Home › Blog › Studies

5 studies

Deliverability Studies

Original research on the state of cold email, measured on live sending infrastructure rather than surveyed. Every figure comes from our own fleet or a full sweep of real domains, and each study says what it does not show as well as what it does.

Share the research

469,331

messages scored

PlacementUpdated monthly6 min readFieldwork: March to October 2026, ongoing

Every placement test we ran on prewarmed inboxes

Every new batch of prewarmed inboxes is seed-tested before it ships, and only stock clearing 95% is deployed. 8 tests, 469,331 messages scored, published whole — a live read on how hard inboxing got through 2026. Every results screen is on the page.

WI-RESEARCH-2026-002 · first published 2026-10-08

Data (CC BY 4.0): inbox-placement-log.json · inbox-placement-log.csv · inbox-placement-log-routes.csv. Method and limitations are on the study page.

Read the study

52

TLDs rated against abuse sources

Domains11 min readFieldwork: November 2026

A study of TLDs for cold email, ranked by how often they recur on abuse lists

52 top-level domains rated by how many of 8 public abuse sources name them, with registration and renewal prices at two registrars, set beside what our own sending measured by TLD.

WI-RESEARCH-2026-005 · first published 2026-11-04

Data (CC BY 4.0): tld-reputation-2026-11.json · tld-reputation-2026-11.csv · tld-reputation-2026-11.pdf. Method and limitations are on the study page.

Read the study

0

tenant blocks at 25 users per tenant

Microsoft10 min readFieldwork: 20 October to 8 November 2026

Microsoft’s Azure tenant crackdown and the 550 5.7.237 bounce: what three fleets showed

From 29 October 2026, senders across sequencers began seeing a new Microsoft bounce, 550 5.7.237, and the worst hit had whole tenants refused with 550 5.7.708. Segment 1, 1,500+ customer tenants at 25 users: no blocks, no licence wipes. Segment 2, 142 dense tenants at 50 or 100 users: outbound refused from 20 October, half of all mail bouncing at worst, back under 1% by November. Segment 3, our own fleet: a step, still around 1%. Tenants younger than 10 to 14 days look like the line.

WI-RESEARCH-2026-004 · first published 2026-11-06

Data (CC BY 4.0): microsoft-azure-tenant-crackdown.json · microsoft-azure-tenant-crackdown-days.csv · microsoft-azure-tenant-crackdown-configs.csv · microsoft-azure-tenant-crackdown-segment-2-days.csv · microsoft-azure-tenant-crackdown-segment-2-traces.csv · microsoft-azure-tenant-crackdown-segment-2-watch.csv. Method and limitations are on the study page.

Read the study

16×

.link against .com

Domains11 min readFieldwork: 30 days to 8 October 2026

A study of top-level domains by sends

3,415,449 sends split across five TLDs and scored against Google’s permanent domain-reputation verdict. .link drew it 16 times more often than .com, and .org twice as often — and the gap holds inside a single client, where the copy and the list are the same. The TLD you buy predicts this better than anything else we measured, and unlike a blocklist listing it costs replies.

WI-RESEARCH-2026-003 · first published 2026-10-10

Data (CC BY 4.0): google-sender-reputation.json · google-sender-reputation-estates.csv · google-sender-reputation-tlds.csv · google-sender-reputation-clients.csv. Method and limitations are on the study page.

Read the study

2,817

domains tested

Blocklists10 min readFieldwork: 30 days to 29 July 2026

Does a SURBL listing actually hurt cold email? We tested 2,817 sending domains

A DNS sweep of every live sending domain, not a sample. 43.3% were listed, and reply rate barely moved: 1.14% to 1.16%. Includes the three mechanisms we diagnosed SURBL grouping domains on — shared keyword, redirects to one master, shared nameserver zones — and the fix for each.

WI-RESEARCH-2026-001 · first published 2026-10-07

Data (CC BY 4.0): surbl-blocklist-study.json · surbl-blocklist-study.csv. Method and limitations are on the study page.

Read the study

How to cite this research

The figures and the data are free to reuse with attribution under CC BY 4.0. Each study keeps its ID for good. The same list is machine-readable at /research.json.

WarmInboxes (2026). Every placement test we ran on prewarmed inboxes. WI-RESEARCH-2026-002. https://warminboxes.com/research/inbox-placement-log
WarmInboxes (2026). A study of TLDs for cold email, ranked by how often they recur on abuse lists. WI-RESEARCH-2026-005. https://warminboxes.com/blog/tld-reputation-cold-email
WarmInboxes (2026). Microsoft’s Azure tenant crackdown and the 550 5.7.237 bounce: what three fleets showed. WI-RESEARCH-2026-004. https://warminboxes.com/research/microsoft-azure-tenant-crackdown
WarmInboxes (2026). A study of top-level domains by sends. WI-RESEARCH-2026-003. https://warminboxes.com/research/google-sender-reputation
WarmInboxes (2026). Does a SURBL listing actually hurt cold email? We tested 2,817 sending domains. WI-RESEARCH-2026-001. https://warminboxes.com/research/surbl-blocklist-study

New studies are added as the fieldwork completes. The placement log is updated every month.

Prewarmed & Fresh Inboxes

Google, Microsoft 365 and Azure inboxes with free prewarmed domains and DNS configured before you log in. Prewarmed inboxes send the day they land; every order is delivered within 48 to 72 hours.

Get Prewarmed & Fresh Inboxes