We Ran 15,867 of Our Own Followers Through a Bot Checker. Here Are the Unedited Numbers.

Why We Did This

There's an obvious question you should ask any follower service: if I audit the followers you deliver, what will I find? Most services survive that question only because nobody asks it. Their followers are bots with default avatars and zero tweets, and any free audit tool exposes them in seconds.

We built one of those audit tools ourselves — TweetScan — and we tell every prospect to run us through it. Which raises the stakes: our own bot detector had better not flag our own deliveries. So we stopped wondering and ran the experiment properly.

The Setup

We picked a client account with a useful property: a US medical professional who had roughly 3,000 organic followers before starting with TweetBoost and about 15,900 followers after months of campaigns. Because X returns follower lists newest-first, the oldest ~3,000 followers are his pre-TweetBoost organic audience and the newest ~12,900 are dominated by campaign-delivered followers.

That gives the experiment a built-in control group: the client's own organic audience. Not a benchmark we invented — real people who found him on their own, scored by the exact same model.

We then pulled the complete follower list — all 15,867 accounts, no sampling — and scored every one with the same public scoring model TweetScan uses on any account you scan: profile completeness, tweet history, account age, follow ratios, and mass-follow patterns.

The Results

The followers we delivered are statistically indistinguishable from the client's organic audience. Same real-percentage (99.1% vs 98.9%), same median quality score (100), nearly identical failure tails. A bot detector cannot tell our campaign-delivered followers apart from people who found the account on their own — because they are people who found the account through a promotion and chose to follow.

The Part a Marketing Page Would Hide

0.3% of the delivered cohort scored as bot-like — around 40 accounts out of 12,867. We looked at them individually: accounts following 7,000+ people with 70 followers and zero tweets. Genuine junk. Follow-bots exist in every large audience (the organic control group had them at 0.4%, slightly more than our delivered cohort), but we'd rather report the number than round it away.

We also ran a second, smaller account from our crypto tier: 96% real, 3% suspicious, 1% bot. Slightly lower — crypto audiences skip bios more often (24.6% of that cohort had none, versus 9.6% in the medical cohort), and anonymous accounts score more cautiously. Still a range most “real followers” services can't reach, and we publish the difference between our tiers instead of pretending it doesn't exist.

As a result of this experiment, we now run this exact audit as an automated weekly quality gate on recent deliveries. If a batch drifts below our thresholds, we catch it and replace it — before a customer's own audit would find it. That's the operational half of the Audit-Backed Guarantee.

Run the Same Experiment on Anyone

The scoring model we used is not a private lab instrument. It's the same one behind TweetScan, free to use on any public X account:

• Scan an account grown by another follower service and look at the bot percentage.
• Scan an account we've grown.
• Scan your own — you'll know exactly where you stand before spending a dollar.

Every service says “real followers.” We define the standard every delivered follower must meet, publish the audit tool, and post our own results. Don't trust us. Test us.