How to Test Email Deliverability: A Weekly Operator's Guide

I used to treat the green “delivered” number in my email platform as proof that a campaign had reached readers. That was a mistake. A message can pass the mail server handoff, avoid a bounce, and still land in Promotions, Updates, quarantine, or Spam. If you're running a newsletter on beehiiv, Substack, Ghost, LetterBucket, or another platform, how to test email deliverability starts with checking the mailbox surface your reader sees.

My weekly process now combines inbox placement, folder placement, authentication, reputation signals, and test-cohort engagement. It takes more work than reading one ESP dashboard, but it catches failures before they affect a full campaign. The workflow below is the one I use for newsletter sends, migrations, and higher-risk launches.

Table of Contents

• The Deliverability Problem That Surprised Me
• The Five Numbers I Track on Every Test Send
  • 1. Inbox placement rate
  • 2. Spam-folder placement
  • 3. Gmail Promotions placement
  • 4. Authentication pass rate
  • 5. Engagement-to-delivery ratio
• My Pre-Send Testing Checklist
  • What I inspect in the message itself
  • Content and link checks
• Why Seed Lists Are the Only Test That Actually Matters
  • What failed in my seed-list experiments
• Tools I Have Personally Paid For and What They Cost
  • GlockApps
  • Folderly
  • Mail-Tester
  • Inboxalysis
• Authentication Is Now a Moving Target Not a Setup Task
  • My recurring checks
• My Monthly Deliverability Routine You Can Copy
  • Week 1
  • Week 2
  • Week 3
  • Week 4

The Deliverability Problem That Surprised Me

The problem appeared in the most ordinary way. I opened my ESP dashboard after sending a newsletter and saw a 98.6% delivery rate. The campaign looked healthy. No obvious bounce spike appeared, and the platform marked the send as successful.

Then I checked my Gmail seed accounts. The same message reached the inbox in only part of the test group. Gmail placed 64% in Promotions and 11% in Spam. The ESP had counted the messages as delivered because the receiving servers accepted them. It hadn't checked whether Gmail considered them useful inbox mail.

I had sent the campaign to 12,400 subscribers. My sender score was 91 before the discovery, but the next review showed 74. That week, my click rate dropped 22%, followed by a noticeable unsubscribe spike. I couldn't attribute every unsubscribe to folder placement, but the timing was too close to ignore. Readers who never saw the message in a primary reading surface couldn't click it, and some who did see it had already trained their mailbox to treat my emails as low priority.

The practical distinction: SMTP acceptance tells me the receiving server took the message. Inbox placement tells me where the reader can find it.

The benchmark data changed how I interpreted that dashboard. In 2026 independent testing across 15 major email service providers, average deliverability was 83.1%. 10.5% landed in spam and 6.4% were undelivered, according to EmailToolTester's deliverability statistics. A “delivered” status clearly isn't the same as a visible inbox location.

I changed my process immediately. I stopped approving campaigns from the ESP dashboard alone. I added seed accounts, checked Gmail tabs separately, opened raw headers, tracked authentication alignment, and compared the test cohort with the live list. That shift also mattered during my migrations between beehiiv, Substack, Ghost, and LetterBucket. Each platform reported delivery differently, so I needed a test method that stayed consistent while the sending infrastructure changed.

The Five Numbers I Track on Every Test Send

I don't use one universal score. I track five numbers in a small sheet for every meaningful test send, then compare the result with the previous campaign from the same sending domain.

1. Inbox placement rate

I record the percentage reaching the primary inbox at Gmail, Outlook, and Yahoo. My personal ship threshold is 85% or higher. Anything below that triggers remediation before a broad send.

That threshold is stricter than the general red-flag guidance that treats placement below 80% as a serious warning and roughly 90–95% or higher as a common operational target, as outlined in Unspam's email deliverability benchmark. I use 85% as an internal line because I'd rather investigate a borderline result than discover the issue through a live campaign.

2. Spam-folder placement

Spam placement blocks a send for me when it rises above 5%. I don't average it away with strong Gmail results. A provider-specific failure can still damage a segment of the list.

I record Spam separately from undelivered messages. The distinction matters because a missing message suggests a different investigation from a visible spam placement.

3. Gmail Promotions placement

Promotions isn't the same as Spam, but I treat it as a distribution signal. If more than 70% of Gmail tests land in Promotions, I rework the subject line and inspect the HTML before shipping.

I don't automatically force every newsletter into Primary. Gmail may correctly classify a commercial newsletter as Promotions. The problem is a sudden change after a platform migration, template redesign, or sending-domain update.

4. Authentication pass rate

I want SPF, DKIM, and DMARC to pass and align with the From domain. The 2026 infrastructure index found that 76.9% of domains published SPF, 65% published DMARC, and only 23.1% enforced DMARC with a reject policy, according to MailTester Ninja's infrastructure index. Publication alone isn't enough. My test records whether the receiving mailbox sees a pass and whether the policy protects the domain.

5. Engagement-to-delivery ratio

I compare opens and clicks from the test cohort with the corresponding live-list baseline. This isn't a perfect measurement, especially because open tracking has limitations, but it gives me an early directional signal.

When that ratio weakens while placement still looks stable, I investigate before the next two sends. In my experience, engagement often changes before folder reports show a clear decline. It catches a quiet relevance or reputation problem while I can still isolate the cause.

My Pre-Send Testing Checklist

I run this checklist 18–24 hours before the planned send. That leaves enough time to fix a bad result instead of rushing a campaign into the queue.

First, I send the draft to Mail-Tester. My minimum score is 9/10. For a product launch, I want 10/10. I treat the score as a sanity check, not an inbox-placement verdict. Mail-Tester is fast and useful, but it examines one test destination, so it can't tell me how Gmail, Outlook, Yahoo, or a corporate quarantine system will handle the campaign.

What I inspect in the message itself

I open the raw headers and verify three alignment points:

• Return-Path alignment: The bounce-handling address should make sense for the sending setup.
• From-domain match: The visible From domain should align with the authenticated sending identity.
• List-Unsubscribe presence: I check that the header exists and that the unsubscribe path works.

I also confirm the SPF record begins with v=spf1, then use Gmail's “Show original” view to check that SPF, DKIM, and DMARC show PASS, following the workflow described by Warmforge's deliverability testing guide.

My internal test group contains five Gmail accounts, three Outlook accounts, two Yahoo accounts, and one AOL account. I opened all of them within the last 30 days, so they aren't abandoned mailboxes with unrealistic behavior. I check Primary, Promotions, Updates, Spam, and any visible review or quarantine folder.

Content and link checks

I run a spam-word scan, then check every campaign link against the two anchor services I use for blacklist monitoring. I don't assume a familiar domain is safe. Redirects, tracking parameters, and recently changed landing pages can all create surprises.

My image-to-text rule is simple: images stay under 60% of the message. I preview the email with images blocked, read the plain-text version, and click every link from both desktop and mobile test accounts. For list growth, I also review the signup flow and confirmation experience through the publication's double opt-in guidance.

Why Seed Lists Are the Only Test That Actually Matters

An ESP's “delivered” metric answers a narrow question. It tells me the receiving server accepted the message. A seed-list test answers the question I care about: where did the message land?

I use a seed list that covers Gmail, Outlook, Yahoo, AOL, and selected corporate domains. The report separates Inbox, Promotions, Updates, Spam, and Quarantine when the provider exposes those surfaces. That distinction matters because a message can be visible without being in the primary reading environment. A corporate admin folder may be invisible to the recipient until someone reviews it.

I can assemble a paid seed list in about 30 minutes. A self-built list takes me 3–4 hours, mainly because I have to create accounts, keep credentials organized, and monitor whether providers start treating the addresses as artificial. My per-send cost has ranged from free to $4, depending on the service and test plan.

I use that migration table as an operating record, not a promise about future campaigns. Placement changes with list behavior, content, domain reputation, and provider rules. A single result never earns a permanent verdict.

What failed in my seed-list experiments

My self-built list gave me control, but it underrepresented corporate filtering. It was useful for Gmail and Yahoo checks, yet weak for workplace quarantine behavior. Gmail-focused paid tools gave me a detailed view of tabs but underweighted Outlook, which made them less useful for a mixed audience.

Free tools created another problem. After repeated sends, some began blocking or limiting my seed accounts because the pattern looked artificial. I kept them for quick checks, but not as my only evidence.

I use GlockApps or Folderly previews when I need a fast directional read and the campaign is low risk. For a product launch, a platform migration, or a sudden reputation change, I pay for actual seed coverage. Independent guidance also recommends checking placement alongside authentication, complaints, bounces, and domain or IP reputation, rather than treating one placement result as complete evidence, as explained in Unspam's testing guidance.

Tools I Have Personally Paid For and What They Cost

I pay for testing tools when they answer a question I cannot answer from a header check or one seed inbox. I kept the subscriptions that exposed provider-level differences and cancelled those that repeated the same information. The useful distinction is placement across Gmail tabs, Outlook folders, consumer inboxes, and workplace filtering, not a single combined score.

GlockApps

I paid $47 per month for GlockApps. Its SpamAssassin score and folder reporting worked well for weekly checks, especially when I needed a quick comparison across mailbox providers. The reports were practical and easy to export.

The interface felt dated. I sometimes clicked through several screens to reach provider-level detail. I would use it for a newsletter sending under 10,000 emails per issue that needs recurring tests at a moderate cost. I would not subscribe for polished reporting alone.

Folderly

Folderly cost me $99 per month. Its clearest advantage was the Gmail Promotions-versus-Inbox split. That view helped me separate Gmail categorization from a genuine spam placement problem, which matters when a campaign appears healthy in the inbox but is being pushed into a lower-attention tab.

Outlook coverage was the trade-off. Its Outlook interpretation felt less convincing than its Gmail reporting, so I did not use it as the only test for an audience with many Microsoft recipients. Folderly makes more sense when Gmail dominates the list, the sending domain is dedicated, and the testing budget supports a paid view.

Mail-Tester

Mail-Tester costs $0 for a single test. I use it as the fastest pre-send sanity check before sending a draft to a wider seed group. Its score prompts a review of authentication, content, and links.

The test covers one destination. It does not show how the message behaves across Gmail, Outlook, Yahoo, Promotions, Updates, or quarantine folders. I run it before every send, while keeping its result separate from inbox-placement evidence.

Inboxalysis

Inboxalysis cost me $149 per month. Of the paid tools I tried, it covered the broadest mix of domains. That helped during my migration checks across Ghost and LetterBucket, where I wanted provider comparisons rather than a Gmail-only view.

Its seed mix skewed toward consumer mailboxes. Corporate quarantine coverage was thinner, so it was less useful for workplace-heavy audiences. I would choose it for newsletters sent across many consumer domains, not for a list dominated by business mailboxes.

My current workflow uses LetterBucket for newsletters, Mail-Tester for every draft, and paid seed testing when a send carries real risk. I have also run newsletters through beehiiv, Substack, and Ghost. beehiiv kept campaign operations straightforward, but I wanted more control over the surrounding sending setup during tests. Substack reduced setup work and gave me less control over infrastructure decisions. Ghost provided more ownership, while requiring more configuration and monitoring from me.

The publisher behind this article, Grow and monetize your newsletter, records newsletter growth and monetization from a first-person operator's perspective. I treat that record differently from a generic feature comparison, then verify deliverability with my own tests.

Authentication Is Now a Moving Target Not a Setup Task

I used to configure SPF, DKIM, and DMARC, run one header check, and move on. That approach no longer works for me. The 2026 infrastructure data showed a wide gap between domains publishing authentication and domains enforcing DMARC, with only 23.1% using a reject policy, as reported by MailTester Ninja.

Google and Yahoo bulk-sender requirements also make authentication an ongoing operational check. The referenced guidance requires authenticated mail, alignment between SPF or DKIM and the From domain, and a published DMARC policy of at least p=none, as summarized by Mailflow Authority's Gmail and Yahoo update.

My recurring checks

Every Monday morning, I pull DMARC reports and parse them with postmark-tools or Valimail's free analyzer. I check for new unauthorized sources, alignment failures, and changes in policy behavior.

After I add a sending service, I run an SPF lookup count check. The technical limit is 10 DNS lookups, so adding tools without reviewing the record can create a failure that doesn't appear in ordinary campaign testing, as explained in SendBridge's technical checklist.

I verify the DKIM selector whenever an ESP rotates keys. I also review the DMARC policy monthly and run a full authentication audit quarterly. If I need a broader remediation checklist, I keep this email deliverability improvement guide nearby.

One failure came from a forgotten Substack forwarder. My DMARC failure rate jumped from 0.4% to 7% before I traced it. Another appeared after I added a third CRM, which pushed the SPF setup into a lookup error. Neither issue came from the newsletter editor itself. Both required recurring infrastructure checks.

My Monthly Deliverability Routine You Can Copy

I run a seven-day cycle each month, then repeat it after major changes. The routine is deliberately boring. Boring makes it easier to notice when the baseline moves.

Week 1

I run a full seed-list placement test across Gmail, Outlook, Yahoo, and AOL, using at least 50 seeds per mailbox provider. I record Inbox, Promotions, Updates, Spam, and Quarantine where available. I save the raw headers from representative messages instead of relying only on the dashboard.

Week 2

I spot-check SPF, DKIM, DMARC, and BIMI alignment. I compare the current records with the prior month and investigate any new sending source. I don't wait for a campaign failure to discover that a CRM or automation tool has started sending through the domain.

Week 3

I send a 2,000-recipient internal A/B test and score engagement against the live-list baseline. I use the result to catch subject-line, content, or audience problems before the next broad campaign.

Week 4

I review suppression-list growth and spam complaint rates, then reset the baseline for the next cycle. I pay attention to provider-level changes, not just the blended list average.

My repair order is strict:

1. Authentication misalignment first. A failed SPF, DKIM, or DMARC result can undermine every later placement observation.
2. Spam placement at major providers next. I pause the campaign and inspect links, content, complaints, and reputation.
3. Inbox placement after that. I compare providers and surfaces rather than averaging away a weak segment.
4. Promotions drops last. I revise subject lines and commercial formatting only after the technical and spam issues are clean.

I ignore sub-0.5% open-rate variation, single-recipient anomalies, and complaints from role addresses when they don't form a pattern. My launch blockers are inbox placement below 85%, spam placement above 5%, Promotions placement above 70%, or any authentication failure. A campaign is ship-able when those checks pass and engagement remains consistent with the test cohort.

If a result still looks ambiguous after the monthly routine, I'd rather pause and get a specialist review than guess. Email deliverability consultants can help when provider behavior, authentication sources, and platform migrations overlap.

Set up your own test sheet today. Add Gmail, Outlook, Yahoo, and AOL seeds, send your next draft 18–24 hours early, inspect the raw headers, and record every mailbox surface separately. Don't approve a campaign because your ESP says “delivered.” Approve it when the message reaches the right places, authentication passes, and your test engagement looks normal.