Dedicated vs shared IP: the real decision

Every email you send leaves through an IP address — the numerical street address mailbox providers like Gmail and Outlook use to decide whether to trust you. Reputation is attached to that address. The question this guide answers is whose reputation: yours alone, or yours blended with everybody else's.

Shared IP pool:your ESP — the email service provider that actually sends your mail, e.g. Braze, SendGrid, Mailgun — sends from IPs that handle thousands of other senders. Your reputation is partly your own, partly inherited from the pool average. A bad day across the pool raises your delivery risk; a great week across the pool lifts yours. You're renting reputation, not building it.

Dedicated IP: an IP that sends only your mail. Your reputation is entirely yours — for better and worse, no other sender can touch it. Requires a warm-up — a managed ramp-up period where you slowly increase volume so mailbox providers learn to trust the new address — before you can send at full volume (the IP warm-up playbook has the ramp schedule), and requires sustained volume to keep ISP trust alive.

Those are the two clean options. Most ESPs offer dedicated and charge extra for it — sometimes a few hundred dollars a year, sometimes thousands per IP depending on the contract. Confirm pricing before you commit. A dedicated IP that costs thousands a year and underperforms the shared pool is a losing trade you'll end up explaining to finance.

Here's the bit ESP sales decks skip past. Mailbox providers — Gmail, Outlook, Yahoo — score reputation by watching patterns over time. Engagement rates, complaint rates, bounce rates, all measured against your sending cadence. They need enough data to draw a line. If the data is sparse, they default to suspicion.

Rule of thumb: you need to sustain 100,000+ emails per day, every day, before a dedicated IP outperforms a well-run shared pool. Below that floor, ISPs (internet service providers — the umbrella term for the mailbox companies whose filters you're trying to please) can't build a reliable reputation signal from your sending patterns, and you end up with an IP that's technically dedicated to you but treated as unfamiliar every time it shows up at their front door.

Spikiness makes it worse. Some teams hit the volume requirement on peak days and fall well below it on quieter ones. That's worse than consistent low volume — ISPs watch for variance, and inconsistent senders are treated as riskier than consistent small ones. If your program does 200K on Tuesdays and 20K on Fridays, you're not a dedicated-IP candidate until the Friday floor comes up.

The corollary, said plainly: dedicated is not automatically better for deliverability. A volume-starved dedicated IP often delivers worse than the shared pool the team just migrated off. That surprises people who read "dedicated" and assume "premium".

Three risks teams underestimate when they sign the upgrade order.

Cliff risk.A bad campaign — high complaint rate, an unusual spike in bounces (bounces are emails rejected by the receiving server, hard or soft) — damages your sole-owned reputation directly. Shared pools absorb some variance because the noise of thousands of senders smooths it out. Dedicated IPs don't. Recovery from a serious cliff is typically 3 to 6 weeks of scaled-back, highest-engagement-only sending while you diagnose and stabilise. In severe cases you'll need to request a fresh IP from your ESP and warm that up from scratch.

Warm-up time cost.A new dedicated IP can't send at full volume from day one. The warm-up is 2 to 4 weeks of managed volume increase. If you urgently need to migrate sending — say, your previous setup is on fire — you're not actually moving on day one. You're starting a month-long transition during which capacity is constrained.

Monitoring overhead.You need to actively watch reputation through Google Postmaster Tools and Microsoft SNDS — free dashboards from Google and Microsoft that report on the health of mail you send into their networks — plus your ESP's deliverability reports. Because you're the only sender on that IP, nobody else is going to notice a dip first. On a shared pool your ESP does most of this for you; that's part of what you're paying the shared-pool tier for.

Three contexts where dedicated IP is the right answer even at volumes below the usual threshold.

Regulated sending.Financial services, healthcare, any program with legal requirements around message integrity. Isolation matters here because a shared-pool incident could technically be attributed to you, and "we were in a pool with a bad actor" is not a defence anyone wants to make to a regulator.

High-stakes transactional sending.Services that need near-100% deliverability on time-critical messages — password resets, order confirmations, 2FA codes (the one-time numbers you type after a password to prove it's really you). The sending pattern is consistent and the engagement is excellent by construction (people open password resets at 90%+), which builds clean reputation faster than a marketing program could.

Enterprise recipient base. B2B programs sending into corporate Microsoft Exchange servers benefit from the predictable reputation of a dedicated IP because corporate filters tend to whitelist known senders. You want to be the IP they recognise on sight.

Most guides present this as a binary. It isn't. The architecture larger programs end up on — usually after a few painful incidents teach them — is split sending. Marketing mail on the shared pool, transactional mail on a dedicated IP. Two different streams, two different reputations, two different sets of risk.

The mechanics: marketing mail rides the shared pool, where it benefits from pool reputation and absorbs the occasional bad day. Transactional mail (consistent, high-engagement, building clean reputation fast) rides the dedicated IP. Requires two ESP configurations and almost always a subdomain split — mail.brand.com for marketing, notify.brand.com for transactional. The subdomains let you authenticate each stream independently, so a marketing complaint spike never bleeds into your password-reset deliverability.

Starting on the split saves the incidents that usually teach it. The Orbit Deliverability Management skill covers the full domain-and-IP split pattern — subdomain setup, SPF/DKIM/DMARC per subdomain (the three authentication records that prove you own the sending domain), monitoring across both streams — so the pieces are in the right order from day one.

The decision rule, end to end: under 100K/day, shared. 100K/day sustained, dedicated worth considering — provided you have someone who'll spend the weekly hours on it. Mixed streams of any size, split. Anything regulated, dedicated regardless of volume. That's it. Skip the rest of the sales deck.