7 Important X (formerly Twitter) Analytics Metrics for Marketing Agencies

X analytics break down into seven things agencies report on: impressions, reach, engagement rate, follower growth, video completion, posting time, and link-driven traffic. Six of the seven need an X Premium subscription at $8 a month, either directly or through the numbers they’re calculated from. The seventh, link clicks, isn’t in the dashboard at all, on desktop or with Premium, and that changes how you report on it.

This guide walks through all seven, what each one tells a client, and where X’s own numbers can mislead you if you don’t know how they’re calculated.

How to Access X Analytics in 2026

Go to your X profile and open Premium from the left-hand menu, then select Analytics. You’ll need an active Premium subscription, starting at $8 a month on the web. Basic, the cheaper $3 tier, doesn’t include it.

Without Premium, you’re not locked out entirely. Open any individual post and tap the three dots for impressions, likes, replies, and reposts on that single tweet. It’s slower than a dashboard, but it’s free, and it’s the only path on mobile regardless of subscription tier.

Once you’re in, you can switch the account overview between a handful of preset date ranges. Screenshot the exact options before you publish anything that names them specifically. X changes this UI often enough that a written description goes stale faster than the numbers do.

1. Impressions

Impressions count how many times a tweet appeared on someone’s screen. X provides the number directly in the dashboard, no calculation needed.

The catch: impressions aren’t unique views. Someone who scrolls past the same tweet three times generates three impressions. A high number looks good in a screenshot and tells you almost nothing about how many actual people you reached, which is exactly what the next metric fixes.

Common pitfall: reporting impressions alone as a headline number. Pair it with reach or engagement rate, or a client will assume every impression was a different person.

2. Reach

Reach is the count of unique people who saw a tweet, regardless of how many times each of them saw it. X doesn’t report this number directly.

The workaround agencies use is:

Say a tweet gets 1,000 impressions on an account with 10,000 followers: that’s a 10% estimated reach. Treat it as a rough gut-check, not a real number. The math assumes every impression came from a follower seeing the tweet once, and neither half of that is true. Retweets and hashtags put your content in front of people who don’t follow the account, and a single follower scrolling past twice inflates the count. The estimate can land far off in either direction, so use it to spot trend direction, not to report a precise reach figure to a client.

3. Engagement Rate

Engagement rate measures how many people acted on a tweet relative to how many saw it:

Likes, replies, reposts, and bookmarks all count as engagements.

Here’s where most agencies get burned. Search “X engagement rate benchmark” and you’ll find numbers anywhere from 0.03% to over 1%, and none of the sources are lying. They’re just answering different questions. This gets its own section below, because the mismatch is common enough to be worth walking through in full rather than a footnote.

4. Follower Growth

Follower growth is net new followers over a period. Total followers minus total following gives you a simple net number, but the trend over time matters more than any single snapshot.

A steady climb means content is resonating. A spike usually traces back to one viral post or a paid push; a dip usually means something changed, whether that’s posting frequency, content direction, or a platform-wide shift affecting the whole account category. Look at the shape of the line before you decide what caused it.

5. Video Completion Rate

Video completion rate is the share of viewers who watch a video all the way through:

A low completion rate almost always means one of three things: the video runs too long for its content, the opening few seconds don’t earn the watch, or the storytelling loses people partway through. Compare completion rate across your own video library before you compare it to an industry number. Benchmarks for this metric vary too widely by format and length to be a reliable external yardstick; your own account’s history is the more honest comparison.

6. Best Time to Post

X’s native analytics shows which days and hours your tweets tend to pick up the most impressions. It’s a starting point, not a strategy on its own.

Test posting times against your own audience’s activity rather than a generic “9 AM Tuesday” rule pulled from a blog post. A B2B account and a consumer brand will show completely different peak windows, and those windows shift as a follower base grows. Revisit this every few months instead of setting it once.

7. Link Clicks and Website Traffic

This is the one metric on this list you can’t pull from X’s dashboard, and it’s worth being direct about why. Link click data isn’t shown in the account overview, in aggregate, or per post, on desktop, even with Premium. The only place it surfaces at all is on an individual post’s stats, and only through the mobile app.

So how do you actually report on X-driven traffic? Tag every link you share with UTM parameters and pull the numbers from Google Analytics instead. That gives you sessions, bounce rate, and downstream conversions tied specifically to X, which is a more useful set of numbers for a client than a raw click count would have been anyway. A thousand clicks with a 90% bounce rate isn’t worth reporting the same way as 500 clicks that convert.

Common pitfall: promising a client “link click tracking” as a standalone X metric. Frame the conversation around website traffic and UTM tagging from the start, and you won’t have to walk it back later.

Why Your Engagement Rate Won’t Match Published Benchmarks

Two engagement rate benchmarks can both be correct and still disagree by 60 times. The formula is the reason, not the data.

X’s own dashboard, and the formula in the metric above, divides engagements by impressions. The 2026 Social Media Industry Benchmark Report, the same benchmark long published under the Rival IQ name before the two companies combined their research, calculates its industry medians by dividing interactions by follower count instead. Those are two different denominators measuring two different things, and a number built one way isn’t a fair comparison for a number built the other way.

That’s why you’ll see an X engagement rate quoted anywhere from a fraction of a percent to well over 1%, depending on which formula and which account size produced it. Smaller accounts post one way and land higher on a followers-based formula; a large brand account with a huge follower count and modest per-post interaction can look artificially low by that same math. Neither number is wrong. They’re not measuring the same thing.

The practical fix: pick one formula, apply it consistently across every client and every reporting period, and compare a client against their own history rather than against a published industry figure calculated a different way. That’s the only comparison that’s actually apples to apples.

How X Reporting Breaks Down Past a Handful of Clients

X’s API returns only the past 30 days of post data. That’s not a lag. It’s a hard cutoff, and it turns a missed reporting cycle into permanent data loss.

Onboard a client in March and pull their X data for the first time in June, and January and February aren’t there to backfill. They’re gone. The same thing happens quietly when a client revokes account access, a token expires, or nobody notices a connection has been sitting broken for five weeks. A tool you check weekly across two accounts will catch that fast. A tool you’re supposed to check weekly across forty accounts won’t, because nobody actually opens forty dashboards every week.

This is where connection monitoring stops being a nice-to-have and becomes the thing standing between you and a permanent gap in a client’s report. Swydo’s Data Health Check flags a broken or stale X connection through an Alert the day it happens, not five weeks later when a client asks why their tweet history looks thin. You find out before the 30-day window closes on data you’ll never get back, instead of after.

Organic and Promoted Are Not the Same Number

A tweet’s performance changes completely depending on whether ad spend pushed it in front of people. One blended engagement number for the whole account hides which result your organic content earned and which one you paid for.

X’s own analytics separates promoted and organic tweets, and Swydo’s X Analytics integration carries that same Promoted versus Organic filter into your reports. Use it. A client crediting their content strategy for a number that was actually bought with ad spend is going to expect that number again next month for free, and you’ll be the one explaining why it didn’t happen.

Matching Metrics to What Your Client Actually Wants to Know

Clients don’t ask about impressions or engagement rate. They ask whether X is working. The metrics above only answer that question once you connect them to the actual business goal behind the ask.

A client asking about brand awareness wants impressions and reach, not a raw follower count. One asking about audience quality wants engagement rate and reply sentiment, not total likes. One asking about pipeline wants website sessions and conversions tied to X traffic through UTM tagging, not a click count that was never trackable to begin with. Get specific about which question you’re actually answering before you pick which number to lead with.

That reframing is most of what turns a metrics dump into a report a client reads and trusts, and it’s the same conversation that shapes broader social media ROI reporting across every channel, not just X.

Building This Into Your Client Reports

A reports-only tool like Looker Studio doesn’t watch a client’s data between rA reports-only tool like Looker Studio doesn’t watch a client’s data between reports. Someone still has to open each dashboard by hand and hope a connection hasn’t quietly gone dark. That routine doesn’t get easier as the client list grows. Five clients is manageable. Fifty means the same manual check, multiplied by ten, with nothing built in to shortcut it.

Swydo’s Monitoring suite is built around that gap:

• Metrics Overview puts up to six key metrics from every client on one screen, so you’re not opening forty dashboards to see forty numbers.
• Goals shows every client’s targets as On Track, Off Track, or Achieved from a single Monitoring → Goals view, instead of doing that math by hand for each one.
• Alerts flags a metric the moment it moves out of bounds, so you’re not scanning for a problem that already happened.
• Data Health Check catches a broken connection the day it fails.

Looker Studio and most tools like it stop at the report. Swydo keeps checking the numbers and the connections after it’s sent.

The report itself takes less setup. Swydo’s X Analytics report template pulls impressions, engagements, engagement rate, profile visits, follower counts, and the rest directly from the connected account, refreshed on whatever schedule you set. Pair it with your client’s Instagram organic metrics and TikTok organic metrics in the same report for a single social view instead of stitching together exports from three platforms by hand.

If the client runs paid on several networks, a Combined Data Sources widget merges up to five ad platforms into one, so X Ads spend lands in the same blended cost-per-result row as Google Ads and Meta. With 38 integrations and more being added, the practical limit on a cross-channel report is usually which platforms the client runs rather than which ones your tool supports.

Giving Clients One Place to Find All of It

Clients don’t want another link. Most of them have a PDF in their inbox, a dashboard URL in a Slack thread from March, and a shared drive folder, and they open none of them consistently.

Swydo’s Client Portal combines Reports, Boards, and Goals into one secure link per client. Reports carry the detailed X performance, Boards give the quick KPI read, and Goals let the client watch progress against their own targets without asking you for a status. Pick exactly what that client sees, preview it before you send it, and add a password when the account warrants it.

If the link needs to be revoked because an in-house marketer left or a contract ended, regenerate it and the old URL stops working immediately. Update the contents whenever the client’s reporting needs change, without issuing a new link.

Open the client’s account, go to the Portal tab, enable it, select the Reports, Boards, and Goals you want to share, and send the link. That’s also where “can you send me the numbers for last Tuesday” stops being an email you have to answer.

Letting Clients Ask the Report Questions

Swydo AI runs inside client-facing dashboards, so a client can ask the report a question instead of asking you. Turn it on under Settings and clients get summaries and insights generated from their own data, including every X metric above.

Understand the cost model before you switch it on across every account. Your plan includes 4,000 AI credits a month, and an average report summary uses around 95 credits. Credits reset monthly and don’t carry over, and you can set a spending limit if you’d rather cap it. You’re only billed for what you use, so an account where nobody touches AI costs nothing extra. Dashboards get a limited set of AI features compared with reports.

On data handling, requests are processed in the AWS cloud using commercially available models, the data isn’t used for training, and the whole thing is GDPR compliant under the existing data processing agreement.

Final Thoughts

Count how many of the seven you’re currently reporting, then check how many of those you’ve captured in the last 30 days. Anything older than that window is already gone, and no tool or upgrade brings it back.

The single strongest reason to fix this now is the 30-day cutoff. Every other problem on this list is a reporting choice you can revisit next month, while a closed data window is permanent.

Pick your busiest X client, confirm their connection is live, and set one Alert on it before your next reporting cycle. That takes about ten minutes and it’s the difference between catching a broken connection on day one and finding out five weeks later.