How to Read Your X Analytics: 5 Charts That Actually Matter

Open an X Analytics export for the first time and you get a spreadsheet with a few dozen columns and no opinion about which ones matter. Most of them don't, at least not at first. Five charts cover nearly everything a person or a small team needs to know: reach over time, engagement rate, follower change, performance by post type, and posting time. Make those, look at them once a month, and ignore the rest until you have a question they can't answer.

Before the charts, a quick word on the data, because it decides what you can plot.

What's in the export

X's analytics has two exports that matter here. The overview export has one row per day, with account-wide totals such as impressions, engagements, profile visits, and new follows (sometimes unfollows too). The content export has one row per post: a timestamp, the text or a link to it, impressions, engagements and a breakdown of the engagement types. Both come from the desktop version of the analytics section.

Access is the annoying part. The full dashboard and its export have generally been tied to a paid Premium subscription, while free accounts see numbers on individual posts only, and X has redrawn that line more than once. Column names drift between versions as well, so check the header row before reusing last quarter's spreadsheet formulas. Also check the time zone of the timestamps. If your morning posts show up in the afternoon, the export is in UTC.

If you need more history, or several accounts, the official X API is the other legitimate source. Don't try to fill gaps by scraping; it breaks X's rules and doesn't give you impressions anyway.

Now the charts.

1. Daily impressions with a rolling average

Start with reach. Plot daily impressions from the overview export as a thin line, then add a 7-day rolling average as a heavier line on top. The daily line will zigzag, partly because weekdays and weekends differ and partly because posts vary. The average shows whether the account is actually reaching more people this month than last.

Annotate the spikes with what caused them. Something like "quoted by a big account" or "conference thread" is the most useful text on the chart, because it tells you what to try again. Resist the urge to read meaning into every dip. A quiet Saturday is just a quiet Saturday.

2. Engagement rate, not engagement count

Raw engagements rise and fall with impressions. A post seen by 30,000 people will collect more likes than one seen by 3,000, even if the second one was better received. Engagement rate (engagements divided by impressions) adjusts for that.

Plot the median engagement rate per week from the content export. Use the median, not the average. A single viral post can double a week's average and make the account look healthier than it is. Set a floor too: posts with only a few dozen impressions produce extreme rates that mean nothing, so leave them out of the rate.

Read the chart against your own history, not someone else's benchmark. If your weekly median has sat between 1.8% and 2.4% for three months and drops to 1.2%, something changed: a new format, a topic your followers care less about, or simply fewer replies to other people's posts. That comparison is worth more than any industry average.

The tweet performance guide goes further into reach versus rate, including the outlier problem and why one viral post can wreck an average.

3. Follower change, split into follows and unfollows

The total follower count on your profile tells you where you are. It tells you very little about what happened. Plot net new followers per day or per week as bars, and if the export includes unfollows, draw follows above a zero line and unfollows below it.

That split catches things the total hides. A giveaway that brings in 300 follows and loses 200 of them a week later looks like modest growth in the total. In the split chart you can see both waves.

A second chart worth making: profile visits next to new follows, per week. It shows how often a visit to your profile turns into a follow.

When the rate drops on a big week, people came for one post and didn't find a reason to stay. That is often a profile problem (bio, pinned post) more than a content problem. The follower growth guide covers cumulative versus daily views, rolling averages and how to compare accounts of different sizes.

4. Performance by post type

Tag each post in the content export by format: text, image, video, thread, link. It takes ten minutes a month and is the most useful column you will add. Then plot engagement rate per post as dots, one row per type, with a tick at each row's median.

This chart answers the question most people actually have ("what should I post more of?") and shows how much evidence sits behind each answer. If threads lead but you only posted six of them, that is a hint, and it shouldn't be treated as a rule yet. Print the count next to each row.

Link posts usually have lower engagement rates because people leave to read the page. That doesn't make them worse; judge them by link clicks instead.

5. When your posts do best

Finally, timing. Build a grid with days of the week down the side and 3-hour blocks across the top, and color each cell by the median engagement rate of posts published in that slot. The engagement heatmap guide has the full method.

Two warnings. First, grey out any cell with fewer than five posts behind it, or the darkest cell will be the 3 a.m. post that happened to go well. Second, a heatmap shows where your good posts landed, which is partly where you chose to put them. Look for broad bands like "weekday mornings" rather than a single winning hour, then test it by scheduling similar posts in two slots for a few weeks.

What to skip at first

A few things look interesting and can wait:

• Hashtag clicks and detail expands, which are small numbers that rarely change a decision.
• Daily follower demographics, which barely move from week to week.
• Network graphs of who replies to whom. They are useful for understanding communities around a topic, but they need API data and careful filtering. The network graphs guide explains how when you get there.

A monthly routine

Export both files at the start of each month, paste them into the same spreadsheet or notebook, and regenerate the five charts. Write one sentence under each about what changed. After three months you'll know what normal looks like for your account, and that is what makes the next spike or slump readable.