promise board / headline 3 · free workflow
Compare Headlines on Promise and Proof, Not on Punch — Free
Score each headline on what it promises and whether the piece actually delivers it, and the board ranks them so the choice is made on proof rather than on which one sounds best in a meeting.
What this replaces
Headline testing picks the one with the most punch, which is how a piece ends up promising something it does not deliver. This scores promise and proof at equal weight, so a headline your article cannot support sorts below one it can.
Tool note. This runs entirely in your browser — no account, no upload, no paid API key. Google NotebookLM is named as the free source-grounded reading path for the material you paste; its published free-tier terms were checked on 2026-09-13 and can change.
The four passes
Collect the smallest useful input
One headline per line with a promise score, a proof score, and its character count.
Let the board rank them
Each gets a weighted score with the three numbers behind it.
Check the proof score honestly
If the piece does not contain it, the score is not a nine.
Pick the top and check the length fits
A long headline truncates in most feeds.
Try it on the promise board / headline 3
The board below runs in your browser. It opens with a worked sample so the result is visible before you paste anything; switch scenarios, edit the text, or reset to the sample.
running the sample…
first-load · sample values are illustrative, not verified facts
Everything you paste stays in this browser tab. Nothing is uploaded, there is no account, and no request leaves the page to generate the result; clearing site data removes anything saved locally.
What breaks, and what to do
Proof scores guessed optimistically
Score what the draft contains, not what you intend to add.
Character counts omitted
Read as zero; count them or the length penalty is meaningless.
Clickbait that technically delivers
High on both and still bad; the score cannot measure tone.
Go deeper — Promise Board
The free board above is the whole loop. The paid tier keeps the arranged output so you do not rebuild it.
- Printable board with the promise and proof columns for an editorial meeting — estimate: 11 minutes saved per piece, basis: 4 headlines re-scored at ~2.75 minutes each
- Saved local history across pieces — estimate: 7 minutes saved per piece, basis: re-entering 4 headlines at ~105 seconds each
$4 one-time