbox placement. A new sending domain starts with neutral-to-no reputation regardless of the IP's warm-up status, and ISPs evaluate both.
Domain reputation builds through the same signals as IP reputation — positive engagement (opens, clicks), low complaint rates, low bounce rates — but the domain reputation persists longer and recovers more slowly than IP reputation. A domain that has built six months of positive reputation history at Gmail has significantly more protection against short-term complaint spikes than a 4-week-old domain, regardless of how well the IP warm-up went.
Best practice: use your sending domain for at least 30 days with small-volume, high-engagement sends before your IP warm-up begins. This gives the domain some reputation foundation before the IP warm-up puts larger volumes through it. Mailgun explicitly recommends that the sending domain be active for at least 30 days before beginning an IP warm-up for this reason.
Our infrastructure team designs and manages the complete warm-up process for every new IP we deploy — including custom ramp schedules, per-ISP monitoring, and real-time adjustment when signals indicate modification is needed. Zero guesswork on your end.
Learn About IP Warming Service →