Why Cold Emails Get Marked as Spam (2026)

Most advice on emailing professors focuses entirely on what to write. That's the right focus for a single email that's guaranteed to land in the inbox, but it skips a separate problem entirely: an email that never gets delivered at all doesn't fail because of bad writing, it fails before the professor ever sees it. Understanding the actual, technical reasons emails get filtered as spam, separate from the reasons they get ignored once delivered, matters more the more professors you're planning to reach out to.

Two completely different failures, often confused

"My emails aren't working" usually means one of two very different things, and the fix for each is different. The first is a deliverability failure: the email gets routed to spam, promotions, or blocked outright, and the professor never sees it, no matter how good the writing is. The second is a relevance failure: the email arrives and gets read, but doesn't get a reply, because it doesn't say anything specific enough to act on. Our guide to the exact structure that gets replies covers the second problem in depth. This post is about the first one, the technical and behavioral reasons a message doesn't even arrive.

The technical layer: authentication records

Every major email provider checks a sending domain's authentication records before deciding whether a message is likely legitimate. Three records matter most:

• SPF (Sender Policy Framework) lists which mail servers are authorized to send email on behalf of a domain. Without it, or with it misconfigured, receiving mail servers can't verify the message actually came from where it claims to, and many spam filters treat that failure as a strong signal to route the message to spam.
• DKIM (DomainKeys Identified Mail) attaches a cryptographic signature to outgoing mail that lets the receiving server confirm the message wasn't altered in transit and genuinely originated from the claimed domain.
• DMARC (Domain-based Message Authentication, Reporting and Conformance) tells receiving servers what to do when SPF or DKIM checks fail, quarantine the message, reject it, or let it through anyway, and gives the domain owner reporting on authentication failures.

These three records are usually already properly configured on university email domains by the institution's own IT department. That's part of why sending from a university address, your own current one, tends to carry a mild deliverability advantage over a personal account, the underlying domain reputation and authentication setup are typically already solid.

The behavioral layer: what your sending pattern looks like

Spam filtering doesn't only look at one message in isolation, it looks at patterns of sending behavior over time. A few behavioral signals matter specifically for anyone reaching out to many professors:

• Sending volume and speed. A burst of many similar emails sent in a short window from the same account, especially to addresses that have never interacted with that account before, starts to resemble the bulk sending pattern spam filters are specifically built to catch. Spacing outreach out, rather than sending dozens of emails back to back in one sitting, reduces this risk.
• Near-identical content sent repeatedly. Templated emails with only a name and university swapped in are both a deliverability risk at volume and, separately, the exact thing that gets an email ignored even when it does land in the inbox. Personalization solves both problems at once.
• Bounce rate. Sending to addresses that no longer exist or were guessed incorrectly generates bounces, and a sending account that racks up too many bounces relative to successful deliveries develops a poor sending reputation that can affect deliverability for every subsequent email, not just the ones that bounced. Verifying an email address before sending matters more than it seems.
• New account with no sending history. A freshly created email account with no prior sending history and no "warm-up" period is treated with more suspicion by receiving servers than an account with an established pattern of normal, low-volume sending over time.

The content layer: what's actually in the message

No single word in a subject line or body guarantees a spam folder placement on its own, modern filters weigh many signals together rather than blocking on one trigger word. That said, certain patterns genuinely correlate with spam, and are worth avoiding for credibility reasons even where they wouldn't trip a filter:

• Subject lines that mimic an existing conversation that never happened, "Re:" or "Fwd:" with no real prior thread, are one of the more reliable spam signals filters look for, separate from how badly it also reads to a human recipient who never emailed you first.
• Financial language, urgency framing, and guarantee-style claims ("free," "guaranteed," "act now") are common trigger patterns, and they're also simply out of place in an academic cold email, so there's no real reason to use them regardless of deliverability.
• Excessive links, especially shortened or unfamiliar-looking ones, can raise suspicion in a first message from an unknown sender. A single relevant link, to your website or LinkedIn, is normal, a message stacked with several is not.
• Content that reads as obviously AI-generated and generic tends to get flagged by spam systems that specifically weight for template-like phrasing, on top of being the same thing that gets a delivered email ignored by the professor reading it. See our comparison of generic vs. personalized AI-drafted emails for what the actual difference in outcomes looks like.

Attachments: a smaller factor than people assume

A single, reasonably sized, clearly named attachment, a CV as "FirstName_LastName_CV.pdf" rather than a generic or unnamed file, is unlikely to be the deciding factor in whether one individual email reaches an inbox. The risk from attachments scales mainly with bulk sending volume and with unusually large file sizes or unusual file types, not with the simple presence of a normal PDF attached to a normal message. If you're unsure whether to attach a CV to a first email at all, that's a separate, content-strategy question rather than a spam-avoidance one.

What to do if you suspect a specific email already landed in spam

There's no reliable way to know for certain, from the sender's side, whether a specific message you sent landed in someone's spam folder rather than simply being ignored once delivered, email systems don't give outside senders that visibility. A few practical signals can still point you in the right direction. If a follow-up sent from the same account to the same person also gets no response, and you know from other signals (a reply to a different professor at the same university, for instance) that your account can reach that domain successfully, the problem is more likely to be content and relevance than deliverability. If you're sending from a brand-new account, sending in a tight burst, or you've had noticeably lower response rates than expected across many recipients at once, a deliverability issue becomes more plausible and worth addressing structurally, by slowing your sending pace and warming up the account with normal, low-volume use, rather than just rewriting the email itself.

Why this matters more the more professors you're reaching out to

A single email to a single professor carries relatively low deliverability risk on its own. The risk compounds specifically when an applicant is reasonably trying to reach a real number of professors, commonly recommended somewhere in the range our guide to how many professors to email covers, since that volume, sent carelessly from one new account in one sitting, starts to resemble exactly the bulk sending pattern spam filters exist to catch. Spacing outreach out over days rather than hours, and treating each email as genuinely individual rather than a copy-paste job, addresses both the deliverability risk and the reply-rate problem at the same time, because the underlying fix, real, specific personalization sent at a human pace, is the same for both.

A practical checklist

• Send from an account with an established history, ideally a university address, rather than a brand-new personal account with no prior sending activity.
• Space outreach out over days rather than sending a large batch of near-identical emails in one sitting from the same account.
• Verify an email address is correct and current before sending, rather than guessing at a likely format, to keep your bounce rate low.
• Avoid false "Re:" or "Fwd:" subject lines, urgency language, and financial or guarantee-style phrasing, both for deliverability and because none of it belongs in a genuine academic email anyway.
• Write each email as if only one person will ever read it, since specific, personalized content is both harder for spam filters to pattern-match as bulk mail and far more likely to actually get a reply once it lands.

Personalization is the one fix that solves both the deliverability problem and the reply-rate problem at once, and it's also the part that's genuinely hard to do well at scale by hand, reading a professor's actual recent paper closely enough to reference it specifically takes real time multiplied across dozens of professors. GradScoutFunding searches professors by field and country, confirms they're actively publishing, and drafts a personalized first email grounded in their real recent paper for you to review and send yourself, it never sends anything automatically. The free tier gives 2 credits with no card required, and paid credit packs are one-time purchases that never expire.

Common questions