Why Emails Go to Spam: Every Root Cause Explained (With Fixes)

You spent two days crafting an email campaign. The subject line is sharp. The offer is real. The design looks beautiful in preview. You hit send — and 20% of your audience never sees it. It went to spam, and you have no idea why.

This is the most frustrating experience in email marketing, and it happens to programmes of every size. The 36% of emails that land in spam folders globally (Unspam 2025) are not all from bad actors. Many are from legitimate marketers who made fixable mistakes — mistakes that look different at the surface but trace back to one of six specific root causes.

This guide names every root cause, explains the mechanism behind each one, and gives you the exact fix. Not vague best-practice advice. Specific, testable actions that address the actual reason your email went to spam.

How Spam Filters Actually Make Decisions

Most explanations of spam filtering describe a checklist: check SPF, check DKIM, check content, check reputation. That description is accurate but misleading. It implies that passing all the checks guarantees inbox placement. It doesn't.

Modern spam filters at Gmail, Microsoft, and Yahoo work as scoring systems with hundreds of weighted signals that combine into a placement decision. Authentication is a prerequisite — fail SPF or DKIM and you're rejected outright. But pass authentication and you still face the full weight of the reputation and engagement evaluation. A technically perfect email from a domain that has been generating complaint rates above 0.10% will land in spam. Every time.

"Lots of perfectly technically compliant mail lands in the spam folder — because it is easily measured as having low value to recipients. Engaging mail delivers. Unwanted mail doesn't."
— Al Iverson, Spam Resource

The implication: spam filters are not primarily filters against spam. They are filters for email that recipients want. An email that recipients consistently open, click, and never complain about earns the right to inbox placement. An email that generates complaints and disengagement loses it — regardless of how technically correct its DKIM signature is.

With that framing established, let's go through the six root causes in the order they should be investigated.

Root Cause 1: Authentication Failures

Authentication is the entry gate. Without it, nothing else matters — your email either gets rejected outright or routed directly to spam before any engagement or reputation signals are considered. Since February 2024, Gmail, Yahoo, and Microsoft (as of May 2025) all enforce authentication requirements that make failure here an immediate and severe deliverability problem.

SPF Failure

SPF (Sender Policy Framework) tells receiving servers which IP addresses are allowed to send email for your domain. If the IP that sent your email isn't in your SPF record, the result is SPF fail or SPF softfail — both of which damage your DMARC evaluation and reduce inbox placement.

The most common SPF failure mode is not a missing SPF record — it's a stale one. Your marketing team switched to a new ESP six months ago. IT forgot to update the SPF record. The new ESP's IPs aren't authorised. Every email from the new ESP fails SPF silently, and nobody notices until inbox placement metrics collapse.

How to check: Send an email to a personal Gmail account, open it, click the three-dot menu → "Show original." Look for spf=pass or spf=fail in the Authentication-Results header. Or use MXToolbox's SPF Lookup tool — it shows every IP range your SPF currently authorises.

DKIM Failure

DKIM adds a cryptographic signature to your email. If that signature doesn't verify — because the key was rotated, the DNS record was deleted, or the ESP changed configuration — authentication fails. The most dangerous DKIM failure is the one that happens mid-campaign: your first 10,000 emails go out with valid DKIM, then a key rotation propagates wrong and the next 50,000 fail.

DKIM has one additional critical requirement that most guides understate: since the MAGY (Microsoft, Apple, Google, Yahoo) sender requirements, DKIM must sign with your own domain, not your ESP's shared domain. Signing with d=sendgrid.net instead of d=yourbrand.com does not satisfy current requirements and does not build domain reputation under your name.

How to check: Look for dkim=pass header.i=@yourdomain.com in the Authentication-Results header of a delivered message. The header.i= value must show your domain, not an ESP shared domain.

DMARC Failure (and the DMARC Alignment Trap)

DMARC is where SPF and DKIM come together. A DMARC pass requires that at least one of (1) the MAIL FROM domain in SPF or (2) the DKIM signing domain is "aligned" — meaning it matches or is an organisational parent of the From: address domain. This is where complex multi-ESP setups often fail in non-obvious ways: SPF passes, DKIM passes, but neither one aligns with the From: domain — so DMARC still fails.

Real example: a company sends from From: newsletter@marketing.brand.com. Their DKIM is signed with d=esp.com (the ESP's shared domain). Their SPF MAIL FROM is bounces.brand.com. SPF and DKIM both technically pass — but neither aligns with marketing.brand.com. DMARC fails. Inbox placement suffers.

Root Cause 2: Sender Reputation Damage

Reputation is the credit score of email sending. Gmail, Microsoft, and Yahoo each maintain their own reputation assessment of your sending domain and IP addresses, updated continuously based on how recipients respond to your email. High reputation means your email gets the benefit of the doubt. Low or Bad reputation means it goes to spam even before the content is evaluated.

Reputation damage happens faster than reputation recovery. A single campaign that generates a 0.20% complaint rate at Gmail can knock your domain from High to Medium reputation in 24 hours. Recovering back to High requires weeks of clean sending at below-0.05% complaint rates — during which every campaign reaches fewer people and generates less revenue than it would have before the damage.

The Complaint Rate Threshold

Gmail publishes its threshold explicitly: keep your Gmail spam rate below 0.10% (measured in Postmaster Tools). The practical target for optimal inbox placement is below 0.05% — the 0.10% threshold is the maximum, not the goal. At 0.10%, you're already experiencing degraded inbox placement. By the time you notice lower open rates, the reputation damage has been accumulating for days.

What Actually Drives Complaint Rate Up

Complaint rates rise predictably from specific sending decisions. The most common drivers: (1) Sending to contacts who don't remember signing up — time gap between signup and first send is the biggest predictor of complaint rate on welcome campaigns. (2) Frequency increases without re-confirmation — doubling campaign cadence without asking subscribers if they still want that frequency generates complaints from previously happy subscribers. (3) Sending to lapsed contacts — anyone who hasn't engaged in 90+ days is complaint-prone regardless of whether they explicitly unsubscribed. (4) Subject line misdirection — promising something in the subject line that the email doesn't deliver creates the "betrayal" feeling that drives complaint rates up.

IP Reputation vs Domain Reputation

These are separate signals that operate differently. Domain reputation (tracked in Gmail Postmaster Tools) is tied to your sending domain and is portable — if you move to a new ESP but keep signing with your own DKIM domain, your domain reputation follows you. IP reputation is tied to the specific sending IP address and is not portable.

In 2026, domain reputation is the more important signal for Gmail inbox placement. Google has progressively de-emphasised IP reputation in favour of domain-level engagement signals. This means: (1) switching ESPs doesn't automatically fix a domain reputation problem, and (2) building domain reputation through consistent, engaged-list sending is the most durable deliverability investment available.

Root Cause 3: List Quality Problems

List quality problems are the most common root cause of spam folder problems for commercial email programmes — more common than authentication failures, more common than content issues. The reason: list quality degrades invisibly. Addresses become invalid. Domains expire. Engaged contacts become unengaged. None of these changes produce an error notification. The list looks the same in your ESP dashboard while its quality quietly deteriorates.

Hard Bounces: The Immediate Danger

A hard bounce means the email address doesn't exist — the receiving server rejected it permanently. Every hard bounce increases your bounce rate, and high bounce rates signal to ISPs that you're not maintaining your list responsibly. The risk threshold: above 0.5% hard bounce rate on any single campaign, investigate immediately. Above 1.5%, expect ISP throttling and reputation impact.

Hard bounces that are not immediately suppressed are particularly dangerous because they compound: every subsequent campaign to that address generates another bounce, each one reducing the sending IP's bounce-rate metric at the ISP level.

Spam Traps: The Invisible List Quality Problem

Spam traps are email addresses maintained by ISPs and blocklist operators specifically to catch poor-quality senders. There are two types:

Pristine traps: Addresses that have never been used by a real person. They can only appear on a list if the sender acquired them through scraping, purchasing, or harvesting — all prohibited practices. Hitting a pristine trap results in immediate blacklisting. No warnings, no grace period.

Recycled traps: Previously valid addresses that were abandoned, then repurposed as traps after a period of inactivity. These are harder to avoid because they may have been legitimate subscribers years ago. Hitting recycled traps indicates inadequate list cleaning of inactive addresses — not necessarily illegitimate acquisition, but still a serious quality signal.

You cannot see spam traps in your list — that's the point. The only defences are: (1) not using list acquisition methods that introduce traps (no purchased lists, no scraped addresses), and (2) regular email verification that identifies addresses matching known trap patterns.

The Lapsed Contact Problem

The most damaging list quality problem for otherwise well-managed programmes is the accumulation of lapsed contacts who were once legitimate subscribers but have become disengaged over time. These contacts don't bounce. They often don't unsubscribe. They sit on the active list generating zero positive engagement and occasionally generating complaints when an email finally prompts them to click "Report Spam" rather than continue ignoring the messages.

The engagement signal that lapsed contacts generate (or fail to generate) directly influences Gmail's per-user spam filter evaluation for your sending domain. A domain where 30% of the active list hasn't opened anything in six months will have systematically lower per-user engagement signals than one where 85% of the active list has clicked something in the past 60 days — and Gmail's filtering reflects that difference in per-recipient placement decisions.

Root Cause 4: Content and Structure Issues

Content issues have become less important as a primary spam filter trigger over the past decade — modern ISPs weight reputation and engagement signals far more heavily than content keyword scanning. But content still matters in specific ways, and for new sending domains without established reputation, content scoring is proportionally more significant because reputation signals haven't accumulated yet.

The Image-to-Text Ratio Problem

Emails that are primarily images — a single large banner image containing all the text and the CTA — score poorly on content quality evaluations at corporate email gateways (Proofpoint, Barracuda) and some ISP filters. The reason: early spam campaigns frequently embedded all content in images to evade text-based content scanning. This pattern is recognised by content scoring systems and treated as a moderate negative signal.

The practical guideline: maintain a text-to-image ratio of at least 60:40 (60% text, 40% images or less). Ensure all meaningful content — the offer, the CTA, the key information — is present as HTML text, not embedded in images. Images should illustrate the text, not replace it.

Spam Trigger Words (The Nuanced Reality)

The "spam trigger word" concept is frequently oversimplified. Individual words like "free," "sale," or "click here" do not individually cause spam filtering in 2026 — ISPs are far more sophisticated than keyword matching. What matters is the combination of content signals in context: a high-urgency subject line combined with an email body full of promotional superlatives combined with a sending domain with Medium reputation combined with a list that has 15% inactive contacts — that combination pushes the content scoring significantly higher than any individual element would.

The practical test: run your email through mail-tester.com before sending. A score below 7/10 indicates content scoring issues worth addressing. Scores above 8/10 are generally safe from content-triggered spam filtering.

HTML Code Quality

Broken HTML — unclosed tags, malformed structure, inline style conflicts — creates content scoring signals that some filters interpret negatively. More practically, broken HTML causes inconsistent rendering across email clients, which degrades the recipient experience and reduces engagement. Lower engagement means fewer positive reputation signals, which means worse inbox placement over time.

The most common HTML quality issue in commercial email: table-based layouts that were designed for one email client but weren't tested in others. The Gmail web client renders HTML slightly differently than Apple Mail, which renders differently than Outlook on Windows. Test in all three before deployment.

Link Quality and Tracking Domain Reputation

Every link in your email — including the click-tracking redirect URL — is evaluated by ISP spam filters. If your click-tracking domain (the domain that click-tracking URLs resolve from before redirecting to the destination) appears on a URL blocklist, all emails containing that tracking domain are scored more aggressively. Many ESPs use shared tracking domains that can become blocklisted due to other customers' behaviour — sending email with a shared tracking domain means inheriting other senders' URL reputation risks.

Fix: Configure a branded click-tracking domain on your ESP (clicks.yourdomain.com). Most ESPs support this through their Domain Authentication or Sending Domains settings. This keeps your link reputation under your control rather than dependent on the ESP's shared domain quality.

Root Cause 5: Infrastructure Misconfigurations

Infrastructure problems are the most technically obscure root cause — and the most commonly missed because marketers don't have visibility into the server configuration layer. Google's 2025 DMARC rejection data analysis revealed that a surprising fraction of email rejections are caused by infrastructure configuration problems, particularly PTR record failures, not by content or reputation issues.

Missing or Broken PTR Records

Every sending IP address must have a PTR (reverse DNS) record that maps the IP to a hostname, and that hostname must resolve back to the same IP in forward DNS (FCrDNS — Forward-Confirmed reverse DNS). Microsoft enforces this strictly; Gmail has tightened enforcement significantly since 2025.

PTR records are set by the IP hosting provider, not the domain owner — which means they often fall through the cracks. A company switches from one data centre to another. The old IP's PTR record is set; the new IP's PTR record is never configured. Every SMTP connection from the new IP is evaluated more aggressively, and some ISPs reject it outright.

How to check:

# Check PTR record for your sending IP
dig -x YOUR_SENDING_IP

# Verify FCrDNS (hostname must resolve back to same IP)
dig A hostname-from-PTR-record

# Both must resolve correctly. Any mismatch = infrastructure problem.

Volume Spikes (The Pattern Recognition Problem)

ISPs track the historical sending volume pattern for each IP and domain. A sender that typically delivers 5,000 emails per day who suddenly sends 200,000 in a single hour triggers volume anomaly signals that increase spam filter scrutiny — not because the content changed, but because the pattern is outside what ISPs expect from a normal sender.

Volume spikes are particularly dangerous in two scenarios: first deployment on a new infrastructure (no volume history to establish expectations), and seasonal peaks (Black Friday campaigns where 10x normal volume lands on ISPs at the same time from the same domain). The fix in both cases is the same: gradual volume ramp that introduces higher volume incrementally rather than abruptly.

Shared IP Neighbour Problems

If your email goes through an ESP's shared IP pools, you share the IP's reputation with every other sender using that pool. A large sender on the same shared pool who generates a complaint spike — entirely outside your control — can cause the shared IP's reputation to decline, affecting inbox placement for all senders on that pool simultaneously.

The protection: use a dedicated IP address once you reach sufficient volume (typically 100,000+ monthly sends) to justify the dedicated IP's setup cost and warmup overhead. Dedicated IPs eliminate shared pool risk entirely — the IP's reputation is a direct reflection of only your sending behaviour.

Root Cause 6: Poor Engagement Signals

This is the root cause that explains why technically perfect email programmes sometimes still struggle with inbox placement — and why the fix is not technical. Engagement signals are the inputs to the reputation models that determine inbox vs spam placement at every major ISP. Authentication passes, reputation history is clean, list is verified — but if the recipients consistently don't open or click the email, Gmail's per-user personalised filtering progressively de-prioritises the sender for each disengaged recipient.

The Gmail Per-User Engagement Model

Gmail doesn't just evaluate domain-level reputation globally. It applies per-user engagement history to individual placement decisions. A recipient who has opened every email from your domain for two years will see your next email in their primary inbox even if their neighbour (who subscribed the same week and never opened anything) sees the same email in spam. The per-user model means that engagement quality within your list is as important as the aggregate reputation metrics.

The practical consequence: an email programme with 50% genuinely engaged subscribers and 50% completely disengaged subscribers will systematically underperform one with 85% genuinely engaged subscribers — because the disengaged half generates weak or negative per-user signals that drag down the domain's overall engagement profile at Gmail.

Why Open Rate Is No Longer the Right Signal to Optimise

Apple Mail Privacy Protection (MPP, active since iOS 15 in 2021) pre-fetches all email images — including tracking pixels — for Apple Mail accounts, generating open events regardless of whether the recipient actually viewed the email. Gmail's Gemini AI does the same for summary generation. As of 2026, an estimated 35-50% of "opens" in email analytics dashboards are machine-generated, not human.

Optimising for open rate in this environment — using open rate data for suppression decisions, A/B test evaluation, or engagement scoring — produces systematically incorrect conclusions. A subscriber with 20 "opens" in the past 30 days and zero clicks may be a completely disengaged Apple Mail user whose MPP auto-generates opens. A subscriber with 2 genuine opens and 2 clicks is far more engaged and valuable.

The engagement signals that remain reliable in 2026: click rate (machines don't generate clicks), reply rate (replies require human action), conversion events (purchases, sign-ups), and complaint rate (spam button presses are human actions). Build your engagement model and suppression logic on these signals, not on open rate.

How to Diagnose Which Root Cause Is Yours

Most email programmes experiencing spam folder problems have a primary root cause — the one that's doing most of the damage — and often one or two contributing factors. Identifying the primary root cause determines the fix. Here's the diagnostic sequence:

The Fix Priority Order

When multiple root causes are present, fix them in this order:

1. Authentication — always first. Nothing else can be accurately diagnosed until authentication is working correctly. Authentication failures add noise to all other diagnostic signals and must be resolved first.

2. Infrastructure — before reputation recovery. A broken PTR record or shared IP blocklist listing will undermine any reputation recovery effort. Fix infrastructure before expecting reputation to improve.

3. List quality — simultaneously with reputation recovery. Cleaning the list (removing hard bounces, suppressing lapsed contacts, verifying addresses) simultaneously reduces complaint rate and bounce rate — the primary inputs to reputation recovery. List quality and reputation recovery must happen together.

4. Engagement strategy — after quality is clean. Once the list contains only genuinely interested contacts with correct addresses, the engagement-focused improvements (better content, better segmentation, better cadence) have a clean surface to work on. These improvements are irrelevant if the list is still contaminated with disengaged or invalid addresses.

5. Content — after everything else. Content improvements are the last priority because content issues rarely cause spam folder problems alone in 2026. They can contribute to higher complaint rates from disappointed recipients, but the fix is not primarily about word choice — it's about sending relevant, wanted content to the right audience. That's fundamentally an engagement and list quality problem, not a content problem.

Spam folder problems are solvable. Every one of the six root causes documented here has a specific, testable fix. The diagnostic sequence above takes less than an hour to complete and identifies the primary cause with high confidence. Once identified, the fix is straightforward — time-consuming for reputation recovery, but not mysterious. Run the diagnosis; apply the fix in priority order; and systematically measure recovery through the metrics that matter: Gmail spam rate in Postmaster Tools, hard bounce rate per campaign, and click rate as the reliable engagement signal in the MPP/Gemini era. The inbox is achievable. The path is documented. The only remaining question is whether the programme will execute it.