← All topicsDeliverability Overview › Mailbox Provider Filtering Logic
Accuracy / correctness · 2
hedge checkScript correctly hedges proprietary knowledge ('nobody outside Google has the secret recipe', 'what's publicly known'), correctly separates auth from placement (Beat 3), and correctly rejects single-signal framing (Beat 1 gauge, Beat 5 myth-fact). No single-signal-tanks drift. Article's own line 'Missing authentication is an instant red flag' is softened to 'a red flag' in script, which is safer and accurate. No overstatement introduced.
claim checkScript says per-recipient engagement can send you to spam 'even while your overall reputation stays fine' and Gmail 'leans harder than most' on post-delivery engagement. Both are directly supported by the article ('even if your overall reputation is fine'; 'Gmail weighs engagement more heavily than most'). Kept as a hedged generalization ('leans harder than most'), not an absolute. Accurate.
Would elevate the video · 2
exampleThe article's concrete threshold detail from aiSummary ('complaint rate must stay under the 0.1 percent threshold Google publishes in its Sender Guidelines') is a real, sourced number the script omits. Worth adding as one hedged line in Beat 4 ('Google publishes a bulk-sender complaint threshold around 0.1 percent'), because it's the one concrete, citable Gmail figure and gives the engagement beat teeth. Frame as Google's published guideline, not a filter secret.
substanceArticle's 'global sender patterns' point (if thousands of Gmail users suddenly report your domain, the filter adjusts in real time; sudden volume spikes get throttled while Gmail evaluates). The script folds engagement in but drops the real-time/global and volume-spike angle. A half-line in Beat 4 or 5 would round out 'the filter is adaptive' without a new beat.
Considered, left out · 2
detailArticle's Bayesian vs neural-network history ('the old Bayesian approach', 'neural networks understand semantic meaning'). Beat 5 keeps the plain-English payoff ('reads for meaning, not just spam words') and rightly drops the ML terminology, which belongs in 002.007.005 (Do filters use AI). Correct cut, avoids jargon cram.
ctaArticle's list-hygiene fix detail (segment out non-openers, send only to engaged subscribers for a few weeks to rebuild) is compressed to one line in NEXT. Fuller remediation belongs in a metrics/list-hygiene video; keeping it short here is right for a 'how does it work' question.
How Gmail's spam filters actually decide where you land
Question: 002.007.001 · How do Gmail's spam filters work (generally)? · ~4:00 · single-question video
COLD OPEN
Same email. Two very different inboxes.

Two ships send the exact same newsletter. One lands in the inbox, one drops straight to spam. Same words, same links. So what is Gmail actually looking at? Here's what's publicly known about how it decides.

⬡ talking-stat, big "billions/day", then four pills fade in: Reputation · Authentication · Engagement · Content
BEAT 1, the one-line answer

Nobody outside Google has the secret recipe, but Google is pretty open about the ingredients. Gmail blends four things: your sending reputation, whether you prove who you are, how people actually treat your mail, and what the message looks like. Then a machine-learning model weighs them all together. There's no single "gotcha" switch. It scores the whole picture.

⬡ gauge-meter, needle settles based on a blend of inputs, not one lever
BEAT 2, reputation is your track record

First ingredient: reputation. Think of it as your ship's logbook. How many people engaged with your past sends? How many hit report spam? A good logbook buys you leeway. And here's the catch, reputation builds slowly over months, but it can drop fast after one bad campaign. You can watch your own Gmail reputation in Google Postmaster Tools once you're sending enough volume.

⬡ timeline, reputation line climbs slowly, then dips sharply after one bad send
BEAT 3, authentication proves you're really you

Second: authentication. SPF, DKIM, and DMARC are the flags that prove the message really came from your domain and not some impersonator flying your colors. Missing them is a red flag to Gmail. One thing to be clear about, passing authentication proves your identity, it does not book you a seat in the inbox. It's the ticket to be considered, not the guaranteed berth.

⬡ auth-flow, SPF, DKIM, DMARC each stamp PASS, then a note: "identity checked, placement still separate"
SUBSCRIBE

If this is starting to make Gmail feel less like a black box, subscribe. We're working through the whole deliverability playbook, one question at a time.

BEAT 4, engagement is Gmail's favorite signal

Third, and this is where Gmail leans harder than most: engagement. Gmail watches what each recipient does after delivery. Opens, clicks, replies, stars, all good. Deleting without opening, or worse, hitting report spam, all bad. And it's per-person. If Captain Kraken never opens you, your next email to Kraken is likelier to slide to spam, even while your overall reputation stays fine. Reply and star, and you're golden with that reader.

⬡ split-compare, LEFT reader opens and replies (mail rises to inbox) vs RIGHT reader deletes unopened (mail drifts to spam)
BEAT 5, content, and why you can't just game it

Fourth: content. Gmail scans your links, your HTML, and your wording. But this isn't the old days of hunting for the word "free." Modern Gmail reads for meaning, so a message that simply reads spammy can get caught even with clean vocabulary. And you can't win on one ingredient alone. A cold blast with great open rates but no authentication still gets filtered. A perfectly authenticated newsletter that ninety percent of people ignore will still drift to spam over time. It's the whole picture, together.

⬡ myth-fact, MYTH "avoid the spam words and you're safe" vs FACT "Gmail scores reputation + auth + engagement + content together"
TAKEAWAY
⬡ title-card

So Gmail isn't one rule, it's a blend that leans hard on how real people treat your mail. Send things your readers actually want, prove who you are, and keep your list healthy. That's the whole game.

NEXT / SUBSCRIBE

Debugging a Gmail drop? Start with authentication, our free SPF checker takes thirty seconds. Then look at engagement, an open rate under ten percent is usually a list problem, not a filter problem. Next video: how Outlook does it differently.

⬡ end-card, Subscribe + Next: "How Outlook's spam filters work" (002.007.002)
DESCRIPTION

How do Gmail's spam filters work? Nobody outside Google has the exact algorithm, but the ingredients are public: sender reputation (your track record, visible in Google Postmaster Tools), authentication (SPF, DKIM, DMARC proving you're really you), per-recipient engagement (Gmail leans on this harder than most providers), and content read for meaning by machine learning. No single signal tanks you, Gmail scores the whole picture together. Plain-English, no jargon left unexplained.

0:00 Same email, two inboxes

0:20 The four ingredients

0:50 Reputation, your track record

1:30 Authentication, proving it's you

2:10 Engagement, Gmail's favorite signal

2:55 Content, and why you can't game it

3:40 Takeaway and next steps

Check your authentication free → reviewmyemails.com/tools/spf-checker

Next: How Outlook's spam filters work → [link 002.007.002]

Full written guide → reviewmyemails.com/emailalmanac

#email #deliverability #gmail

CONNECTIONS
• next: 002.007.002 How Outlook's spam filters work
• related: 002.007.004 What types of filtering do mailbox providers use · 002.007.005 Do spam filters use AI
• prerequisite: 002.007.004 Types of filtering (the layers underneath this)
• vocab: sender reputation, SPF, DKIM, DMARC, engagement signals, Google Postmaster Tools