X analytics: the complete guide

You open X analytics and find a dozen numbers. Impressions are up, the engagement rate is down, profile visits did something odd, and someone wants a report by Friday. Which of these numbers deserves your attention? If you report to a client, start from our monthly X report template. Two numbers deserve their own guides: vanity metrics you can ignore and what profile visits tell you.

X analytics is the set of figures X gives you about your posts and your account. This guide covers where to find them, what each metric means, what the API adds, what X does not show, and how to turn it all into a monthly report built on your own baseline. Where we could not confirm a detail from X, we say so.

Where do you find X analytics

X shows analytics in a few connected places. Which views you can open depends on your account, so check your own. These are the ones X's pages describe.

Professional accounts add Professional Home analytics, according to X's help. Our guide to the X professional account covers what else changes when you switch.

What the main X analytics metrics mean

Six numbers cover most of what you need: impressions, engagements, engagement rate, profile visits, link clicks and followers. Each answers a different question, and none of them answers all of them. To keep link tags consistent across people and campaigns, use our UTM naming conventions.

These definitions come from X's own pages: its developer docs for impressions, its Export CSV Definitions for engagements and link clicks, and its Ads glossary for engagement rate. For the wider vocabulary, see our X glossary.

The one that causes most confusion is engagements. It is bigger than the likes, replies and reposts you can see under a post, because it also counts clicks on the post, on media, on your name and on your photo.

Here is how that reads on a single made up post.

Say the post shows 2,400 impressions and 72 engagements. The engagement rate is 72 ÷ 2,400 × 100, which is 3.0%. If 18 of those 72 engagements were clicks on the link, the link click rate is 18 ÷ 2,400, or 0.75%, and the link clicks are already inside the 72.

The full method, including the choice between dividing by impressions or by followers, is in our guide to calculating engagement rate on X.

Which number answers which question

• Was it shown? Impressions.
• Did people react? Engagement rate by impressions.
• Did it send traffic? Link clicks first, then sessions in your web analytics.
• Did it build the audience? Profile visits, then followers.
• Did a video hold attention? The playback milestones, which are unique users who played through 25, 50, 75 and 100 per cent. Our video on X guide covers formats.

How the Post Activity Dashboard and CSV export work

The Post Activity Dashboard lists your posts with their impressions, engagements and an engagement rate, and it can export a CSV. As X's help page reads in October 2026, it opens on the last 28 days and you can change the date range.

The Video Activity Dashboard works the same way for video, and X says both can export a file with post level data and permalinks, split into organic and promoted metrics. The 36 hours matters: X says its systems take up to that long to stabilise, so a slight fluctuation is normal.

A clean export takes six steps.

1. Pick one date range

   Choose the dates the post went out, up to 30 days at a time. For a month, export the whole month, not the last 28 days.

2. Export the CSV

   If you reply a lot, replies count towards the 3,000 post cap. X suggests splitting a long range into smaller ones when the file will not download.

3. Convert the times

   The export is in UTC, so a post you sent at 9 in the morning may show at another hour. Convert to your own time zone before you compare slots.

4. Leave out the newest posts

   Posts under 36 hours old are still settling. Better, wait until every post in the file is the same age, such as 7 days.

5. Keep organic and promoted apart

   The file splits them. Add them together only if the report covers paid and unpaid posts as one.

6. Name the file by its range

   Save it as x_posts_2026_09.csv or similar, in a folder you own. This is the copy you will trust in a year.

What the X API shows, and why the 30 day window matters

The X API lets a program read the same kinds of counts. Public metrics cover any post. Non public metrics, which include URL clicks, profile clicks and the engagements total, are only for your own posts, and X says they are only available for posts created within the last 30 days.

That window is the single most important API fact for reporting. After day 30, the private counts for that post are no longer there to read. Three reporting guides go further: measuring brand awareness, measuring community health and reporting results to executives.

X's developer docs sort the metrics like this.

The docs also describe a Get Posts Analytics endpoint, and a handful of facts shape how you use the API.

• What you send. Up to 100 post IDs and a start and end time, with results by hour, day, week or in total.
• What comes back. Fields such as impressions, engagements, likes, replies, bookmarks, detail expands, URL clicks, profile clicks, media views, follows and unfollows.
• How far back. That page states no maximum look back, and the metrics page states 30 days, so we plan for the shorter one.
• No rate field. The metrics page lists counts only. You divide them yourself, as in our engagement rate guide.
• Followers is a running total. The API gives the current followers count on a user. To see growth, save the count on the same day each month.

Reading through the API costs credits: X's docs say credits are bought in advance and deducted as requests are made.

Our posts on the X API and X API pricing explain how that works. If you would rather not build this, Kairo Post's analytics read post and account metrics from the official X API, and because X charges for reading post metrics, you choose how often they refresh and see the price first.

What X analytics does not show

X analytics does not show who saw your post, what people did on your website afterwards, or the private numbers of other accounts. Knowing the gaps stops you drawing conclusions the data cannot hold.

• People. An impression is a view, not a viewer. We found no unique viewer count for organic posts on the pages we read.
• Your website. X records the click. What happens on your site, such as how long people stay or whether they buy, is visible only to your own analytics.
• Other accounts' private numbers. Link clicks, profile clicks and engagements need user context and are for your own posts.
• Old private counts through the API. After 30 days the non public metrics for a post are gone from the API.
• A finished rate. The API lists counts and no engagement rate, so the rate in a report is always one somebody calculated.
• A reason. X shows what happened to a post, not why. The why comes from comparing posts you changed on purpose.

How often to check each number

You do not need every number every day. Match the check to how fast the number moves and what you would do about it. Most accounts need a short weekly look and one proper review a month.

Profile visits and followers deserve one extra habit. Read them together. Profile visits show how many people looked at you, and followers show how many of those stayed. If visits climb and followers do not, the posts are interesting but the profile is not finishing the job. Our profile optimisation guide covers what visitors see.

Followers also need a growth rate, not just a total. Subtract the followers at the start of the month from the followers at the end, then divide by the start. In the example later, 142 new followers on 3,120 is 4.6%. A total of 3,262 on its own cannot tell you whether that was a good month.

How to find your own baseline

Your baseline is what a normal post of yours does, measured the same way each time. Take the median engagement rate of your last 20 to 30 posts, all read at the same age. A new post is then "above", "near" or "below" normal, which a single number cannot tell you.

We give no average engagement rate for X here, on purpose. We found none from X, and none from research that discloses its method and sample.

1. List 20 to 30 recent posts

   Use the export, so the numbers share a source. Leave out giveaways and paid posts if they are not your normal output.

2. Pick one measuring age

   Read every post at the same age, such as 7 days after it went out. Which age matters less than keeping it fixed.

3. Work out each post's rate by impressions

   Engagements divided by impressions, times 100. Write the formula at the top of the sheet.

4. Take the median and the middle half

   The median is the middle value. The range from the lower to the upper quarter shows how much a normal post varies.

5. Compare, then change one thing

   A post above the middle half beat your normal. Test one change at a time, so you know what caused it.

Our free engagement rate calculator does this for a pasted list of posts. The median beats the average because one very good post drags an average up, and a median does not move.

When your baseline settles, it also gives you something to report that does not depend on anyone else's account.

How to track clicks from X to your website

To track clicks from X, put UTM tags on every link you post, then compare X's link clicks for each post with the sessions your web analytics shows for that tag. The two counts measure different things, so they will not match. Our guide to tracking clicks from X has the full method.

Google asks you to always use utm_source, utm_medium and utm_campaign. Values are case sensitive, so x and X become two sources. A tagged link for an X post looks like this:

https://example.com/autumn?utm_source=x&utm_medium=social&utm_campaign=autumn_launch&utm_content=post_01

Build the tags with the free UTM builder, which lowercases values and has presets for a post, a thread and a profile link. Use the same names every month, because a report that changes its labels cannot be compared with last month's. To compare two versions of a post without fooling yourself, see how to A/B test X posts fairly.

How to build a simple monthly X report

A good monthly report fits on one page: six numbers, one line on what changed, one line on what you will do next. Anything longer is read once. Here is a routine that takes under an hour once you have done it twice.

1. Export the month

   On the first working day of the month, export last month's posts. Wait until the newest posts are at your measuring age if you can.

2. Fill in the six numbers

   The six metrics from earlier: impressions, engagements, engagement rate, profile visits, link clicks and followers at the start and end of the month. Add posts published and the sessions from your tagged links as context.

3. Put each number next to a comparison

   Last month and your baseline. A number alone tells nobody anything.

4. Pick one post that beat your baseline and one that fell short

   Link both. Write one line each on what you think made the difference.

5. Write the decision

   One sentence: what you will keep, change or stop next month.

6. Save the report next to the export

   Same folder, same file name pattern. In six months you can read the whole story in order.

Here is what the numbers might look like. The account and every figure are made up.

The pooled rate sits above the median because a few strong posts carried the month, a pattern worth a line in the report.

If you manage accounts for others, the guide to managing X for clients covers the wider routine. If you want a regular report, Kairo Post can save a daily or weekly one as a private file you can share.

What to do when a post beats or misses your baseline

Treat one unusual post as a lead to follow, not a verdict. The first job is to rule out a boring explanation, and the second is to find what was different about the post.

The split in the second row matters most. A post can have a strong rate on few impressions, which says the idea landed with the people who saw it, or a weak rate on many impressions, which says it travelled to people who did not care. Those two cases call for different next steps, and the engagement rate alone hides the difference.

If you export a month of posts, sort the sheet by impressions, then by rate, and read the top and bottom five of each. You will usually see the same few differences repeating. That short list is worth more than any single number in the report, and our guide to how the X algorithm works covers what X has published about how posts are shown.

Mistakes that make X analytics misleading

Most bad readings of analytics come from comparing numbers that were never comparable. These are the ones we would check first.

The way you write a result matters as much as the reading. Compare these two lines for the same made up month.

The first line names the formula, the age, the baseline and the change. The second says nothing that next month's report could be checked against.

Your next step: a 30 minute analytics setup

Do this once and the monthly report gets much easier. It also gives you a record that survives the 30 day limit.

From here, our guide to growing on X and the X content strategy show how to act on what the numbers say. If you plan content around topics, content pillars for X gives you a way to see which topic earns its place.