11 Factors That Affect Email Deliverability

Email deliverability rarely collapses because of one forbidden word.

It usually declines because several signals begin pointing in the wrong direction. Things such as using an old list, a sudden jump in volume, weak authentication, rising complaints, and messages that no longer match what people expected when they subscribed can all be attributed to declining deliverability.

Mailbox providers do not publish a simple scoring formula. They evaluate patterns. These are the 11 factors I would examine first for an ecommerce sender.

1. Email authentication

SPF identifies which systems may send on behalf of your domain. DKIM adds a cryptographic signature that helps prove the message was authorized and unchanged. DMARC checks whether SPF or DKIM aligns with the domain customers see in the From address and tells receiving systems how to handle failures.

[https://support.google.com/mail/answer/81126](Google requires) SPF or DKIM for all senders to personal Gmail accounts. Senders exceeding 5,000 messages a day must use SPF, DKIM, and DMARC. Google also requires a DKIM key of at least 1,024 bits and recommends 2,048 bits when the provider supports it.

Authentication is baseline infrastructure, not an inbox guarantee. When I audit it, I check every legitimate source, not only the main marketing platform. Receipts, help-desk replies, review requests, and loyalty tools can all send as the brand.

2. Domain and IP reputation

Your history follows the domain and sending infrastructure used by your email program. Consistently wanted mail builds trust. Repeated complaints, invalid addresses, suspicious volume patterns, and blocklist appearances weaken it.

Many smaller ecommerce brands use shared sending IPs managed by an email service provider. Google notes that activity from other senders on a shared IP can affect everyone using it. A dedicated IP provides more control, but it also makes you solely responsible for building and maintaining its reputation.

I would not move to a dedicated IP simply because deliverability fell. First determine whether the real problem is the list, consent, authentication, cadence, or content. New infrastructure does not erase the behavior that damaged the old infrastructure.

3. Permission and acquisition source

A subscriber who intentionally signs up on your product page is different from an address acquired through a vague co-registration giveaway.

List source predicts future behavior. Bought and scraped lists are especially dangerous because recipients did not request your email. Google explicitly advises senders not to purchase addresses or send to people who did not sign up, because those recipients may report the messages as spam.

My practical test is simple: can the store explain when and where it obtained the address, what the person agreed to receive, and which brand the person expected to hear from? If the answer is unclear internally, it will probably be unclear to the recipient too.

4. List quality and age

Invalid, abandoned, misspelled, and unmonitored addresses create bounces and weak engagement. Even a once-valid list becomes riskier when it sits untouched for a long period.

Google recommends confirming addresses before subscription, periodically checking whether recipients still want the mail, and considering the removal of people who remain disengaged. Those are useful principles, but there is no universal “clean after exactly 90 days” rule. A weekly fashion promotion and an annual replenishment reminder have different natural engagement windows.

Use validation at capture, suppress hard bounces, and run a genuine re-engagement effort before sunsetting long-term nonresponders. Keep the detailed subscriber-record process in your list-management system; here, the purpose is protecting sender reputation.

5. Spam complaints

A complaint is stronger than an unsubscribe because the recipient is telling the mailbox provider that the message should not be there.

Google recommends keeping the Postmaster Tools spam rate below 0.10% and avoiding 0.30% or higher. Put differently, one user complaint per 1,000 delivered Gmail messages is already the upper edge of Google’s preferred operating range. The 0.30% level is a line to avoid, not a performance target. You’ll want lower than that.

Complaints often reveal an expectation mismatch: the person did not recognize the sender, forgot signing up, received too many messages, or expected different content. I look for complaint concentration by signup source, campaign, mailbox provider, and list age instead of treating the whole database as equally risky.

6. Bounce and deferral behavior

A hard bounce usually indicates a permanent failure, such as a nonexistent mailbox. Suppress it. A soft bounce or deferral is temporary, but repeated temporary failures still deserve investigation.

The exact server response matters. Google’s current guidance says that if messages begin bouncing or being deferred, senders should reduce volume until SMTP error rates decline and examine the individual errors before ramping up again.

I trust that evidence more than a universal bounce-rate benchmark. A sudden jump from your established baseline, concentrated at one provider or after one change, is usually more actionable than an all-industry average built from a different list mix.

7. Engagement and relevance

Clicks, replies, purchases, deletes, ignores, and complaints help providers infer whether people want your mail.

This is where segmentation affects deliverability. A recent buyer receiving care instructions has a clear reason to engage. A subscriber who ignored 60 promotions probably should not receive promotion 61 at the same frequency.

Do not diagnose this from opens alone. Google says low open rates are not necessarily evidence of a deliverability or spam-classification problem, and privacy features can inflate tracking-pixel activity. Compare clicks, orders, revenue, complaints, and provider-level changes with the open trend.

8. Sending volume and consistency

Large, sudden changes can look suspicious, especially on a new domain or IP.

Google advises high-volume senders to send at a consistent rate, avoid bursts, begin with engaged users, and increase volume gradually. Its guidelines warn that immediately doubling previous volume can lead to rate limiting or reputation drops.

Imagine a skincare store that usually sends 8,000 messages on Tuesdays. A creator giveaway adds 70,000 leads, and the store immediately sends the entire group a three-email launch sequence. Even if many addresses are real, the sending pattern and audience quality changed overnight.

I would separate recent, explicit opt-ins from uncertain leads, verify the source, prioritize engaged recipients, and ramp carefully while watching provider-level results. I would not begin by hunting for “spam words.”

9. Message content and link reputation

Content matters, but not through simplistic rules such as “never use FREE.” Providers evaluate the message in context.

Problems include deceptive subject lines, mismatched or obscured links, compromised landing pages, broken HTML, hidden text, and an identity the recipient does not recognize. Google specifically warns against fake Re: or Fwd: prefixes, concealed content, misleading display names, and links whose destination is unclear.

Honest urgency is fine. Deception is not. I check the domains behind buttons and tracking links as carefully as the visible copy, because one unsafe or misconfigured destination can affect the entire message.

10. Unsubscribe experience

Hiding unsubscribe does not retain subscribers. It pushes frustrated people toward the spam button.

Google requires marketing and subscribed messages from senders over its 5,000-message daily threshold to support one-click unsubscribe and include a clearly visible unsubscribe link. Its technical requirement uses the List-Unsubscribe-Post and List-Unsubscribe headers; a footer link by itself is not the same thing.

For U.S. commercial email, the FTC’s CAN-SPAM compliance guide also requires a clear opt-out method and says requests must be honored within 10 business days. A preference center can let people reduce frequency or choose categories, but it should never trap someone who wants to leave.

11. Infrastructure and configuration

TLS, valid forward and reverse DNS, message formatting, DKIM keys, tracking domains, and the configuration of every third-party sender can affect acceptance.

Google requires valid forward and reverse DNS and TLS for mail sent to personal Gmail accounts. It also requires messages to follow the RFC 5322 internet message format. These details often belong to the email platform or technical team, but the brand still owns the outcome.

This becomes important when several tools send as your store: the ecommerce platform sends receipts, the ESP sends marketing, a help desk sends support messages, and a review app sends requests. Inventory every source, decide whether it should be authorized, and authenticate it correctly.

A quick diagnostic framework

When performance changes, I group the possible causes before making edits:

Layer	Questions to ask
Identity	Are SPF, DKIM, DMARC, alignment, DNS, and sending domains correct?
Audience	Where did these contacts come from, and did they knowingly opt in?
Reputation	Did complaints, bounces, deferrals, or provider-level reputation change?
Behavior	Did volume, cadence, targeting, message type, or links change recently?
Outcome	Did clicks, orders, and revenue fall with opens, or did only opens change?

That structure keeps troubleshooting grounded. Start with the timeline, confirm the likely cause with evidence, and change the highest-risk variable first.

Deliverability is not one metric. It is the accumulated result of your technical setup, list practices, sending behavior, and relationship with subscribers.