Why X monthly active user counts vary between sources

Build a B2B campaign on X and the Ads Manager reach estimate often lands 20-40% below what a headline user count predicts. That gap is not a bug in the ad platform. X stopped reporting monthly active users after Q1 2019, and every figure published since counts a different population.

Twitter Monthly Active Users: the Numbers in Conflict

Sources disagree about Twitter monthly active users because they are counting four different populations, and only one of those populations was ever audited. The verified floor is 237.8 million monetizable daily active users, filed with the SEC in Twitter's Q2 2022 10-Q. Sensor Tower measured 252 million monthly active mobile users in Q4 2024. BankMyCell's aggregated estimate put 2024 global MAU at 335.7 million. Musk said 600 million in May 2024. Each figure is defensible on its own terms. Not one of them is interchangeable with another.

The data splits into three pools that do not speak to each other. Pool one is regulated disclosure: audited mDAU, which counts logged-in users on ad-eligible surfaces on a single day. Pool two is third-party panel measurement, meaning mobile app sessions observed from outside the company, which by construction cannot see desktop web, logged-out reading, or third-party clients. Pool three is self-report, a number posted publicly by the owner of a private company with no methodology attached. Stacking figures from all three pools into one chart is the most common error in X user statistics, and most aggregator pages make it.

What changed in October 2022 was not the user base. It was the verification mechanism. Before the acquisition closed, every user figure Twitter published carried an audit trail, a stated accounting policy, and legal exposure if it was wrong. Once the company went private, the SEC filing obligation ended. No post-2022 figure has that backing, which means the honest answer to how many people use Twitter is a range with an error band rather than a number.

We build scheduling and engagement automation, so this gap reaches us as a forecasting error rather than an academic dispute. Engagement rate benchmarks derived from the 600 million figure come in hot every time. The active, reply-generating audience for B2B content on X behaves closer to the 250 million mDAU range. If your content model assumes 600 million, your denominator is roughly twice the size of the population that can see the post and respond to it.

That is why we treat the disagreement as a planning input. A team budgeting against 600 million and a team budgeting against 237.8 million will build different content calendars, run different paid tests, and hold different definitions of a disappointing quarter. The first team will conclude its content is failing when the content is fine and the forecast was wrong. That failure pattern is hard to diagnose after the fact, because nothing in the reporting says the denominator was inflated.

The dispute also persists because every party has an incentive in it. X benefits from the largest defensible number when selling ads. Third-party measurement firms benefit from being the ones who caught the discrepancy. Statistics aggregators benefit from ranking for the question, and ranking rewards a single confident figure over a range. Almost nobody in that chain is paid to say the number is unknowable within a 2x band, so almost nobody says it.

Why Did Twitter Stop Reporting Monthly Active Users in 2019?

Twitter stopped reporting MAU because the metric had stopped growing and had started shrinking. MAU sat flat at 320-330 million for years and was declining through 2018 as bot and spam purges removed accounts from the count. Q1 2019 was the last quarter Twitter published both figures side by side. From Q2 2019 forward, monetizable daily active users became the primary public metric, and the company never restored MAU as a headline disclosure.

The spam purges are the part most summaries skip. Removing inauthentic accounts is correct hygiene and terrible optics, because it produces a quarter where the top-line user number falls for a reason that is good news. Reporting a metric that punishes you for cleaning your own platform is a bad long-term position for a public company, and Twitter was in it.

Then there is the gap the final dual-reporting quarter exposed. In Q1 2019, MAU was approximately 330 million while mDAU was only 134 million. Fewer than half of the monthly users were in a monetizable logged-in state on a given day. That single comparison is the most useful number in this entire topic, because it is the only time anyone published both definitions against the same period with an audit behind them. It tells you the size of the error you introduce by treating monthly reach and daily monetizable reach as the same thing.

The switch also created a broken time series, which is the source of most contradictions you will find across sources today. Any chart titled Twitter active users by year that runs continuously from the early 2010s through the 2020s is splicing pre-2019 MAU onto post-2019 mDAU. There is no published conversion factor between the two. The Q1 2019 overlap gives you one quarter where both existed, and treating that single ratio as a permanent constant has no basis in anything Twitter disclosed.

So when you see a smooth line covering both eras, the line is fabricated somewhere in the middle. The fabrication is usually not malicious. Someone needed a continuous series for a chart, the two metrics were both labeled users, and nothing in the spreadsheet warned them that the definition changed under their cursor.

When we set our own audience-sizing assumptions, we split the series at Q1 2019 and refuse to interpolate across the break. Two short series with a documented discontinuity are more useful than one long series that quietly lies. Any year-over-year comparison that crosses 2019 is a comparison between two different questions, and you should say so out loud when you present it.

The 237.8 Million Benchmark Nobody Can Dispute

237.8 million mDAU is the last X user figure that was independently audited and filed with the SEC. It comes from Twitter's Q2 2022 10-Q, the final quarterly report before Musk completed the acquisition and took the company private in October 2022. Every user number published after that quarter is either self-reported or a third-party estimate with no regulatory backing behind it.

The distinction is worth being precise about, because people use the word verified loosely. A figure in a 10-Q went through an accounting policy, internal controls, and an auditor, and carried securities-law consequences if it was materially wrong. A figure in a public post went through one person. Both may be accurate. Only one of them was constructed under a system designed to catch inaccuracy, and when the filing obligation ceased, that system stopped applying to X user data entirely.

For practitioners, 237.8 million has a second property that matters more than its audit status: it describes the population automation can reach. X's API requires authenticated sessions and enforces rate limits on them, so any scheduling, monitoring, or engagement tool interacts only with logged-in accounts on the surfaces the platform exposes. That is the mDAU population. It is not the MAU population, which includes logged-out readers arriving from search and social embeds plus users of third-party clients the API cannot attribute.

That makes mDAU the only figure with operational meaning for automation-driven B2B outreach. A logged-out reader cannot be followed, replied to, or messaged. They can read your post and never enter any funnel you control. If your outreach model is built on a number that includes them, you have inflated your addressable audience with people your tooling is structurally incapable of touching.

The obvious limitation is recency. The figure describes Q2 2022 and nothing has audited X user data since, so it cannot be presented as current truth. We use it as a floor and a sanity check rather than a live estimate. When a new claim appears, the first question is whether it is plausible relative to the last audited count and in which direction the third-party trend lines point.

In practice that gives you a bracket instead of a point estimate: 237.8 million mDAU as the audited floor from Q2 2022, and the 550 million MAU figure from the SpaceX S-1 covering March 31, 2026 as the most recent number to appear in a regulated document. Those two are the only X user figures ever filed with the SEC in either era. Everything else in circulation sits outside that bracket's evidentiary standard, which is a useful way to sort sources quickly.

Three Incompatible Definitions Drive the X Monthly Active User Debate

mDAU counts logged-in users on ad-eligible surfaces on a specific day. It excludes logged-out readers, third-party app users, and any surface that does not serve ads. That definition was written for an ad business, not for cross-platform comparison, and it makes mDAU non-comparable to the MAU and DAU figures Facebook, Snap, and other platforms report. Anyone placing X's mDAU next to another platform's DAU in a competitive deck is comparing two different measurement philosophies and calling the difference market share.

Sensor Tower's number measures something narrower still. It reported 252 million monthly active mobile users for X in Q4 2024, covering mobile app installs and sessions, and it found U.S. mobile DAU down 18% year-over-year as of February 2024. X disputed that reporting publicly. Both positions can hold: Sensor Tower was describing mobile app behavior, while X was describing a total audience across surfaces a mobile panel cannot observe. The disagreement is about scope before it is about accuracy.

Similarweb measures differently again, and reached a different number: 132 million daily mobile users as of June 2025, down 15.2% year-over-year, with Threads surpassing X in mobile DAU in September 2025. Similarweb's panel methodology is not Sensor Tower's, and neither maps onto X's mDAU definition. Three credible measurement organizations covering overlapping periods produce three figures that cannot be reconciled, and none of them is wrong within its own frame.

Then there is a fourth definition, and it is the one that takes your money. X Ads Manager applies its own active-and-targetable filter on top of any published user count, stripping out stale, low-engagement, and ad-ineligible accounts in real time. When we pull reach estimates for B2B segments filtered by job title, industry, and follower lookalikes, they come back consistently 20-40% below what back-of-envelope math from published MAU figures would suggest. That is not the ad platform being pessimistic. It is the only number in this topic that reflects accounts the system is willing to serve an impression to.

The failure mode here has a shape worth naming: surface mismatch. A practitioner reads a mobile-panel figure, compares it against a self-reported all-surface figure, concludes that one source must be lying, and then picks whichever number supports the plan they already wrote. The correct move is to ask which surface each figure covers before asking which one is right, because the answer to the second question is usually both.

Once you sort figures by surface, the contradictions mostly dissolve. Mobile panels will always read lower than platform totals because they cannot see desktop web, logged-out reading, or third-party clients. Self-reports will always read highest because they include every surface and face no audit. mDAU sits in between by design, since it is deliberately restricted to the monetizable slice. The numbers are not fighting. The labels are.

Musk's 600 Million Claim vs. Third-Party Estimates

In May 2024, Musk publicly claimed X had 600 million monthly active users, with roughly 300 million using the platform daily. The claim came via a public post. No methodology was disclosed, no third-party audit accompanied it, and no SEC filing supported it, because as a private company X had no obligation to file one. That figure is nonetheless the most widely repeated X user number across search results, which is how an unverified claim becomes the default answer to a factual question.

Independent estimates for the same calendar year point the other way. BankMyCell's aggregated estimate placed 2024 global MAU at 335.7 million, down 5.14% from 353.9 million in 2023. Note that the disagreement is not only about magnitude. One source describes a platform near 600 million and growing; the other describes a platform near 335 million and shrinking. Those are not two measurements of one trend. They are two incompatible stories about the same year.

The full spread of 2024 figures runs from roughly 335 million to 600 million, a gap of nearly 2x, and that gap is driven entirely by methodological differences and the absence of any verification mechanism. There is no way to adjudicate it from the outside. Nobody can audit a private company's internal definitions, and the company has not published them.

Bot and multi-account inflation sits underneath every figure in the debate. During the 2022 acquisition dispute, the share of inauthentic accounts was argued in the 10-20% range and never settled with a public audit. If that range is even approximately right, every MAU figure in circulation should be presented with an uncertainty band on top of its definitional ambiguity. None of them are. Aggregator pages publish point estimates to one decimal place for a quantity whose authentic share was litigated and left unresolved.

For anyone running automation, that dilution is a safety problem before it is a measurement problem. Dormant and bot-operated accounts dilute engagement signals, meaning likes, replies, and reposts per impression run lower than a clean-audience model predicts. Tools that throttle activity based on engagement-rate triggers have to calibrate against the real active pool rather than a self-reported MAU figure. A throttle tuned to an inflated denominator reads ordinary performance as failure, and then either pushes volume up to compensate or backs off a healthy account for no reason. We calibrate against observed session-level engagement instead, because it is the only pool we can measure directly.

Our working resolution is unglamorous and has held up well: treat the low end of the credible range as the planning assumption and the high end as a ceiling you never spend against. If a campaign only clears its targets under the 600 million assumption, it does not clear its targets. That rule costs you some upside on the rare occasions the optimistic figure is closer to true. It also stops you from building a quarter of content strategy on a number nobody can check.

When X User Data Returned to an SEC Filing

On May 20, 2026, SpaceX's S-1 IPO filing disclosed 550 million monthly active users for X as of March 31, 2026. That is the first X user figure to appear in a regulated SEC document since Twitter's Q2 2022 10-Q. For a topic where every intervening number came from a public post or an outside panel, a figure inside a registration statement is a meaningful change in evidentiary quality.

It does not close the question, though, and it is worth being clear about why. An S-1 is company-prepared and self-defines its operating metrics. The filing is submitted under securities law, which raises the cost of stating something false, but it is not the same as an operational metric audited quarter after quarter under a public company's reporting controls. The disclosed definition is whatever the company says it is, and there is no external party reconciling it to prior-era methodology.

The second limitation is continuity. The S-1 figure is MAU. The last audited figure was mDAU. They belong to different pools, so 550 million in 2026 does not extend the 237.8 million series from 2022; it starts a third one. If you want a year-over-year story out of X user data, you still do not have one, and no amount of arranging these figures left to right will produce it.

The direction is still informative. The 550 million disclosed in a regulated filing for March 2026 sits below the 600 million claimed in a public post for May 2024. Read that carefully rather than dramatically: two different definitions, two different periods, one of them written under legal exposure and one not. What it does not support is the smooth growth narrative that the self-reported figure implied.

Why does the credibility of a user count carry commercial weight at all? Because advertisers price against audience size. X generated $2.5 billion in revenue in 2024, down 13.7% year-over-year from a $5.08 billion peak in 2021, and the advertiser exodus behind that decline was partly driven by the inability to verify the audience being sold. When the only figures available are unaudited, media buyers discount them, and a discounted audience estimate becomes a discounted budget.

The practical read for practitioners is narrow and useful. The S-1 figure is now the best available anchor for current platform scale, it belongs in a different bucket than audited mDAU, and it should be labeled as company-prepared every time you cite it. Putting it in a deck next to the Q2 2022 audited number without that label recreates exactly the definitional splice that makes this topic a mess.

How to Read X Monthly Active User Data Without Getting Misled

Use mDAU as your operational baseline, not MAU. Two figures have appeared in regulated documents: 237.8 million mDAU from Twitter's Q2 2022 10-Q, which is your audited floor, and 550 million MAU from the SpaceX S-1 covering March 31, 2026, which is the most recent regulated estimate. Everything else you encounter is a third-party panel measurement or a self-report. Sorting figures by that test takes ten seconds and eliminates most of the confusion in this topic.

Next, identify which surface a source can see. Sensor Tower and Similarweb measure mobile app activity, so they will always read lower than platform totals, because desktop web sessions, logged-out reading, and third-party client activity are invisible to a mobile panel. That does not make their numbers wrong or their trend lines unreliable. It makes them answers to a narrower question, and the narrower question is often the one you care about.

For paid planning, stop deriving audience size from published user counts entirely. Pull the estimate directly from X Ads Manager with your target job titles, industries, and seniority levels applied, and treat what comes back as a ceiling rather than a starting point. Expect it to land 20-40% below what MAU math would predict, because the platform is applying its own active-and-targetable filter. The Ads Manager figure is the only number in this topic that describes accounts the system will serve your impression to.

For organic planning, the relevant benchmark is not platform scale at all. The practical engaged audience for a B2B account on X runs around 5-15% of follower count per post in our data, and it moves with post recency and engagement velocity rather than with any headline user figure. That range is the clearest demonstration of how little MAU claims tell you about reach: your content competes for attention inside a small slice of an audience you already earned, not inside a 600 million pool.

Remember the structural constraint on tooling, because it decides which population is reachable rather than merely visible. X's API requires authenticated sessions and enforces rate limits on them, so automation platforms, including ours, only ever interact with logged-in accounts on API-accessible surfaces. Reach numbers reported by social media management tools will therefore differ from both self-reported MAU and third-party panel estimates, and the difference is not an error in any of the three. Each is measuring a different slice, and the slice a tool can measure is the slice it can act on.

When a new X user count appears in coverage, run three filters before you repeat it. Is it MAU or mDAU? Was it filed in a regulated document or self-reported? Does it cover all surfaces or mobile only? Those three answers place the figure in its comparison bucket, and a figure in the wrong bucket is worse than no figure, because it looks like evidence.

The uncomfortable conclusion is that there is no correct number to publish here, only a correct way to handle an uncertain one. Build your models on mDAU-scale assumptions, pull reach from the ad platform with your real filters applied, benchmark organic against your own follower base rather than platform totals, and label every citation with its definition and its verification status. Practitioners who do that will forecast X campaigns more accurately than practitioners quoting a bigger, rounder, more confident figure from an aggregator page.

Frequently asked questions

Why do different websites report different numbers for Twitter monthly active users?

Each source measures a different population. X's mDAU counts only logged-in, ad-eligible users on a daily basis. Third-party tools like Sensor Tower measure mobile app sessions only. Musk's self-reported figures cover all surfaces without a disclosed methodology. Because no two sources use the same definition, figures for the same period can differ by nearly 2x.

What is the difference between Twitter MAU and mDAU, and why does it matter?

MAU counted any account that logged in at least once in a 30-day window. mDAU counts only logged-in users on ad-eligible surfaces on a specific day. In Q1 2019 Twitter reported both: MAU was 330 million while mDAU was only 134 million. The gap matters because mDAU is the population X's ad products actually reach; MAU-based projections consistently overshoot.

Why did Twitter stop reporting monthly active users in 2019?

Twitter's MAU had been flat at 320-330 million for years and was declining through 2018 as bot and spam removal reduced the count. mDAU reflected only the monetizable engaged user base, which aligned better with the ad-supported business model. Twitter announced the switch with its Q1 2019 earnings and never returned to MAU as a primary disclosed metric.

How many monthly active users does X (Twitter) have in 2026?

The most recent figure in a regulated document is 550 million, from SpaceX's S-1 IPO filing covering the period ending March 31, 2026. Third-party estimates for 2024 ranged from 335 million (BankMyCell) to 600 million (Musk's self-report). Sensor Tower's Q4 2024 mobile-only figure was 252 million. None of these use the same methodology, so direct comparisons between them are not valid.

Are Elon Musk's X user count claims independently verified?

No. Musk's May 2024 claim of 600 million monthly active users was made via a public post with no disclosed methodology, third-party audit, or SEC filing. The SpaceX S-1 from May 2026 placed MAU at 550 million in a regulated document, but that figure is company-prepared rather than independently audited.

What was the last officially audited Twitter user count?

237.8 million mDAU, reported in Twitter's Q2 2022 10-Q filing with the SEC. That was the final quarterly report before Elon Musk completed the acquisition in October 2022 and took the company private. As a private company, X has no regulatory obligation to file audited operational metrics.

How do third-party tools like Sensor Tower and Similarweb measure X users differently than X does?

Sensor Tower and Similarweb track mobile app sessions using panel-based methodologies; they cannot observe desktop web traffic, logged-out reading, or third-party client activity. X's mDAU counts logged-in users across all ad-eligible surfaces. As a result, Sensor Tower reported 252 million monthly active mobile users for Q4 2024 while X claimed a substantially larger total audience for the same period.

Does X's 600 million MAU claim include bots and fake accounts?

Unknown. X does not disclose what share of its user base consists of inauthentic or automated accounts. During the 2022 acquisition dispute, bot prevalence was central but never resolved with a public audit. If 10-20% of active accounts are inauthentic, as was debated then, every published MAU figure carries an undisclosed uncertainty band that no source acknowledges.

How has X's user count changed since Elon Musk acquired Twitter in 2022?

There is no audited time series after October 2022. The last verified figure was 237.8 million mDAU from Q2 2022. Musk reported 600 million MAU in May 2024; the SpaceX S-1 listed 550 million as of March 2026. Third-party data points differently: Similarweb reported a 15.2% year-over-year decline in daily mobile users as of June 2025, and Threads surpassed X in mobile DAU in September 2025.

Why does the X Ads Manager audience estimate differ from published MAU figures?

X Ads Manager applies a real-time active-and-targetable filter that removes stale, low-engagement, and ad-ineligible accounts. When B2B targeting criteria (job title, industry, seniority) are applied, the resulting estimate typically runs 20-40% below what published MAU math would produce. The platform is measuring the practically targetable audience at campaign setup, not the headline reported user base.

Sources and further reading

• Twitter's Q1 2019 SEC earnings release, where mDAU was formally defined and both MAU and mDAU were reported in the same filing for the last time
• Twitter's full SEC filings archive, including the Q2 2022 10-Q containing the last independently audited user count
• Social Media Today's coverage of Musk's 600M monthly active users claim, including the absence of any disclosed methodology