Module 7 — Staying Out of the Spam Folder

You can write the best pitch of your life and have it never reach a human. Email providers run aggressive spam-filtering, and they err on the side of false positives — better to silently dump a borderline email than risk a user calling support about an actual scam. Industry-wide, roughly 8 to 12 percent of legitimate commercial emails in the US land in spam folders.

This module is about staying in the 90 percent.

The two ways your pitch ends up in spam

There are two paths a pitch can take into the spam folder. Each requires a different fix.

Algorithmic filtering. A spam filter scored your email above the threshold based on content, sender reputation, formatting, links, attachments, and behavioral signals. The email was tagged before anyone saw it.

Manual reporting. A human recipient marked it as spam. Each one of these damages your sender reputation more than 100 silent algorithmic flags do.

Both paths are avoidable. Let’s walk through each.

Avoiding the algorithmic filter

Send slowly, not in bursts. A new sender suddenly sending 200 emails in an hour looks like a spammer. A new sender ramping from 5 emails a day to 30 over six weeks looks like a person. If you’ve never cold-pitched from your email address before, build the volume gradually.

Plain text by default. Spam filters are deeply suspicious of HTML email — colored backgrounds, embedded images, custom fonts, fancy buttons. They’re the visual fingerprint of marketing emails, and marketing emails are spammed disproportionately. A cold pitch should look like an email a person typed. Plain text, sentences, two paragraphs, sign-off. The rare time an inline image is justified, it’s a cropped screenshot of a single specific thing — never a logo, never a header.

Avoid spam-trigger words. There’s a vocabulary that spam filters are trained to mistrust: free, guarantee, act now, limited time, congratulations, winner, click here, risk-free, no obligation, make money fast, double your, while supplies last. You probably weren’t planning to write any of these in a pitch to a Series B founder anyway, but the list is bigger than you’d guess. The agent in this course catches them automatically.

Keep links to two or fewer. Three or more outbound links in a short email is a spam signal. One link to a relevant sample is great. Two — sample plus your site — is fine. Three is risky. Five and you’re cooked.

Use descriptive URLs. Spam filters distrust trackers and shorteners. A link to bit.ly/something is much worse than a link to yoursite.com/samples/saas-blog. If your email-tracking service offers a custom-domain option (most do), use it; if it doesn’t, accept that tracking comes at a deliverability cost.

No attachments. Period. Attach nothing. If you have a portfolio, link to it. If you have a one-pager, link to it. Attachments to unsolicited emails are the single highest-confidence spam signal there is, and most antivirus systems will quarantine them on the way in. Your beautiful PDF brochure will never be opened.

Verify the email address before sending. Sending to invalid addresses produces bounces, and bounces wreck your sender reputation faster than almost anything else. Run prospect emails through a verification service (Hunter.io, NeverBounce, Bouncer) before they go out. The agent does this automatically.

Don’t send to a backlog of dead addresses. Related: if you’ve collected an email list over years, don’t blast the whole thing. A high bounce rate on the first send classifies you as a spammer for weeks afterward, even on your good addresses.

Authenticate your domain. SPF, DKIM, DMARC. Three DNS records on your sending domain. They tell receiving servers yes, this email really did come from this domain. Setting them up takes thirty minutes once. Without them, you’re sending from a domain that, as far as the receiving server can tell, anyone could be spoofing. Take the half-hour.

Avoiding the manual report

Algorithmic filtering can be tuned around. Manual spam reports cannot. Once a recipient hits Report Spam on you, your reputation with that recipient’s email provider takes a hit that recovers slowly. A handful of manual reports across recipients can make every email you send for a month land in spam.

The defenses are simple but absolute:

Personalize. This is principle #1 from module 4 and it’s also the single biggest defense against manual spam reports. Recipients almost never report a clearly-individualized email — even one they dislike. They report the ones that obviously went to a hundred other people identical-stamped.

Be CAN-SPAM compliant. The US law requires four things in any commercial email: (1) accurate from/subject, (2) identification as a solicitation if applicable, (3) a physical mailing address, (4) a clear opt-out method. None of these are hard. The agent’s signature block includes the address and a “tell me to stop and I will” line by default.

Proofread. Typos, broken sentences, weird formatting — these read as “low effort, probably automated.” Recipients report low-effort outreach more than high-effort outreach by a wide margin.

Don’t anonymize your links. If your link looks like a tracker (r.example.com/abc123) the recipient may report you on suspicion alone. Use real, descriptive URLs.

Be ethical. This isn’t only an aesthetic preference. Fake urgency, fabricated social proof, false “RE:” subject lines, pretending you’ve talked before — all of these radically increase the manual report rate. The recipient sees the trick, feels insulted, and clicks the spam button. You earned that.

A short checklist

Before any pitch goes out, the email should be:

• Plain text (no images, no fancy HTML)
• Personalized in at least two specific lines
• Under 150 words
• Linking to at most two URLs, both descriptive
• Not containing any spam-trigger words
• Going to a verified, valid address
• Signed with a real physical mailing address and an opt-out method
• Sent from a domain that has SPF, DKIM, and DMARC set up

This is six bullet points, and you can run them mentally in 20 seconds, or — as the rest of this course assumes — you can let the agent run them automatically.