Twitch viewbots: how to spot fake viewers before you pay

A streamer's viewer count is the number many sponsorship deals are priced on, and it is the easiest one to fake. Viewbots inflate it for a fee; the brand that pays a flat fee per stream pays for those viewers too. Twitch has been tightening detection since July 2025, but no platform filter turns a viewer count into proof that people saw your ad. Here is what a viewbot is, what Twitch has actually done about it, and how to check an audience before you buy.

What a viewbot is, and what it is not

A viewbot is software, usually sold as a service, that opens many fake sessions on a live stream so the concurrent viewer count goes up. Some services also run chat bots: accounts that post generic messages so the chat looks as busy as the number suggests. In an X post on 28 July 2025, Twitch CEO Dan Clancy separated two kinds: streamers who pay third parties to inflate their numbers, and bots sent to a channel to harass the streamer. Twitch itself says viewbotting can be out of the streamer's control.

Not every viewer who isn't watching is a bot. According to Twitch's support documentation, as summarized by Net Influencer (31 July 2025), the viewer count is the number of people watching a live video at one time, whether or not they are signed in. Muted streams, background tabs and minimized mobile streams count as legitimate views. “Users in Chat” is a different number: it counts only signed-in accounts connected to chat.

Between the two sits a grey zone. In 2023, Fortune described streamers embedding their streams on third-party sites, such as pages with gaming tips, where visitors often don't know they are being counted. Streamer Gothalion estimated embeds can lift viewership by 80%; Twitch declined to share embed data and said the vast majority of streamers use the embed player as intended.

Why fake viewers cost the advertiser, not the platform

Fortune reported that Twitch doesn't place its own ads on the embedded player, so it says it doesn't directly profit from those views. The party that does pay is whoever prices a deal on concurrent viewers. In 2025, Clancy said “thousands of small streamers” use inflated view counts to generate “bogus revenue”, as Tubefilter recalled on 8 May 2026. In the same Fortune article, Gothalion said he knew of a marketer who was fired after an underperforming campaign with a creator whose viewership had been inflated.

In the US, the FTC's rule on fake reviews and testimonials, in effect since 21 October 2024, prohibits selling or buying fake indicators of social media influence, including bot-generated ones, when the seller or buyer knew or should have known they were fake and they are used to misrepresent influence for a commercial purpose (16 CFR 465.8). The FTC's own Q&A adds that hiring an influencer who happens to have fake followers is not covered. The rule targets whoever sells or buys the fake numbers, not the brand that hires the streamer; the wasted spend, however, stays with you.

What Twitch changed in 2025 and 2026

The size of the August 2025 drop is disputed. PPC Land cited unnamed analytics showing viewership 24% below the previous week on 22 August; Twitch said viewership was “by no means in decline or in free fall” and called some third-party figures misinformation. When numbers climbed back days later, Twitch said: “We haven't rolled back any of our viewbot tools or removal efforts.” Treat every percentage here as an outside estimate.

The 2026 cap is based on the creator's historical non-viewbotted traffic and gets longer with repeat violations; the streamer is notified and can appeal, Tubefilter reported. Individual enforcements are not announced publicly, according to Net Influencer (11 May 2026). Clancy says details stay private because viewbot makers adapt fast.

How big is the problem? In a whitepaper reported by Tubefilter on 26 September 2025, Streams Charts estimated that in Q2 2025 at least 10% of Twitch accounts averaging 50 or more viewers showed clear signs of persistent viewbotting, and more than 16% on Kick. Those are flags of likelihood, not confirmed cases, and the paper was co-published with an influencer marketing firm.

Warning signs you can check yourself

“Viewbot checker” sites and third-party trackers work from public numbers, which Twitch said will shift as its filtering changes. Use them for history, not as a verdict, and look for patterns:

No single sign proves viewbotting. Two or three together are a reason to ask for data before you sign. Which audience metric to buy on in the first place is covered in CCV, average and unique viewers.

Questions to ask a streamer or agency before you buy

1. Analytics, first-hand. Ask for a screen share of the channel's Twitch analytics for the last 90 days, not a screenshot of a tracker site.
2. History across August 2025. How did the average audience change around the detection update, and has it recovered?
3. Traffic sources. Is the stream embedded on other sites, and does it autoplay there?
4. Enforcement. Has Twitch ever notified the channel of a viewer cap or other action?
5. Past campaign reports. Were screenshots captured at the moment of display, or sent by the streamer afterwards?
6. The unit you pay for. A flat fee priced on concurrent viewers, or a price per view that is checked before it is billed?

How APIQ checks each view it bills

APIQ doesn't sell a streamer's viewer count. It sells verified views at one CPV rate, with a guaranteed volume. Here is what “verified” means in practice:

• A screenshot for every counted view. The system captures the stream at the moment of display and records the streamer, time and audience size.
• Computer vision. It checks that the creative is present, intact and in the right position, and that clickable elements work; hidden overlays and broken creatives don't count.
• Anti-fraud across 15 parameters. Signals include chat activity, the ratio of viewers to chat participants, audience dynamics and stability, sudden spikes, how regular and long a channel's streams are, whether viewer profiles have an avatar and a description, clicks and conversions. The full list is available on request.
• Consequences. Suspicious traffic is excluded from the report and the invoice. If bots are found, those views aren't paid for and the streamer is removed from the platform.
• Outcome signals. Each channel gets its own link, CTR is tracked online, and a streamer who falls below the threshold agreed with you is switched off automatically.

The limits, precisely: this checks that your creative was on screen and that the audience behind each view passes quality checks. It doesn't replace Twitch's own detection, and it isn't a promise that no bot ever reaches a stream; it means suspicious views don't reach your invoice. Your own attribution through AppsFlyer, Adjust, UTM tags or pixels adds an independent check. How to audit the report is explained in How to check that your ad actually ran, and what screenshots show about context in Brand safety in live streams.

The media plan arrives within 24 hours, and the launch follows 1–2 days after the streamer and creative list is approved. More on the brands page.