Twitter Analytics on X: What It Shows and What It Doesn't

To see Twitter analytics on X, tap the analytics icon under one of your own posts on mobile, or open Analytics from the left menu on desktop, where the account dashboard reportedly needs X Premium. Twitter is now X, so "Twitter analytics" and "X Analytics" are the same tool. It shows how often your posts appeared on screen and what people did with them, for a recent window only.

This guide covers where to find it, what each metric means, what X does not keep, and how to hold your own history. X changes this tool often, so treat anything below as accurate at the time of writing and check your own account.

Where to find analytics on X

Analytics on X lives in three places: the per-post view on mobile, the Analytics page on desktop, and the Post Activity Dashboard.

Per post, on mobile. Tap the analytics icon under one of your own published posts. Sociality.io reports that post-level analytics stay available without Premium on mobile.
Account dashboard, on desktop. Choose Analytics in the left menu. Social Media Today reported on 13 June 2024 that the redesigned page was available to Premium subscribers, and Brandwatch repeated that in 2025.
Post Activity Dashboard. X's business help page on Post and Video Activity Dashboards describes dashboards that show the last 28 days by default, let you change the date range and export to CSV. The same page says these dashboards are free with an ads account at ads.x.com or by registering at analytics.x.com.

Our fuller breakdown of these screens, and where the sources disagree, is in X Analytics history: how far back, and how to keep it.

What the metrics mean

The core X metrics are impressions, engagements, engagement rate, profile visits, link clicks and video views, and each one counts something narrower than it sounds.

Metric	Definition	Where to find it
Impressions	Times a post appeared on a user's screen. Not unique: the same person viewing twice counts twice, per X's API docs.	Per-post view, dashboards
Engagements	Total times someone interacted with a post, including clicks anywhere on it, such as links, media, username, profile photo or expanding the post, per X's business help page	Per-post view, Post Activity Dashboard
Engagement rate	Engagements divided by impressions, per X's business help page	Post Activity Dashboard
Profile clicks (user profile clicks)	Clicks on the author's name, @handle or profile photo on a post, per X's business help page	Post detail, dashboards
Link clicks	Clicks on a URL in the post	Post Activity Dashboard
Video views	Views of a video, aggregated across all posts containing it, per X's API docs. Retention and completion appear in the Video Activity Dashboard.	Video Activity Dashboard

Two cautions when reading these numbers.

Engagement rate has no single formula. X divides engagements by impressions. Some published benchmarks divide by follower count instead, so a figure quoted in an article is not comparable to the one on your screen until you check which formula it used.

Impressions are not readers. Because impressions are not deduplicated and say nothing about whether the post was read, a high count tells you about exposure, not attention. Hovering over a metric in the post detail view shows X's own definition.

What X does not keep or show

X shows a rolling window rather than a permanent record, and several things never appear in it.

A single retention figure. In the sources we found, X's own page states a 28-day default view and nothing more precise. Guides report ranges from 90 days to a year, and one describes 28-day retention. They are not reconciled anywhere.
Daily change. A dashboard or export tells you where a post stands now. It does not show how its impressions grew day by day.
Older API metrics. X's API docs say the non-public, organic and promoted metrics are limited to posts from the last 30 days, which is one reason tools built on that API see a short window too.
Reach prediction. Analytics describe what happened. They do not forecast what the next post will do.

None of this means data disappears on a set day. It means you should not depend on X to hold it for you.

Exporting and keeping your own history

The export is a CSV from the Post Activity Dashboard, and the way to keep history is to export on a schedule and store every file.

According to X's business help page, each export covers up to 30 days and 3,000 posts, with timestamps in UTC.

A routine that holds up:

Export more often than your window. If an export covers 30 days, export every two to three weeks so ranges overlap.
Name each file with its export date and never edit the raw files.
Once a quarter, merge them into one sheet keyed by post link, keeping the latest row for any post that appears twice.
Note anything that affects comparison, such as a renamed metric.

The remaining gap is that each file is one snapshot. To see how a post's numbers moved over time you would have to export repeatedly and keep all of it. If you also want context on what X's ranking system rewards, read how the X algorithm works and the X ranking weights page.

Asking an AI assistant about your own post results

An AI assistant can only answer questions about your posts if it can read your results, and pasting exports into a chat is the manual version of that.

LinkIntel is an MCP server that makes it automatic. Once you connect X, it reads your own recent original posts and their metrics daily and stores dated snapshots. Your assistant, such as Claude, can then cite them when you ask which of your recent posts did best or how a topic performed. The get_owned_performance tool reads only already stored snapshots and makes no live X requests. The snapshots build up from the day you connect, so the daily history is kept for you without a calendar of exports.

What it does not do matters as much. It never publishes, never searches X and never predicts reach. It costs $39 USD a month, billed monthly in advance, with no free trial and a full refund within 7 days. For the three ways to get post data in front of Claude, including a free manual one, see how to give Claude your X post data. Install steps are in the MCP documentation.

Limits

LinkIntel covers your own account only, and it cannot restore history from before you connected.

Own account only. It reads your own recent original posts. It does not report on competitors or on accounts you do not own, and it is not a monitoring tool.
Recent original posts. Replies and reposts are not the focus, and your whole account history is not imported.
Going forward. Snapshots start when you connect X. Earlier periods are not recovered, so export from X first if you need them.
No reach forecasts. Stored results describe your past posts, not what the next one will do.

FAQ

How do I see Twitter analytics on X?

On mobile, tap the analytics icon under one of your own published posts. On desktop, the account dashboard is under Analytics in the left menu and reportedly needs X Premium, while X's Post Activity Dashboard is free with an ads account at ads.x.com or by registering at analytics.x.com.

Is Twitter analytics the same thing as X Analytics?

Yes. Twitter was renamed X, and the analytics tool people still search for as Twitter analytics is now called X Analytics. The metrics are the same kind: impressions, engagements, engagement rate, profile clicks, link clicks and video views.

How far back do X analytics go?

X's business help page describes a 28-day default view with a changeable date range. Other guides report longer ranges, and none we found states a hard retention limit, so check the date picker in your own account and export regularly.

Can I export my X analytics?

Yes, as a CSV from the Post Activity Dashboard. X's business help page reports a limit of 30 days and 3,000 posts per export, with timestamps in UTC.

Can an AI assistant read my X analytics?

Only if it has a source to read. LinkIntel is an MCP server that, once you connect X, reads your own recent original posts daily, stores dated snapshots and serves them to your assistant. It costs $39 USD a month, never publishes and never searches X.