Which Mailboxes to Buy for Cold Email (Match Your Senders to Your List)

The Mistake: Buying Infrastructure Before Reading the List

The usual sequence is backwards. Someone decides to do cold email, buys a batch of inboxes on whichever platform they read about that week, warms them up, then loads a list and sends. Results come back uneven: strong reply rates from one slice of the list, near silence from another. The instinct is to blame the copy, so they rewrite the copy, and the pattern repeats.

The copy usually is not the problem. The problem is that a single lead list is never a single audience from a deliverability point of view. It is a mix of recipients sitting on Google Workspace, Microsoft 365, a long tail of smaller providers, and a slice hiding behind a security gateway. Those four groups evaluate your message on different signals, and the sending infrastructure that clears one of them is not the infrastructure that clears the others.

So the honest answer to "which mailboxes should I buy" is: check who you are emailing first. It takes about a minute, because the answer is sitting in public DNS.

How to Read Your List's Provider Mix

Every recipient domain publishes MX records that name the mail servers accepting its mail. Those records tell you exactly who is filtering on the other end. Google-hosted domains point at google.com, Microsoft at outlook.com, and gateways announce themselves with names like mimecast.com or pphosted.com.

You can check one domain by hand, but a real list needs bulk. Our free ESP segmentation tool reads every domain in a pasted list or uploaded CSV, splits it into those buckets, and then converts the mix into a mailbox shopping list. The rest of this article explains the reasoning behind what it recommends.

Segment 1: Google Recipients, Buy Google Inboxes

Google to Google is the strongest performing route we measure, and it is not close. When both ends of the conversation sit on Google infrastructure, the sending signals Google trusts most are the ones it can verify directly. A properly configured Google inbox with clean authentication reaching a Google-hosted recipient clears filters that the same message struggles with from any other origin.

For most B2B lists in the US and Western Europe this is also the largest single segment, which makes it the segment worth optimising first. If you only get one thing right in your infrastructure, get this pairing right.

Segment 2: Everything Else, Buy Azure

Zoho, Yahoo, Proton, iCloud, regional hosts, self-hosted mail servers: individually small, collectively often 20-30% of a B2B list. Filtering across this long tail is inconsistent, which makes it the wrong place to spend premium infrastructure budget and the right place for volume economics.

Azure tenants are the answer here because of how they price and isolate. A single tenant covers 100 inboxes, so even at a deliberately low per-inbox pace the tenant produces serious aggregate volume at a low cost per send. Just as importantly, it keeps this mixed-quality traffic on separate infrastructure from the inboxes carrying your Google and Microsoft segments, so a bad outcome in the long tail does not contaminate the segments that matter most.

Segment 3: Microsoft Recipients, Buy Aged Domains (on Google or Azure)

This is the counterintuitive one, and it is where most advice you will read is wrong. The common recommendation is to mirror the recipient: Microsoft recipients, Microsoft senders. In practice that under-performs, because it optimises the wrong variable.

Microsoft's inbound filtering leans heavily on sending domain reputation and domain history. A brand new domain sending cold mail into Microsoft gets treated with suspicion regardless of which platform it sends from. The variable that moves outcomes for this segment is domain age, not the sending platform.

Practically, this means keeping a pool of aged domains and routing your Microsoft-hosted prospects through them. It also means Microsoft-heavy lists need planning further ahead than Google-heavy ones, because you cannot manufacture domain age on demand. If your ICP skews Microsoft, that is a reason to acquire and warm domains earlier than you think you need them.

Segment 4: Gateway-Protected Domains, Do Not Target Them

Roughly 10-20% of B2B domains sit behind a secure email gateway: Mimecast, Proofpoint, Barracuda and the rest. These are not mailbox providers, they are dedicated filtering layers bolted in front of one, and they are considerably more aggressive than Google or Microsoft on their own.

Our recommendation for this segment is simple: do not target it. The organisations running gateways tend to be larger, more compliance-driven, and slower moving, and the reputation damage from repeatedly hitting aggressive filters is not worth what the segment returns. Segment them out and spend the volume elsewhere.

If your ICP genuinely lives behind gateways and skipping them is not an option, the rules change rather than relax:

• •Aged accounts only. Never send fresh infrastructure at a gateway.
• •Low volume. Well below the pace you run elsewhere, treated as a separate campaign rather than folded into your main send.
• •Cycle often. Rotate the accounts serving this segment more frequently than the rest of your fleet, and expect a shorter useful life from them.

How Many of Each: The Math

Once you know the mix, sizing is arithmetic. Three inputs: how many contacts in each segment, how many emails each contact receives across the sequence, and how many days you want the campaign to run.

The fourth number is the safe daily pace per inbox. These are the paces we send at:

• •Google: 15 sends per inbox per day.
• •Microsoft: 8 sends per inbox per day.
• •Azure: 2 sends per inbox per day, offset by 100 inboxes per tenant.

The formula is then straightforward:

A worked example

Take a 10,000-contact list running a 3-step sequence over 30 days, with a fairly typical mix: 55% Google, 25% Microsoft, 12% other providers, 8% gateway-protected.

• •5,500 Google contacts × 3 emails = 16,500 sends ÷ (15 × 30) = 37 Google inboxes, about 13 domains at 3 inboxes each.
• •2,500 Microsoft contacts × 3 emails = 7,500 sends ÷ (15 × 30) = 17 inboxes on aged domains, roughly 6 domains.
• •1,200 other contacts × 3 emails = 3,600 sends ÷ (2 × 30) = 60 Azure inboxes, which is well within a single 100-inbox tenant.
• •800 gateway contacts: excluded. Send these somewhere else or not at all.

That is a materially different order than "buy 50 inboxes and start sending", and it is the difference between infrastructure sized to the job and infrastructure sized to a guess. The ESP segmentation tool runs this calculation against your actual list rather than an assumed mix, and lets you adjust the campaign length and sequence depth to see how the requirement moves.

Prewarmed or Fresh?

Sizing tells you how many inboxes. The other decision is whether they should be prewarmed or fresh, and the answer is usually both, in a specific order.

Prewarmed inboxes carry established sending history, so they send at full pace the day you order rather than after a 14-21 day warmup. They are the right choice for the capacity you need now: launching a campaign, replacing burned infrastructure, or covering a deadline. They also ship with a free prewarmed domain, so nothing touches your brand domain.

Fresh inboxes on your own domains cost meaningfully less per inbox and are the right choice for the capacity you will need in a month. The standard play is to launch on prewarmed inboxes, start warming fresh ones behind them, and let the fresh fleet take over the baseline volume as it matures while prewarmed capacity handles spikes and recovery.

For Microsoft-heavy lists this sequencing matters more, because the aged domains that segment needs cannot be created on demand. Start that pool earlier than the rest.

Get the Number for Your Own List

Everything above is a general rule; your list has a specific mix. Paste it into the ESP segmentation and mailbox calculator and it will read the MX records, split the list into these four segments, and tell you how many inboxes of each platform to buy for the campaign length you have in mind. It runs free, exports a CSV with your original columns intact plus the provider data appended, and requires no account to use.

Two other tools pair with it: the Domain & Inbox Planner for converting a target volume into a domain and inbox count with a ramp schedule, and the Inbox Rotation Planner for splitting a fleet into rotating batches so no single inbox carries the campaign alone.