Tweet Tester

The tester is a composer, a simulated Twitter feed and a score panel on one page. You write the draft you were about to send, the feed shows it the way a reader meets it, and the panel puts a number on it and explains the number. Nothing is posted anywhere, nothing is saved to an account, and the other posts in the column are invented.

Type or paste your draft into the composer at the top of the page, up to 280 characters, then press Enter or click Post. The draft drops into the feed as the newest post and the counters start moving. Edit it, run it again, and compare the two scores.

The panel next to the post gives the draft a score from 0 to 100 and three multiples: likes, replies and reposts. Under the number sit two short lists, what helped and what hurt, written in plain English rather than jargon. Read the lists first. They tell you which sentence to change, while the number only tells you whether the change worked.

The photo button in the composer takes an image or a video, and the follower field in the widget header sets the account size the counters are scaled to. Set it to your real number so the estimates land near your usual range. Your file stays on your device; only the fact that a post carries media is sent along with the text.

The feed is a one-for-one rendering of the Twitter home timeline. Line breaks stay where you put them, and an attached image or video renders in the same box a reader sees it in. The column runs light or dark from the toggle in its header, and your draft sits between other posts, which is the honest way to judge a first line.

If you want a picture of a post instead, for a slide or an article, use the tweet to image tool or the fake tweet maker. If you only need to check the draft against the 280 limit, the Twitter character counter is quicker.

The score is one number for one draft, and it is a comparison rather than a forecast. It answers a narrow question: does this read better or worse than a normal post from an account like yours? It cannot know when you will post, who will see it first, or what else will be in the feed that hour.

Three hundred likes is a great day for a 2,000-follower account and a quiet one at 500,000, so a fixed target would be meaningless. The tester works out what a normal post from an account your size tends to do, then reads your draft against that. A 70 means the draft looks clearly better than your usual; a 40 means it looks weaker. To see your own averages in real numbers, the Twitter metrics calculator works them out from your recent posts.

Every score comes with reasons. The helped list names the things working for the draft: a result of your own, a clear opening line, something a reader can answer. The hurt list names what is costing it, usually an ask, a link that sends people away, or phrasing that reads assembled rather than written. Up to three items each, in the order that matters most.

Likes track how easy the post is to agree with. Replies track whether there is an obvious way in. Reposts ask a reader to put your words in front of their own audience, which depends on who you are to them, so that counter carries a low confidence tag. Read it as a range, not as a figure.

None of this is a formula, and none of it beats having something to say. These are the patterns that show up in the fitted data when the same idea is written two ways and one version travels further. They are tendencies across many accounts, not rules for yours, so treat them as a checklist to argue with rather than a recipe to follow.

A result you got, a credential you hold, or something you watched happen first hand tends to read better than the same advice with nobody behind it. Numbers help, because they are checkable. So does admitting the part that did not work. Posts worth passing to a group chat are usually the ones carrying something only you could report.

Attaching an image or a video moves the counters, in the fitted data more for some kinds of post than for others. The tester reads only whether the file is an image or a video, so it tells you the direction, not whether your particular screenshot is a good one.

Asking people to comment, drop a link or follow back does not lift the score. When a draft leans on an ask, the panel says so in a heads-up line above the reasons. Those are warnings about how the post reads, not deductions, and the rest of the draft is read as normal.