Market-share stories about email service providers usually run on vendor press releases and survey panels. Our dataset is blunter: 192 DNS snapshots of the Tranco top-1M domains, taken between January 2016 and the 2026-07-05 snapshot, with each domain's ESP classified from the include: mechanisms in its SPF record. 623,370 domains publish SPF today. When a company adds an ESP, it edits DNS; when it leaves, the record eventually gets cleaned up. Over a decade, those edits add up to the most honest churn chart in the industry.
Three careers below tell the whole story: a giant that was scheduled to die and did not, a specialist that quietly shrank, and a legacy brand that quietly grew.
Mandrill: the sunset that never set
In 2016, Mailchimp announced that Mandrill would stop being a standalone product — the classic "sunset" moment after which analysts write a platform off. In our first snapshots that same year, Mandrill was at its peak: 5.58% of SPF domains, one of the largest senders on the internet. A decade later it still stands at 3.37%. That is a decline, but a remarkably slow one — roughly forty percent of the peak share eroded over ten years, while the product survived as a Mailchimp add-on.
The mechanism is simple: transactional email is plumbing. It is wired into password resets, receipts and notification pipelines that nobody wants to touch. Migrations happen when something breaks or a contract forces the issue, not when a roadmap slide changes. Mandrill's regional footprint makes the point even harder — it remains the #1 ESP by SPF share in Germany, the Netherlands, Belgium and Switzerland at 25–27% of each country's SPF domains. Legacy infrastructure does not die; it amortizes.
Postmark and Constant Contact: two quiet trajectories
Postmark, the transactional specialist, peaked at 0.76% of SPF domains in 2017 and has drifted to 0.47%. Nothing dramatic happened — no outage, no scandal — the category simply consolidated around cheaper infrastructure senders. Meanwhile Constant Contact, a brand most SaaS commentary filed under "last decade," moved from 0.09% of SPF domains in 2016 to 0.57% today. A six-fold share increase for a company that gets almost no coverage is a useful reminder that the SMB market buys on distribution and bundling, not on what the industry considers current.
The winners, for context
Against those slow curves, the decade's structural shift is the rise of raw sending infrastructure:
- Amazon SES: 1.51% (2016) → 2.86% (2022) → 6.21% — now the #1 ESP in the dataset, having multiplied 2.2× since 2022 alone. We unpack that in the SES story.
- SendGrid: 3.67% (2022) → 4.75%; Mailgun: 2.61% → 4.10% — the developer-pipe tier keeps compounding.
- Newcomers: Brevo went from 0.1% in 2016 to a 1.42% all-time high; MailerSend from 0.13% at its 2017 debut to a 1.27% all-time high.
- The one true cliff: MailChannels, which peaked at 2.94% in March 2021 and fell to 1.07% — a 64% collapse and the decade's only fast exit, covered in the MailChannels cliff.
Even the era's most-discussed decline is gentle by software standards: Mailchimp peaked at 5.03% of SPF domains in November 2022 and has eased to 3.69% — meaningful churn, but spread over years, not quarters.
An ESP line in an SPF record represents integration work: templates, suppression lists, webhooks, analytics, and a team's muscle memory. The switching cost is real, so share erodes at the pace of rebuilds and replatforms — which for most companies means every five to ten years. That is why a DNS-level census sees a different market than quarterly earnings calls do.
What a decade of churn means for your ESP decision
For a marketer choosing or re-evaluating a provider, three operational takeaways fall out of the ten-year view:
- Discount the obituaries. A vendor being "out of fashion" predicts little about whether it will keep delivering your mail. Mandrill has outlived its own sunset by a decade.
- Watch the slope, not the headline. A slowly declining share usually means a stable platform in harvest mode. A fast drop — the MailChannels pattern — is the signal that demands a migration plan. The series for every major provider updates daily in the daily email infrastructure report.
- Assume you will change ESPs less often than you think. Since the integration will likely outlive several strategy cycles, weight deliverability fundamentals and data portability over feature checklists.
And because share numbers describe populations, not your list: whatever provider you run on, measure your own placement directly. A seed test across the major mailbox providers before and after any ESP change turns this market data into a decision about your mail, not the average sender's.