Follower growth is only half the story – why you need to see your losses too

You know the feeling. The stream went well, the dashboard says "12 new followers", you go to bed happy. Two weeks later you look at your total and wonder where the 40 followers went that you picked up in the meantime.

Follower counts without losses are only half the story. Analytics now shows both – and with Premium, who left.

The answer is uncomfortable but boring: some of them left. And nobody told you.

The problem: every tool only counts upwards

Twitch shows you follower gains. Unfollows never appear in the creator stats, and there is no API event that fires when someone unfollows you. So most analytics tools build exactly what the interface hands them: a line that only goes up.

For you, that is a blind spot in a really inconvenient place. Losing followers on Twitch is not an accident, it is a signal. It rarely happens evenly – it happens in waves, after specific streams, after specific decisions. If you can't see the waves, you can't react to them.

How Streamboard handles it

As of the latest release, Twitch follower analytics in Streamboard show both directions.

Free, for everyone: the "Follower development" card replaces the old growth curve. For the last 30 days you get daily gains going up, daily losses going down, and a net badge for the whole window. One glance tells you whether the month was net positive – and which day tipped it.

Premium: unfollower insights. This is where it stops being a number and starts being a name. You see who unfollowed, since when they had been following you (where Twitch provides that date), and when we detected the unfollow. In other words: an actual answer to "who unfollowed me on Twitch", instead of a shrug.

Both are Twitch-only right now. Kick and YouTube don't expose per-person follower lists through their APIs, so we can't show you names there – and we're not going to invent placeholder data to make the card look busier.

How it actually works, no buzzwords

Two layers, two methods.

For the gain/loss timeline we take a daily snapshot of your follower count and record how many follows were registered that day. The losses fall out of the difference: if 15 people followed you but your counter only ended the day 9 higher, 6 left. It's cheap to compute, it needs no personal data, and that's exactly why everyone gets it.

For named unfollowers there is no shortcut. Roughly every six hours we pull your complete follower list from the Twitch API and compare it against the previous state. Anyone who was in the last comparison and is missing now has unfollowed. Twitch offers no unfollow event, so this list diff is the only trustworthy method available. The trade-off is honesty about timing: it is not real time. An unfollow can take up to six hours to show up, and the timestamp you see is when we detected it, not the second someone clicked. That's why the table says "detected at".

Fairness: your tracking window starts with Premium

One thing we'd rather say out loud than bury in a footnote: unfollower tracking starts when your Premium becomes active. The first run records a baseline of your current followers – only the run after that can detect unfollows.

There is no retroactive data. Not because we don't want to give it to you, but because it doesn't exist: Twitch keeps no unfollow history that anyone could reconstruct after the fact. Whoever left last month stays unknown. The earlier you start, the more history you'll have later.

The same logic runs the other way. The comparison only happens while your Premium is active. When it ends, we stop the tracking and delete the cached follower roster. The unfollows we already detected stay stored and remain visible with Premium. If you come back, a new window starts with a fresh baseline – the gap in between stays a gap. That keeps the whole thing light on resources, light on stored data, and easy to describe accurately.

What unfollow data is actually good for

The numbers aren't decoration. They answer questions you'd otherwise answer with vibes:

• Content decisions: did switching to that new category cost you reach, or was it just the time slot? A loss spike on the same day is a stronger argument than any chat opinion.
• Raid and event effects: big raids bring follows – and often people who are gone again two weeks later. Seeing how many stick around tells you what a spike is really worth.
• Comparing formats: if one format quietly produces a net loss every time while another grows, you have a basis for your schedule instead of a hunch.
• Staying calm: sometimes the most useful insight is that it's just background noise. A handful of unfollows a week is normal, and knowing your normal keeps you from panicking over three of them.

Go take a look

The gain/loss timeline is already in your analytics – no upgrade, no setup. The unfollower card is visible in a blurred state without Premium, so you can see what's behind it before deciding.

And if you do want names, remember how the baseline works: the best day to start tracking is the day you read about it.