Case study · Email deliverability · 2026

Suspended
to 390K.

ResearchInn, an international open-access publisher, had its bulk email account suspended mid-send after one campaign bounced at 14%. Over nine weeks I found the cause, rebuilt the programme and scaled it to 390,561 emails across its five journals, with the hard-bounce rate cut by a factor of 48.

Visit researchinn.org↗ Email deliverability & lifecycle · August – October 2026

Client: ResearchInn · five open-access journals
Programme: 37 campaigns, August – October 2026
Platform: OJS 3.4 journals + bulk email provider
My role: Deliverability, lists, campaigns & the journal platform

researchinn.org

ResearchInn open-access journals homepage

Hover the browser to scroll the journals homepage

390,561

Emails delivered

Across 37 campaigns for five journals, from an account that had been suspended mid-send.

48×

Fewer hard bounces

From 6.74% before the work to 0.14% in the re-engagement phase.

99.46%

Peak delivery

On a single 93,000-email campaign.

0.005%

Spam complaints

Across the whole programme: 18 complaints in 390,561 emails.

16×

The original volume

Scaled up while deliverability improved, not at its expense.

~7^pts

From pacing alone

Identical content and audience. Spreading the same volume over time was worth about seven percentage points of deliverability.

01 · The problem

Dead mailboxes on live domains.

Three campaigns in August bounced at 6.74%, one of them at 14.01%, and the email provider suspended the account mid-send. An analysis of 27,691 contacts showed the cause wasn't spam complaints (there were none). It was dead mailboxes on domains that still worked: addresses taken from academic metadata, where the university domain still resolved but the person's mailbox had been closed. Domain-level checks can't see that.

The strongest predictor was how recent the paper was. Contacts from 2020 papers bounced at 9.04%; contacts from 2025 papers at 3.32%. Recency, not domain quality, decided whether an address would still accept mail.

02 · What I changed

Verify, suppress,
pace, schedule.

Mailbox-level verification replaced domain checks, cutting bounces from 6.74% to 1.64% on first contact.
Suppression lists built from the send data: every hard bounce, complaint and unsubscribe excluded on the server before the re-engagement phase.
Send pacing diagnosed from SMTP response codes. Gmail's 421 4.7.0 deferrals proved that throttling, not bad addresses, was costing 1,208 emails in a single campaign.
The provider's own scheduling replaced hand-built batches, spreading each campaign across 24 hours automatically.

August · before6.74%

September · first contact1.64%

October · re-engagement0.14%

Hard bounce rate by phase

03 · Results by phase

Every phase
cleaner than the last.

Phase	Campaigns	Sent	Delivered	Hard bounce	Opens	Complaints
August · before the work	3	12,828	90.72%	6.74%	18.21%	4
September · first contact	29	211,629	96.91%	1.64%	5.44%	12
October · re-engagement	5	166,104	99.46%	0.14%	1.86%	2
Total	37	390,561	—	—	—	18

October went only to contacts who hadn't opened before, so its open rate is low by design. Its delivery rate is measured across the four completed campaigns.

04 · What pacing was worth

Same emails,
better timing.

Send shape	Delivered	Soft bounce
4 campaigns in one hour	92.66%	5.30%
Same volume, 3-hour spacing	98.36%	0.55%
93k campaign, 24-hour spread	99.43%	0.40%

Identical content and audience quality. The only variable was how the volume was spread over time.

05 · Scale & segmentation

Five journals,
everyone reached.

Journal	Audience	First contact	Re-engagement	Total sent
Society, Policy and Human Development	131,738	131,737	93,147	224,884
Digital Health and Clinical Intelligence	24,992	24,992	22,320	47,312
Climate-Smart Agrifood Systems	23,238	23,238	21,338	44,576
Sustainable Energy and Applied Innovation	17,563	17,563	16,019	33,582
Applied AI	14,102	14,102	13,280	27,382
Five journals	211,633	211,632	166,104	377,736

Every contact in all five audiences was reached at least once. Re-engagement went only to the 195,833 who hadn't opened, deduplicated at recipient level across 30 campaigns.

06 · Beyond email

The platform
behind it.

I also run ResearchInn's journal platform. I installed and migrated Open Journal Systems (OJS) with a custom theme, set up the server and moved it to new domains and hosting, and fixed the failing issue exports, XML errors and Google Scholar indexing problems that were blocking discoverability. Mail runs through SMTP with SPF, DKIM and DMARC plus server-level anti-spam, and the outreach lists come from my own extraction pipelines, cleaned with local LLMs.