How to spot clickbait before you waste 40 minutes on it

“Is this clickbait?” is the wrong question, and it is why people are bad at this.

Almost everything is clickbait now, in the sense that almost every title and thumbnail on the platform has been optimised to be clicked. That is what the incentive rewards. Asking whether a title is doing work to get clicked will get you a yes nearly every time, which is not a useful signal.

The question that actually discriminates is: does the video keep the promise its packaging made?

The promise includes the thumbnail

Most people evaluate the title and ignore the thumbnail, which is backwards. The thumbnail usually carries a second, louder promise that the title never makes.

A title reads “My thoughts on the new framework.” The thumbnail says “I WAS WRONG” in ninety-point type. Nothing in the title promised a reversal. Everything about the thumbnail did. If the video is a measured overview with no reversal in it, the promise was broken — and if you only audit the title, you will not be able to say where.

Read both. The thumbnail is usually the real claim.

Five patterns worth recognising

The unanswered question. The title asks something the video never answers. Frequently because answering it honestly takes ninety seconds, and ninety-second videos do not pay. The tell is a title that is a question where the answer would be short and boring.

The buried lede. The promised content is real and arrives at minute 31 of 38. Not dishonest, exactly — but the runtime was inflated around the payload. The tell is a video substantially longer than its topic warrants.

Escalated stakes. “This changes everything” over a modest incremental improvement. The facts are accurate; the framing has been inflated by an order of magnitude. This is the most common pattern and the hardest to catch, because nothing in the video is technically false.

The recycled take. The title implies novel analysis. The content restates what four other videos said last week. Common on fast-moving news topics.

Manufactured conflict. “X DESTROYS Y” describing a polite disagreement, or a “response video” to a criticism nobody made.

Checks that take ten seconds

Open the transcript. The ”…” menu under the player, then Show transcript. Ctrl-F for the key noun in the title. If the video’s central subject appears three times in a 40-minute transcript, it is not really about that.

Check the duration against the topic. Any topic has a natural length. A comparison of two phones is maybe fifteen minutes. At forty, something is padding it.

Read the top comments. The single highest-value shortcut on the platform. If a video buries its lede, the top comment is reliably a timestamp with “you’re welcome.” A comment section full of people thanking someone for the timestamp is a complete answer to whether you should watch linearly.

Look at whether the thumbnail text appears in the video. If the thumbnail says “I WAS WRONG” and nobody ever says they were wrong, you have your answer.

Judge the channel, not the video

Any single video can be a miss. The pattern across a channel is the real signal, and it is far more predictive.

Some creators wear loud packaging over consistently good work — they have learned the platform rewards it, and their videos deliver anyway. Others use loud packaging to cover thin content, every time. After five or six videos the difference is unmistakable, and it lets you skip the evaluation entirely: you already know whose titles to discount and by how much.

This is the part that is genuinely tedious to do by hand, because it requires remembering how a dozen channels performed over months. Briefed’s clickbait detector scores each video’s promise-to-delivery gap automatically, so following a few channels surfaces the pattern within a couple of weeks without you tracking anything.

And separately: was it true?

Delivering on a promise and being correct are unrelated properties. A video can keep its promise perfectly and be confidently wrong throughout, or oversell wildly and be scrupulously accurate.

If accuracy is what you care about, that is a different check — claim-by-claim verification against live sources rather than a promise-versus-delivery comparison. Worth keeping the two questions separate, because a video that passes one can fail the other badly.