If your open rates dropped without warning, or if customers keep telling you they never received their promo code, you do not have a copywriting problem. You have a deliverability problem. Email deliverability is the percentage of your sent emails that actually reach the inbox instead of the spam folder or a silent black hole at the receiving mail server.

Most guides give you a generic list of best practices. This one is built as a diagnostic checklist. Each of the nine fixes below starts with the symptom that reveals the problem, explains what to change, and tells you exactly which metric should move afterwards. Work through them in order, because the technical layer has to be solid before behavioural signals can save you.

Quick diagnosis: match your symptom to the right fix

Symptom you are seeing	Most likely cause	Go to
Emails land in spam at Gmail but fine at Outlook (or vice versa)	Authentication gaps or per-mailbox reputation damage	Fix 1, 2, 3, 5
Open rate fell 30% or more in a few weeks	Sender reputation decline from low engagement	Fix 5, 7
Hard bounce rate above 2%	Dirty list, purchased data, typo addresses	Fix 6
Spam complaints above 0.3%	Unclear consent, hidden unsubscribe, over-sending	Fix 7, 8, 9
Transactional emails (order confirmations) delayed or missing	Marketing and transactional traffic share one domain	Fix 4
New sending platform, terrible first results	No warm-up on a cold domain or IP	Fix 4

Fix 1: Publish a correct SPF record

The symptom

Your email headers show spf=fail, spf=softfail or spf=none. Send a test message to a Gmail address, open it, click the three dots and choose “Show original”. If SPF is not a clean PASS, stop everything else and fix this first.

What to do

SPF (Sender Policy Framework) is a DNS TXT record that lists which servers are allowed to send email using your domain. An online store typically sends from several sources at once: the email marketing platform, the e-commerce platform for receipts, the helpdesk, and the review request tool.

• Keep one single SPF record per domain. Two SPF records equal an automatic failure.
• Stay under the 10 DNS lookup limit. Every include: statement counts. If you exceed it, the record returns permerror and fails.
• End with -all (hard fail) once you are certain every legitimate sender is listed. Use ~all only while testing.
• Remove includes for tools you stopped using two years ago. They are unnecessary risk and wasted lookups.

A typical record looks like this: v=spf1 include:_spf.yourplatform.com include:servers.mailprovider.net -all

Metric to watch

SPF pass rate in your DMARC aggregate reports should reach 100% for your legitimate sources within 48 hours of the DNS change.

Fix 2: Sign every message with DKIM

The symptom

Headers show dkim=none, or your emails display the dreaded “via mailprovider.net” tag next to your sender name in Gmail. Recipients see a mismatch between who you claim to be and who technically sent the message.

What to do

DKIM adds a cryptographic signature to each message so the receiving server can verify that the content was not altered and that the sending platform was authorised by your domain.

1. Generate the DKIM keys inside every sending platform you use. Each one gets its own selector.
2. Publish the CNAME or TXT records your provider gives you in your DNS zone.
3. Use a 2048-bit key when the provider offers it.
4. Verify that the signing domain (the d= value) matches your own domain, not your provider’s. This is what creates DKIM alignment, which DMARC needs.
5. Rotate keys at least once a year.

Metric to watch

DKIM alignment rate at 100% in DMARC reports, and the disappearance of the “via” label in Gmail.

Fix 3: Move DMARC from none to enforcement

The symptom

You have no DMARC record at all, or it sits at p=none and has done so for years. Since 2024, Gmail, Yahoo and Microsoft all require bulk senders to publish a DMARC policy. No policy means throttling or rejection at the largest mailbox providers.

What to do

DMARC tells receiving servers what to do when SPF and DKIM alignment fail, and it sends you reports about who is using your domain.

• Start with v=DMARC1; p=none; rua=mailto:[email protected]; and collect reports for two to four weeks.
• Read the reports with a parsing tool. Identify every legitimate source, fix its alignment, and spot spoofing attempts.
• Move to p=quarantine; pct=25, then raise the percentage gradually.
• Finish at p=reject. This is the level that best protects your brand from phishing that damages your reputation.

While you are in DNS, consider adding BIMI once you reach p=quarantine or higher. It displays your logo next to your emails in supporting inboxes, which lifts recognition and open rates.

Metric to watch

Percentage of messages passing DMARC (target above 99%) and the volume of unauthorised sending sources, which should fall to zero.

Fix 4: Separate your streams and warm up properly

The symptom

Order confirmations and shipping notifications start arriving late or landing in spam right after a big promotional campaign. Or you switched platforms last month and your numbers collapsed. Much the same conclusion turns up on braze.com.

What to do

Mailbox providers build reputation at the subdomain and IP level. Mixing a low-engagement promotional blast with your transactional traffic drags the critical emails down with it.

Stream	Suggested subdomain	Why
Receipts, shipping, password resets	mail.yourstore.ca	High engagement, must never be blocked
Newsletters and promotions	news.yourstore.ca	Variable engagement, isolates risk
Re-engagement and win-back	reach.yourstore.ca	Lowest engagement, highest complaint risk

For warm-up on a new domain, subdomain or dedicated IP, ramp volume over three to four weeks and always start with your most engaged contacts:

1. Days 1 to 3: 200 to 500 emails per day to people who opened something in the last 30 days.
2. Days 4 to 10: double daily volume every two days, still 90-day openers only.
3. Days 11 to 21: extend to 180-day engagement, keep doubling.
4. Day 22 onward: full volume, but hold back anyone inactive for more than 12 months.

Note that a dedicated IP only makes sense above roughly 50,000 emails per month with consistent frequency. Below that, a well-managed shared IP pool is usually safer.

Metric to watch

Delivery rate per domain (Gmail, Outlook, Yahoo, iCloud tracked separately) and deferral rate. Rising 4xx deferrals mean you are ramping too fast.

Fix 5: Monitor sender reputation with real data, not guesses

The symptom

You know something is wrong but you cannot tell where. Your platform reports a 99% “delivered” rate while sales from email are down 40%. Remember: delivered means accepted by the server, not placed in the inbox.

What to do

• Set up Google Postmaster Tools and verify your sending domain. It shows domain reputation, IP reputation, spam rate and authentication pass rates for Gmail.
• Register for Microsoft SNDS and JMRP if you use a dedicated IP, to get Outlook and Hotmail data plus complaint feedback.
• Subscribe to feedback loops where available so complaints automatically suppress the contact.
• Check blocklists monthly (Spamhaus, SpamCop, Barracuda). A listing explains sudden total blocks.
• Run seed list tests before large campaigns to see actual inbox versus spam placement across providers.

Gmail’s rules are explicit: keep your spam complaint rate below 0.3% and ideally under 0.1%. Crossing 0.3% even once can trigger filtering that takes weeks to recover from.

Metric to watch

Domain reputation in Google Postmaster Tools should sit at High or Medium. Reported spam rate below 0.1%.

Fix 6: Clean the list and stop letting garbage in

The symptom

Hard bounces above 2% on a normal send. Or you notice a large chunk of addresses that have never opened a single email since the day they subscribed.

What to do

List hygiene works in two directions: what you remove, and what you prevent from entering.

Cleaning the existing list:

• Suppress every hard bounce immediately and permanently.
• Suppress soft bounces after three or four consecutive failures.
• Remove role addresses such as info@, sales@, admin@, contact@. They generate complaints and rarely buy shoes.
• Run a validation service before re-activating any list you have not mailed in six months or more.
• Never import purchased or scraped lists. They are riddled with spam traps, which are recycled or pristine addresses that mailbox providers use to catch senders with bad acquisition practices.

Keeping it clean at the source:

1. Use double opt-in for newsletter signups. Yes, your list grows slower. Your revenue per email grows faster.
2. Add real-time email validation on your signup forms and at checkout to catch typos like gmial.com.
3. Protect forms with a captcha or honeypot field to block bot submissions.
4. Never pre-check the marketing consent box at checkout. That consent is worthless and it generates complaints.
5. Store the date, source and IP of every consent. Canadian anti-spam law (CASL) requires you to prove it.

Metric to watch

Hard bounce rate below 0.5%, total bounce rate below 2%, and unknown-user rejections trending toward zero.

Fix 7: Build engagement segments and a sunset policy

The symptom

Your list is big, your open rate is small, and it keeps falling every month. Mailbox providers read that pattern as “nobody wants this” and route you to spam, including for the people who do want it.

What to do

Engagement is now the strongest deliverability signal after authentication. Gmail and Outlook watch opens, replies, clicks, deletions without reading, and moves to the spam folder. Sending to unengaged contacts is the fastest way to lose the inbox.

Segment your list by recency of engagement and adjust frequency for each tier:

Segment	Definition	Sending frequency
Active	Opened or clicked in last 30 days	2 to 4 per week maximum
Warm	Engaged in last 90 days	1 to 2 per week
Cooling	Engaged 90 to 180 days ago	2 per month, best content only
Dormant	No engagement in 180 to 365 days	Re-engagement sequence only
Sunset	No engagement in 12+ months	Suppress completely

A re-engagement sequence that actually works is three emails over three weeks, sent from your re-engagement subdomain:

1. “Are we still your size?” A plain, mostly text email asking if they still want to hear from you, with one clear button to confirm.
2. “One last offer” with a genuinely strong incentive and a visible preference centre link.
3. “We are removing you on Friday” with a single stay-subscribed button.

Anyone who does not click gets suppressed. Do not argue with the data. A 20,000-contact list where 8,000 people engage will outperform a 50,000-contact list where 8,000 people engage, every single time.

Metric to watch

Open rate (target above 25% for e-commerce after cleaning), click-to-open rate above 10%, and revenue per recipient, which is the number that proves list pruning was profitable. A fuller account is out there.

Fix 8: Make unsubscribing easier than complaining

The symptom

Complaint rate creeping toward or above 0.3%, while unsubscribe rate stays suspiciously low. That gap means people cannot find the exit, so they hit the spam button instead. One spam complaint hurts roughly a hundred times more than one unsubscribe.

What to do

• Implement one-click unsubscribe (List-Unsubscribe and List-Unsubscribe-Post headers, RFC 8058). This is mandatory for bulk senders at Gmail, Yahoo and Microsoft. The request must be processed within two days.
• Put a visible, readable unsubscribe link in the footer. Dark grey 8-pixel text on a light grey background is a complaint generator.
• Offer a preference centre: fewer emails, only sales, only new arrivals, pause for 60 days. Many people want less, not nothing.
• Never require a login to unsubscribe. That practice alone can destroy a sender reputation.
• Add a “Why am I receiving this?” line reminding people where and when they subscribed.

Metric to watch

Complaint rate below 0.1% and a healthy unsubscribe-to-complaint ratio of at least 10 to 1.

Fix 9: Fix the content signals filters read

The symptom

Authentication is perfect, the list is clean, engagement is decent, and specific campaigns still get filtered while others do not. The problem is inside the message.

What to do

• Use a consistent From name and address. Changing it every campaign resets recognition. Avoid noreply@ addresses, and make sure the reply-to inbox is monitored, because replies are a strong positive signal.
• Keep a reasonable text-to-image ratio. A single giant image with no text is a classic spam pattern. Always fill alt text.
• Send a plain text version alongside your HTML. Multipart messages score better.
• Avoid link shorteners such as bit.ly in marketing emails. They are heavily abused and often blocklisted.
• Use your own tracking domain (links.yourstore.ca) rather than the shared one from your platform.
• Watch the wording. Modern filters do not reject on single words, but stacking “FREE!!! 80% OFF LIMITED TIME ACT NOW” with all caps, red fonts and five exclamation marks in the subject line still triggers them.
• Keep the HTML clean and light. Under 102 KB avoids Gmail clipping. Do not paste content from Word.
• Be consistent in volume and rhythm. Sending 500 emails weekly then suddenly 80,000 for Black Friday looks like a compromised account. Ramp volume up over the two weeks before peak season.

Metric to watch

Inbox placement rate from seed testing (target 95% or higher) and per-campaign spam rate in Postmaster Tools.

Your 30-day deliverability recovery plan

Week	Actions	Expected outcome
Week 1	Audit SPF, DKIM, DMARC. Connect Postmaster Tools and SNDS. Check blocklists.	Full authentication pass, baseline data collected
Week 2	Suppress bounces and role addresses. Validate the list. Build engagement segments. Send only to 30-day actives.	Bounce rate under 1%, open rate climbing
Week 3	Add one-click unsubscribe and preference centre. Launch the re-engagement sequence on the dormant segment.	Complaint rate falling below 0.1%
Week 4	Suppress non-responders. Extend sending to 90-day segment. Run a seed test. Set DMARC to quarantine.	Domain reputation at Medium or High, inbox placement above 90%

Benchmarks for e-commerce senders

Metric	Healthy	Warning	Critical
Delivery rate	above 98%	95 to 98%	below 95%
Inbox placement	above 95%	85 to 95%	below 85%
Spam complaint rate	below 0.1%	0.1 to 0.3%	above 0.3%
Hard bounce rate	below 0.5%	0.5 to 2%	above 2%
Unsubscribe rate	below 0.3%	0.3 to 0.7%	above 0.7%

Frequently asked questions

How to fix email deliverability issues fast?

Start with authentication, because it is the only fix that produces results within hours. Verify that SPF, DKIM and DMARC all pass on a test message, then immediately stop sending to anyone who has not engaged in the last 90 days. Those two actions alone resolve the majority of sudden deliverability drops in online stores. Reputation repair after that takes two to six weeks of consistent, engaged-only sending.

What is the difference between delivery rate and deliverability?

Delivery rate measures messages accepted by the receiving server, which includes everything filed in the spam folder. Deliverability, or inbox placement, measures messages that actually reach the primary inbox. You can have a 99% delivery rate and a 40% inbox placement rate at the same time. Only seed testing and Postmaster Tools reveal the difference.

What is the 30/30/50 rule for cold emails?

It is a volume guideline used in cold outreach: roughly 30 emails per day per mailbox during the first 30 days of warm-up, scaling toward about 50 per day per mailbox once the domain is established. It applies to prospecting, not to permission-based e-commerce marketing. For an online store, volume should be driven by your engagement segments and your warm-up curve rather than a fixed number, and every recipient should have opted in.

What is the 3 email rule?

The 3 email rule says that if a topic takes more than three back-and-forth emails to resolve, you should switch to a call or a chat. In marketing automation it also describes a common three-message structure, such as the three-step re-engagement sequence in Fix 7 or the classic three-email abandoned cart flow.

What are the 5 C’s of email?

Clear, concise, correct, courteous and complete. Applied to deliverability, they matter because messages that are easy to read and honest about their purpose get opened and clicked instead of reported, and engagement is exactly what mailbox providers measure.

Does a free email deliverability test tell me everything?

A free test is useful for checking authentication records and spotting obvious content flags, but it only reflects one message at one moment. Real deliverability monitoring means tracking domain reputation, complaint rate and inbox placement across providers over time. Use free tools to diagnose, use Postmaster Tools and seed lists to monitor.

Should I use a dedicated IP address?

Only if you send at least 50,000 emails per month with a regular cadence. Below that volume you cannot generate enough consistent signal to build a reputation, and a cold dedicated IP will perform far worse than a well-managed shared pool. Domain reputation now matters more than IP reputation at most major providers anyway.

How long does it take to recover from being marked as spam?

With authentication corrected and a strictly engaged-only sending list, expect visible improvement in two to three weeks and full recovery in four to eight weeks. Recovery is slower if you were blocklisted or if your complaint rate stayed above 0.3% for an extended period. There is no shortcut: the only thing that rebuilds reputation is a sustained record of people opening and clicking your emails.

The bottom line

Knowing how to improve email deliverability comes down to proving two things to mailbox providers: that you are who you say you are, and that people genuinely want your emails. Authentication handles the first. List hygiene, engagement segmentation and easy unsubscribes handle the second. Work through the nine fixes in order, track the metric attached to each one, and treat deliverability as a monthly routine rather than an emergency you handle once a year.