When you choose an email service provider, you are not just buying features — you are parking your sending reputation inside someone else's infrastructure and betting on their trajectory. Shared IP pools, postmaster relationships, abuse-desk staffing: all of it scales with the health of the vendor's business. The strange thing is that this health is publicly measurable, and almost nobody checks it. Every domain that uses an ESP declares it in an SPF include, and SPF records are public DNS. Count them daily across the Tranco top-1M and you get a live customer census for the entire industry. That is exactly what our dataset does: as of the 2026-07-05 snapshot, it tracks 623,370 SPF-publishing domains across 192 snapshots going back to January 2016.
Three curves, three diagnoses
The chart above shows why trend shape matters more than absolute rank. Three providers, three very different stories:
- Mailgun: steady compounding. From 2.61% of SPF domains in 2022 to 4.10% today. A curve like this means net customer inflow year after year — growing pools, growing investment, and a vendor whose incentives are aligned with keeping deliverability strong.
- Mailjet: quiet, consistent growth. From 1.33% in 2022 to 2.12%. Not spectacular, but monotonic — and regional strength (Mailjet leads France at 35.88% of the country's ESP-classified SPF domains) gives the curve a durable base.
- MailChannels: the cliff. A peak of 2.94% in March 2021, then a long slide to 1.07% — a 64% decline in share. We will not assert a cause from DNS data alone; what the data shows is thousands of domains removing the include, month after month. For any customer watching the curve, the message arrived years before any migration deadline did. The full case study is here.
Why a falling curve is your problem, not just theirs
A shrinking ESP is not a neutral fact about someone else's business. It propagates into your deliverability through concrete mechanisms:
- Shared-pool composition worsens. Disciplined senders migrate first; senders with nowhere to go stay. The reputation of the IPs you share concentrates around the remaining base.
- Unit economics shift. Fewer customers fund the same infrastructure, which historically pressures either prices or the postmaster/abuse functions that keep pools clean.
- Your exit becomes a migration under duress. Moving ESPs costs weeks of warmup and re-alignment. Doing it on your own schedule is a project; doing it after a vendor event is an emergency.
The reverse also holds: a rising curve like Amazon SES's (now 6.21% of SPF domains, the #1 share of any ESP) tells you the platform is absorbing the market — with its own trade-offs, which we cover separately.
A due-diligence checklist from open data
Before you sign or renew with an ESP, thirty minutes with public data answers questions no sales deck will:
- Pull the vendor's share trend. Our daily email infrastructure report charts SPF share for every major provider back to 2016, with a machine-readable api/latest.json. Sustained growth, plateau or decline — the shape is the signal.
- Check the slope since 2022. The post-pandemic window separates providers that kept compounding (Mailgun, Mailjet) from those that peaked (Mailchimp at 5.03% in November 2022, now 3.69%) and those that cratered (MailChannels).
- Look for regional fit. A provider that is #1 in your target market's country is a different bet from one that is globally mid-table but locally absent.
- Sanity-check with your own placement data. Share curves are a leading indicator of vendor health, not of your personal inbox rate. Run a seed test through the provider — ideally during a trial — and let real placement across Gmail, Microsoft and the rest confirm what the trend line suggests.
SPF share measures authorisation, not volume, and it undercounts providers whose customers flatten SPF into raw IPs. A dictionary update can also produce a step-change that looks like sudden growth. Use trend direction over years, not month-to-month wiggles, and read the methodology notes in the report before quoting a number.
Watch the curve after you sign, too
Vendor due diligence is not a one-time event. The dataset updates nightly, which means a deteriorating provider shows up in the data while you still have time to plan. Pair the external signal (your ESP's share curve) with the internal one (scheduled placement tests of your own streams), and you will never learn about a platform's decline from a bounce spike or a migration deadline again.