Is your marketing email actually reaching people, or is it going to spam? The question sounds simple, but it is hard to answer from inside your email platform, because the reports say delivered for messages that never touched an inbox. Validity's 2026 Email Deliverability Benchmark Report measured average global inbox placement at 87.2 percent across 2025. The company's summary of the 2026 DMA report reaches the same conclusion in plainer terms, that roughly one in eight legitimate, permission-based marketing emails never reaches the inbox.

A newsletter to 5,000 subscribers loses around 600 of those sends every time, and the bill arrives regardless. The compounding part is worse. Readers who do receive the mail open it less, the ones who forgot they subscribed start hitting the spam button, and each campaign gives the filters another reason to score the next one lower.

This guide covers marketing and lifecycle email sent to people who asked for it, the newsletters, nurture sequences and announcements that go out from the platform you already pay for. When emails keep going to spam in that context, the problem is rarely the words in the message. Filters judge who you are, how you send and what readers do with the mail, and they keep that judgement until something changes. The routine that changes it works on the sender before the message. You prove you own the domain you send from, rebuild sending behaviour gradually with the readers who still want mail, prune the list, and keep complaints near zero. This post starts with the free check-up that tells you where you stand today, then walks through each part.

Emails going to spam: what the filter is actually judging

Every mailbox provider runs a reputation system in front of the inbox, and the biggest three score the sender on a small set of inputs before the content is even considered. Gmail and Yahoo began enforcing formal sender requirements in 2024, as Salesforce's explainer on email sending reputation documents, and the rules gave everyone a shared vocabulary for what filters weigh.

Identity is what receiving servers verify first. Authentication records called SPF, DKIM and DMARC, published in your domain's DNS settings, prove that mail claiming to come from your domain actually comes from you. Gmail's sender guidelines now require every sender to pass SPF or DKIM, bulk senders to add DMARC, and marketing messages to carry one-click unsubscribe. Yahoo's sender best practices demand the same, with a DMARC policy of at least p=none.

Then comes your sending history. Filters watch the volume that comes from your domain and IP addresses, and a sudden jump looks like a compromised account or a bought list. Large volume increases raise suspicion even for established senders, which is a point AWS's guide to IP and domain warming makes explicitly.

Reader behaviour rounds out the judgement. Gmail does not track opens, so an open is not part of the score. Complaints, spam marks, unread deletes and unsubscribes are. The complaint rate carries the most weight, and the published numbers are not abstract. Gmail's guidelines tell senders to keep the spam rate reported in Postmaster Tools below 0.10 percent and to avoid ever reaching 0.30 percent. Cross that ceiling and you become ineligible for mitigation until the rate stays below 0.30 percent for seven consecutive days, per Gmail's sender guideline FAQ.

The judgement is slow to build and quick to break, and it follows your domain rather than your email tool. Move from one platform to another and the reputation travels with the From address, which is why Signet's breakdown of sender reputation notes that domain reputation now outweighs IP reputation at Gmail. The fixes in this post target the domain and the list.

Where your mail lands today: the free check-up

Before changing anything, run a 20-minute check-up with two free tools. It tells you whether the problem sits with authentication, with your sending server, or with reader behaviour, and it gives you a baseline to compare against next month.

The first tool is mail-tester.com. You open the site, it shows you a unique test address, and you send one real campaign from your platform to that address. The report scores the message out of ten and lists what is misconfigured, covering SPF, DKIM, DMARC, blacklist status and content checks. One detail matters here. If the test message never arrives, the server you send from is probably blacklisted, which mail-tester's FAQ flags as a sending-method problem rather than a content problem.

The second tool is Google Postmaster Tools, the free dashboard for Gmail delivery. You add your sending domain, publish a small DNS record to prove you own it, and the dashboard shows your spam rate, domain and IP reputation ratings, authentication pass rates and delivery errors. Google's setup guide walks the verification in a few minutes. The catch is volume. Postmaster Tools only populates above a few hundred Gmail messages a day, so very small senders see not enough data, and mail-tester still works at any volume.

You are looking for three things in the results. Authentication failures mean the records in the next section are your first fix. A spam rate reading above 0.10 percent means behaviour is part of the problem, and a low or bad reputation rating means recent history is bruised, so the sending ramp is the answer.

Publish the three records that prove you own your domain

Receiving servers need proof that mail claiming to come from your domain really comes from you. The proof is three DNS records, and your email provider hands you the exact values to publish. You add them where you manage the domain, at the registrar or DNS host you bought it from, and each record takes a few minutes. Your provider's setup guide walks the same steps with their values, so the descriptions below are about what each record does rather than how to type it.

DKIM proves you wrote the message. The provider generates a key pair, you publish the public key as a text record, and the provider signs your outgoing mail with the private key. Google's plain-language guide to authentication puts it simply, DKIM says you wrote the email, SPF says your server sent it, and DMARC decides what happens to messages that fail either check. Setting DKIM up also stops others from impersonating your domain in phishing mail.

SPF lists the servers that are allowed to send for your domain. Google's SPF setup guide carries the rule that trips most teams up, a domain can publish only one SPF record. Every time you add a sending service, a new platform or a transactional provider for receipts and alerts, that service must be added to the record, or its mail fails the check.

DMARC tells receiving servers what to do with messages that fail SPF or DKIM. Google's DMARC setup guide recommends starting at p=none, which delivers the mail and sends you reports, then tightening to quarantine and reject as you confirm legitimate mail passes. Allow 48 hours after SPF and DKIM are live before enabling DMARC, and keep the report address so you can see what is failing.

A surprising share of deliverability problems start and end here. The business moved platforms, someone added a new sending tool, or the records were never set up, and the mail has been going out unauthenticated for years. Filters treat unauthenticated mail as an impersonation risk, so it is routed away from the inbox no matter who is actually sending it. Fixing the records is the cheapest part of this routine and the one with the fastest visible effect.

Rebuild a bruised sender reputation with a sending ramp

Records prove identity, but filters also remember history. If your mail has been landing in spam folders, or you have been sending to a list that has not engaged for months, expect a cautious reception for a while. The fix is a ramp, a deliberate return to normal volume aimed at the readers most likely to engage.

Start with the people who opened or clicked in the past 30 to 90 days. ActiveCampaign's first-30-days deliverability guide makes this the first rule for a new sending setup, and the same logic applies when you are recovering an old one. Send in batches small enough that you could read the complaint count by hand. Mailchimp's four-week domain warm-up plan sends week one to the most engaged 5 to 10 percent of the list and only reaches the least engaged part in week four, holding the ramp until open rates sit above 30 percent and clicks above 1 percent.

The principle matters more than the exact schedule. Send engaged readers first, grow volume gradually, and step back the moment anything spikes. A fresh domain or IP needs the same treatment, with AWS's warming guide recommending openers from the last 30 days before anyone else sees mail. Watch complaints and bounces daily while you ramp. An unsubscribe during the ramp removes someone who would have complained later, which is the outcome you want, while a complaint trains the filter against you.

Prune the list before it prunes you

Lists decay on their own. Bento's list hygiene documentation puts the yearly loss at 20 to 30 percent through job changes, abandoned addresses and shifting interest. The danger is what happens to those dead addresses. Providers recycle abandoned mailboxes into spam traps, so an address that was real five years ago can damage your reputation the moment you send to it again. Mailchimp's explainer on email sender reputation adds the other side of the ledger, hard bounces signal invalid addresses and purchased lists, and enough complaints can get an account suspended entirely.

The prune list is short and mechanical.

Address validation services tell you an address exists. They cannot detect a spam trap, and they cannot tell you whether someone still wants the mail, which is why the engagement rules in the next section do the real work.

Keep complaints low with re-engagement and sunset rules

Complaints are the heaviest signal in the filter's judgement, and the tolerance for them is thin. Mailgun's breakdown of the Gmail and Yahoo requirements makes the arithmetic plain, three complaints per 1,000 sends already puts you at the 0.30 percent ceiling. Sendability's operational data, drawn from more than a billion messages a month, shows the cost of living inside the warning band. Senders running between 0.10 and 0.29 percent see 12 to 18 percent of mail routed to spam, where the same senders saw 4 to 6 percent before the 2024 enforcement.

One badly targeted campaign can do that damage, and the list rules exist to stop that campaign from ever going out. The structure is a sunset policy, and the vendors who manage high-volume sending describe the same shape. Klaviyo's sunset flow guidance sends people who never opened in 30 to 45 days a single short email, gives previously engaged but lapsed subscribers a two to four message re-engagement series, and suppresses anyone who still does not respond, with full suppression after 12 to 18 months of inactivity. Mailjet's sunset policy guide sets the disengaged threshold at three to six months of silence for typical sending cadences.

Two details decide whether a sunset policy helps or hurts. A click, a reply or a purchase counts as re-engagement, while an open alone should not, which is the distinction CampaignKit's tiered sunset framework draws explicitly. And the win-back series is worth running before suppression, since the same framework puts recovery of at-risk contacts at 5 to 15 percent. If you have not emailed someone in over a year, the cleanest move is a re-opt-in, one message asking whether they want to stay on the list, with anyone who does not confirm suppressed. Salesforce's recovery guidance names exactly this move, a permission pass to engaged segments, as the way back from a damaged reputation.

The complaint side has a structural fix as well. Make unsubscribing obvious. Gmail and Yahoo require marketing messages to support one-click unsubscribe through a List-Unsubscribe header, with a visible link in the body, and requests must be honoured within two days. Most email platforms enable this in settings, and it converts a future complaint into a clean unsubscribe. When the quiet part of your list still feels like an opportunity, our guide to email re-engagement campaigns walks the full sequence, and for customers who bought once and went quiet, the customer win-back workflow covers the recovery side of the same problem.

The monthly deliverability check-up that keeps it fixed

Reputation is a trend, and trends only respond to repeated behaviour. Book 30 minutes on the first Monday of the month and run the same five checks, every one of them free.

  1. Send a mail-tester message from a real campaign template and compare the score with last month's. A drop usually means a configuration change or a new blacklist entry.
  2. Open Google Postmaster Tools and read the spam rate, the reputation rating and the authentication pass rates for the month. The spam rate belongs under 0.10 percent, and the reputation rating should sit at high or medium.
  3. In your email platform, pull complaints, bounces and unsubscribes for the past 30 days, and divide complaints by total sends. Under 0.10 percent is the target, above 0.30 percent is the emergency line.
  4. Compare your engaged segment with the total list. If the total grows while the engaged portion stays flat, the list is filling with people who will not read, and the prune needs to be harsher next month.
  5. Log the numbers in the same sheet every month. The trend line matters more than any single reading, because filters judge months, not days.

Escalate deliberately when a check comes back red. A spam rate in the 0.10 to 0.29 percent band means cut the next campaign back to engaged readers until it drops. At 0.30 percent or above, pause broadcasts and let seven clean days rebuild eligibility. If the mail-tester message never arrives, the sending server is blacklisted and the conversation belongs with your email provider, who manages the shared infrastructure. Most months the check-up takes ten minutes and changes nothing, which is exactly what a fixed routine looks like.

Frequently asked questions

How long does it take for emails to reach the inbox again after a fix?

DNS records normally propagate within 48 hours, and Google recommends allowing 48 hours after SPF and DKIM are live before you turn on DMARC. The records fix identity quickly. Reputation is slower because filters keep weeks of history, so expect two to six weeks of ramped sending to engaged readers before placement settles. Postmaster Tools usually shows the first improvement within days of your complaint rate dropping.

Do I need a new domain or a dedicated IP to recover?

Not for most senders. Domain reputation follows the From address and outweighs IP reputation at Gmail, so a dedicated IP does not reset the judgement. A new domain starts with no reputation at all and needs the same ramp. Fix the records, prune the list and rebuild engagement first. For a list that has not been emailed in a long time, a re-opt-in is the cleanest recovery move.

Will better subject lines or fancier templates stop the spam placement?

Once the sender is trusted, content rarely blocks delivery. The exceptions are all-image emails and misleading subject lines, which Omnisend's rundown of why emails land in spam lists alongside the sender-side causes. Before the sender is trusted, no subject line fixes placement, because the filter's decision happens before the words are read.

How do spam complaints happen if people opted in?

People forget they subscribed, and a brand they have not heard from in a year reads as spam in the moment. Unengaged readers are where complaints come from, which is why sunset rules matter more than most senders think. Make the unsubscribe link visible, support one-click unsubscribe, and honour requests within 48 hours.

Is cold email the same problem?

No. Cold outreach runs under different consent rules and volume norms, and providers judge it separately from permission-based marketing mail. This routine covers newsletters, nurture and lifecycle email sent to people who opted in. If your problem is cold sending, our cold email personalisation guide covers that separate workflow.

How long does a deliverability fix take to set up?

A Supernodes pilot can have the foundation live in two weeks, covering the three records, the list pruning, the re-engagement and sunset flows in the platform you already use, and the monthly check-up as a standing routine.