X RadarX Radar All tools
FreeTracking algo xai-2026-07-04

Free X post analyzer check reach before you post.

Paste a draft and see what X's ranking algorithm does to its reach - link demotion, hashtag spam thresholds, dwell-time length bands, media bonuses and author-diversity decay.

6

structural signals checked

0

bytes leave your browser

No

sign-up required

0 chars · 0 links · 0 hashtags
Attached media
Position in thread
1

1 = standalone or first post

Your post # today
1

How many you've already posted

Your structural reach breakdown appears here as you type.

What this tool actually measures

X open-sourced its recommendation algorithm, which means the mechanics that decide your reach are readable rather than guessable. This analyzer models the structural half of that: the signals that can be read directly off a draft without knowing anything about the writing.

That distinction matters, and it is why this tool is free. Structural signals are deterministic - a link either is or isn't there, and the demotion that follows is arithmetic. Judging whether a sentence earns a reply needs a model, costs money per call, and lives in the full scorer.

The link penalty

The single largest structural penalty most drafts carry. The out-of-network scorer demotes posts containing external links, because a link off X costs the platform the session it was about to get. We model it at 0.7× reach. The fix is well known and genuinely works: keep the post self-contained and put the link in the first reply, so the post earns the distribution and the reply carries the click.

The hashtag threshold

3 or more hashtags reads as spam to the Grox classifier, which screens a post before ranking even begins - so this is a penalty applied upstream of everything else on this page. Zero or one is safe. The deeper point is that hashtags stopped being a discovery mechanism on X some time ago; search and the follow graph do that work now, and the characters are better spent on the hook.

Length and dwell time

Dwell time is one of the strongest signals in the ranker, and post length is the main lever you have over it. Two bands work: 150-280 characters reads fully in the feed while still registering as a read, and 800-1200 characters is long enough that finishing it is itself a strong completion signal. The region between them is the trap - long enough to get truncated, not long enough to pay that back.

Native media

Native media lifts dwell and unlocks engagement heads that text cannot reach. Images model at 1.15× and native video at 1.25×. "Native" is load-bearing: a YouTube link is an external link and takes the link penalty instead of the video bonus.

Author-diversity decay

The one mechanic here published as an exact formula rather than an estimate. Every additional post you place in the same ranking window is attenuated toward a floor, so your second post of the hour does not add to your first - it competes with it. This is why posting more is the most common piece of advice that quietly makes things worse, and why spacing beats volume.

What it does not measure

Worth being blunt about, because most tools in this category are not. A post can score perfectly here and still go nowhere. Structure sets a ceiling on reach; it does not create demand. Nothing on this page evaluates whether your opening line earns attention, whether the claim is worth replying to, or whether a non-follower would follow you after reading it - and those decide most of the outcome.

The numbers themselves are labelled honestly too. Of the 31 signals X Radar models, 27 are confirmed mechanics traceable to a named file in the published source; the rest are community estimates that fit the code but are not themselves published. The current build is pinned to xai-2026-07-04. There is a fuller account of what we do and don't claim to know on the about page, and the source is at xai-org/x-algorithm.

Frequently asked questions

Does this tool send my draft anywhere?

No. The analysis runs entirely in your browser using the same open-source multipliers the app is built on. Your text is never uploaded, logged or stored - you can watch the network tab stay empty while you type.

Do links really reduce reach on X?

The out-of-network scorer demotes posts carrying links off the platform, because an external link costs X the session. We model that as roughly a 0.7× reach multiplier. The standard workaround is to keep the post self-contained and put the link in the first reply.

How many hashtags is too many on X?

3 or more reads as a spam signal to the classifier that screens posts before they are ranked. Zero or one is safe, but hashtags no longer drive discovery on X - search and the follow graph do that job now.

What is the ideal length for an X post?

Two bands work. 150-280 characters reads fully in-feed while still registering dwell time, and 800-1200 characters is long enough that finishing it is itself a strong signal. The gap between them is the weak spot: truncated without the depth to justify it.

Why does my second post of the day get less reach?

That is author-diversity decay, and it is one of the few mechanics published in full. Each additional post you place in the same ranking window is attenuated toward a floor, so consecutive posts compete with each other instead of adding up. Spacing posts out is strictly better than batching them.

How is this different from the full X Radar scorer?

This tool reads the structure of a post - length, links, hashtags, media, position. Those are deterministic, so they need no model. The full scorer additionally estimates a probability for every engagement head to judge whether the writing itself earns a reply, a bookmark or a follow, which is where most of the reach actually comes from.

Keep reading

Score the writing, not just the structure

X Radar runs your draft through all 31 ranking signals, estimates a probability for each engagement head, and shows the arithmetic - then rewrites the draft in your own cloned voice.

Sign in with X

Read-only access · revoke anytime · free tier included