I stopped guessing which subject line would slip past the spam filter

I write a lot of launch emails for small products. For a long time my subject-line process was: think of one line, maybe a second, and hope. Hope it is not too long for the inbox preview, hope a filter does not choke on a word, hope it gets opened. The open rate was the only feedback I ever got, and by then the send was already gone.

What SubjectLab does is narrow and deliberately boring. You paste a short offer brief — what the email is about, the offer, maybe a tone note — and it returns at least six subject-line variants. Each variant is scored on two things that are actually checkable before you send: length (too long gets truncated in most previews, and a chopped line reads as spammy) and spam-risk flags (phrases and patterns that commonly trip filters). It is decision-support. You read the scores, you pick the line that fits, you send it through your own tool. That is the whole loop, and that is the point.

The reason I built it this way instead of promising more is that the part I cannot measure is the only part that actually drives opens: your sender reputation, your list hygiene, the time of day, the subject's fit with the body. A scorer that pretended to predict your open rate would be lying to you with a decimal point. So the tool stays in its lane — it tells you which lines are long or risky, and leaves the rest to you and your ESP.

The "at least six variants" floor matters more than it sounds. When you only write one or two lines, you are anchored to the first idea and the spam-risk flag on it feels like a verdict. Six forces you to look at the pattern across options: if three of them trip the same flag, the problem is your offer phrasing, not one unlucky word. That is the real output — not a single "safe" line, but a read on where your instincts drift toward risky or bloated wording. You still choose; you just choose with the pattern in front of you instead of in your head.

What it deliberately does not do

• It is not an ESP and it does not send anything. There is no list, no send button, no open tracking. It cannot and will not tell you your open rate.
• The spam-risk flags are heuristics, not a guarantee. They flag patterns that commonly trip filters. They are not a contract with Gmail or any other provider about what will or will not land in the inbox.
• Scores are about the line, not the campaign. Reputation, list quality, and timing move open rates far more than wording does, and none of that is in scope here.
• No claim that a high-scoring variant will beat a low one for your audience. Scoring reduces obvious risk; it does not replace your judgement or a real A/B test.

SubjectLab is $9/mo, runs in the browser, no account required to try. If a flag it raises ever looks wrong to you, that is the conversation I want to have — the tool is only useful while you trust the flags.