Charting X follower growth

Follower growth needs two charts, and most people only make one. A line of total followers shows the account's size and long-run pace. Bars of net new followers per day show what actually happened, day by day: the launch, the viral thread, the week someone stopped posting. If you plot only the cumulative line, the most interesting days shrink into small kinks that nobody notices.

Everything below works with data from the account's own analytics export or the official X API. You don't need anything else, and you shouldn't use anything else: scraping X breaks its rules and gives you worse data anyway.

Getting the numbers

X's analytics section has an overview that tracks follows over time, and from the desktop view you can download it as a CSV with one row per day. Access has changed several times. The full dashboard and its export have generally been tied to a paid Premium subscription, while free accounts see stats on individual posts only, so check what your plan currently includes before you plan a report around it. Column names also shift between versions. Look for a date column plus something like new follows and unfollows; some exports include only one of the two, and some give no running total at all.

If the export lacks a total, you can rebuild one: take today's follower count from the profile and work backwards by subtracting each day's net change. Expect it to drift a little from what the profile showed in the past, because X removes spam and suspended accounts from follower counts in batches, and those removals don't always show up as unfollows.

For several accounts, or for a longer history than the export window allows, the official X API is the route. It is paid, and its access rules change, so read the current terms before building anything that depends on it.

Cumulative line vs daily bars

A cumulative line rises steadily by design. Even an account that gains followers in fits and starts draws a smooth-looking ramp, because each day adds a small amount to a large total. That is fine when the question is "how big are we, and are we growing faster than last quarter?"

Daily net change answers a different question: "what happened, and when?" Plotted as bars, a normal day of a dozen new followers is a short stub, and a day of 180 towers over the rest.

Use bars for daily change while the period is short (a month or two). For a year of daily data the bars get too thin, and a line chart of daily or weekly net change reads better.

Follows and unfollows

Net change hides two flows. An account that gains 20 and loses 10 a day looks the same as one that gains 60 and loses 50, but the second one has a churn problem. When the export gives both columns, plot follows above a zero line and unfollows below it, with the net as a label or a thin line on top.

This view is especially useful after campaigns. Giveaways, follow-for-follow drives and paid promotion often bring a burst of follows and then, a few days later, a burst of unfollows. If you only see the net, the second burst looks like a bad week with no obvious cause.

Annotating spikes

A spike in daily follows almost always has a cause you can name: a post that traveled, a mention from a larger account, a press piece, a conference talk. Write it on the chart. "+182 the day the hiring thread went viral" is worth more than the bar itself, because it tells the reader what to repeat.

Keep the annotation short and put it next to the mark, not in a legend. If several spikes compete, label the biggest one or two and leave the rest. The labels and annotations guide covers placement in more detail. Unexplained drops deserve a note too, even if the note is "spam purge by X, not real unfollows."

Rolling averages

Daily follower numbers are noisy. Weekends are usually quieter, and a single good post can double a day. A 7-day rolling average smooths the weekday pattern without hiding real changes in pace; a 28-day average shows the trend for monthly reports.

Draw the raw daily values as thin, muted bars or a light line, and the average as a single heavier line on top. Readers see both the noise and the signal, and they can tell which is which. Be clear in the label which window you used, and remember that a trailing average lags: a spike on the 14th lifts the 7-day average until the 20th.

Be careful with log scales

A log scale is tempting for follower counts, especially when an account grew from 500 to 50,000. On a log axis, equal vertical distances mean equal percentage growth, so early and late periods become comparable.

The trouble is that most readers don't know they are looking at one. They read the flattening curve as "growth slowed," when it may simply mean the account kept adding the same number of followers each month on a larger base. If you use a log scale, say so in the title or axis label ("followers, log scale"), and mark the gridlines with real values (1k, 10k, 100k). For a general audience, indexed growth below is usually clearer.

Comparing accounts of different size

Put a 48,000-follower brand account and a 2,000-follower founder account on the same axis and the smaller one hugs the bottom. Nothing it does will register.

There are two fair fixes:

• Index to 100. Divide each account's count by its value on a shared start date and multiply by 100. Every line starts at 100, and 165 means 65% growth since the start.
• Percent growth per period. Plot month-over-month growth as a percentage. This shows acceleration better, but it jumps around more for small accounts.

Either way, put the raw sizes somewhere visible, in the labels or the subtitle. Readers should know that 65% growth on 2,000 followers is 1,300 people.

Rule of thumb: if two accounts differ in size by more than about five times, don't put their raw counts on one axis.

A checklist before you publish