How do you improve email deliverability? A guide to finding and fixing the problem

To improve email deliverability, first distinguish rejection, delay and spam placement from weaker customer engagement. An open-rate drop alone does not tell you which happened, or establish that there is a delivery failure.

For a store sending retention campaigns, the useful question is specific: which messages, to which customers, at which mailbox provider, stopped arriving as expected? That gives you something to fix and something to check afterward.

How to improve email deliverability. Check delivery errors and actual message authentication first. Send marketing only to opted-in recipients, remove invalid addresses, and make unsubscribing easy. Adjust sending volume to provider feedback. Verify recovery with delivery diagnostics and inspected messages; a higher open rate or a passing DNS check alone does not establish inbox placement.

What should you check before changing anything?

Check the full error code and description, affected provider, campaign and time, and original headers of an affected message. Separate server acceptance, folder placement and customer response before choosing a fix.

The M3AAWG industry guidance distinguishes temporary and permanent delivery failures and advises reading the full error description. Your email platform’s summary of “hard” or “soft” bounces may leave out the detail needed to diagnose them.

Use this table to choose the next investigation. The symptoms suggest checks; they do not prove a cause.

Collect an affected message and an unaffected comparison where possible. Record the campaign, date, recipient provider, sending domain and exact failure text. If only one provider is rejecting mail, preserve that distinction instead of treating your whole list as one delivery result.

How do you check authentication and sender requirements?

Open an affected message’s full header and check its SPF, DKIM and DMARC results and the domains involved. In desktop Gmail, select More beside Reply, then Show original. Published DNS records alone do not establish a pass: for DMARC on direct mail, a passing SPF or DKIM domain must align with the visible From domain.

Google documents the full-header inspection path. Use a message sent through your actual campaign system, then check the configuration behind it.

SPF checks whether the sending server is authorized for the envelope-sender domain. DKIM validates a domain’s message signature. DMARC requires a passing SPF or DKIM mechanism to align with the domain in the visible From address. For bulk sending, a DMARC pass through one aligned mechanism does not remove a provider’s separate requirement for both SPF and DKIM to pass.

The rules differ by recipient provider:

These are the providers’ own Gmail FAQ, Yahoo requirements, Yahoo bulk-sender definition and Microsoft error guidance. The consumer-service rules are not a blanket description of every corporate mailbox using Google or Microsoft software.

Gmail also requires valid forward and reverse DNS and encrypted transmission through TLS. If an error points to the sending IP or transport, ask your email provider to check its infrastructure rather than copying an unrelated DNS fix.

For bulk marketing mail to Gmail, check RFC 8058 one-click unsubscribe headers as well as a visible unsubscribe link in the body. A preference-page link alone does not meet the header requirement. Google’s FAQ calls for processing unsubscribes within 48 hours; Yahoo specifies two days.

Passing these checks removes identified compliance failures. It does not establish successful inbox placement: Google explicitly says authentication alone cannot guarantee delivery.

Which subscribers should you keep sending to?

Keep sending marketing to people who asked for your mail and still match what they signed up for. Yahoo recommends honoring the frequency promised at signup and promptly removing invalid addresses. An address that exists is not evidence that its owner wants your newsletter.

M3AAWG treats informed opt-in as the industry best practice. Check the audience attached to the problem campaign: where the subscribers came from, what they opted into, whether they unsubscribed, and whether your current sends match that promise. Suppress confirmed invalid addresses. For a rejected message with a valid address, investigate the described policy or configuration failure before deciding the address itself is bad.

During diagnosis, a practical way to narrow promotional sending is to combine opt-in status with relevant signals such as clicks, orders, replies or confirmed preferences. Choose the period those signals cover from the buying cycle and mailing frequency. A monthly newsletter and a frequently reordered product do not justify the same inactivity cutoff. The email segmentation guide covers how to group audiences for different purposes.

Avoid deciding solely from “opened in the last few weeks.” Apple’s Mail Privacy Protection can fetch remote content regardless of whether someone engages with the message. That limits what image-based opens establish about a human reader.

An inactive audience also deserves a separate decision from your regular campaign audience. If you choose a win-back email, keep the eligibility and stop conditions explicit. A recovery send still needs to respect opt-in status and delivery feedback.

Should you change sending volume, domain or IP?

Change what the evidence implicates. Google recommends a consistent sending rate and reducing volume when messages bounce or are deferred. A new domain or dedicated IP needs its own warming, so neither is a default recovery step.

Google’s sending guidance recommends a consistent rate, starting with engaged recipients and increasing gradually. When messages bounce or are deferred, it advises reducing volume until the SMTP error rate falls, then increasing slowly again. The provider response determines the adjustment; a fixed ramp schedule is not proof that the next increase is appropriate.

Before a large sale campaign, check whether the planned audience or volume departs sharply from your established sending pattern. If it does, review the audience and pacing with your provider rather than use the sale as an experiment in maximum sending volume.

Infrastructure changes need their own justification:

• A new domain: M3AAWG recommends warming new sending domains and subdomains carefully. It discourages artificial warming and using a lookalike domain to bypass reputation problems.
• A dedicated IP: M3AAWG says shared environments can support consistent aggregate traffic. Dedicated environments need their own warming and ongoing volume consistency. Ask what the move solves for your actual traffic before making it part of a recovery plan.
• Different mail streams: Yahoo recommends separating marketing from transactional or user traffic by IP or DKIM domain. Review how your provider implements that separation. Do not assume a new subdomain makes all provider rules independent; Gmail still aggregates its bulk threshold across the primary domain.

How do you know whether deliverability is improving?

Check the failure you set out to fix, using the same provider and mail stream. A campaign with more clicks may have a better offer or a different audience; that alone does not show improved placement.

Google Postmaster Tools reports personal-Gmail authentication, reputation, delivery errors and spam reporting. Low-volume days can have incomplete data. An empty dashboard is therefore an information gap, not a clean bill of health.

Its spam rate also needs careful interpretation. Google explains that user-reported spam can be low when its filters already send substantial mail to spam: fewer messages reach the inbox where users could report them. Read the rate alongside reputation, errors and actual messages.

Google recommends keeping its Postmaster spam rate below 0.1% and avoiding 0.3% or higher. Yahoo requires its own inbox-based complaint rate below 0.3%. Those are provider-specific rates; do not substitute complaints divided by your platform’s total sends and assume the result is equivalent.

Keep a short change record: affected provider, campaign, audience definition, errors before the fix, authentication results, change made, and results afterward. If you inspect messages in test or customer mailboxes, record their actual folder placement. That establishes what happened to those messages, not a placement percentage for your whole audience.

What should you send your email provider if the problem persists?

Send a specific failure packet so support can investigate the same event you are seeing:

• The affected campaign, timestamps and recipient providers.
• Full rejection or deferral codes and their descriptive text.
• Original headers from affected messages, including authentication results.
• Recent changes to domain, sending configuration, audience or volume.
• Relevant provider diagnostics and an unaffected comparison, if available.

Ask support to identify the failed requirement, infrastructure issue or reputation signal and the evidence that would show it has improved. Resume expansion against that evidence. Compliance checks, wanted messages and controlled sending belong together; none supplies a universal recovery deadline.