Bot rate by account category, in one table
Every guide to social outreach tells you to filter out bots. None of them tell you how many to expect.
We just looked at 17,182 distinct X follower profiles across 22 tracked accounts (B2B SaaS sales tools, solo and small-team software founders, B2C SaaS products, political accounts, marketing publications, and a few others). The bot rate varied by 3x across categories, in ways that follower count alone does not predict.
Bot rate on a social media account is the share of follower profiles that signal-based detection flags as non-human (newly-created accounts with no posting history, low-effort bios cloned from templates, follow patterns that do not match human behaviour, and so on). It is not a perfect measure - any heuristic system makes mistakes in both directions - but it is consistent enough across thousands of profiles to surface real category-level patterns.
Here is the breakdown across the 22 accounts in our sample:
- Political / diplomatic (2 accounts): 61.1% bot rate
- Solo / small-team software founders (7 accounts): 54.2% bot rate
- Tech / marketing media (3 accounts): 49.2% bot rate
- B2C SaaS (2 accounts): 45.1% bot rate
- Community / events (1 account): 23.6% bot rate
- B2B SaaS (5 accounts - Apollo, ZoomInfo, Lusha, Hunter, Leadfeeder): 17.6% bot rate
- Personal, non-public-figure (2 accounts): 11.9% bot rate
The categories with only 1-2 accounts are illustrative, not conclusive. The B2B SaaS vs solo founder comparison rests on the most data: 5 and 7 accounts respectively, 17,121 follower profiles between them.
Follower count is part of the story, not all of it
The follower-count gradient is real. Below 10K followers, bot rates in our sample cluster between 9% and 22%. Above 1M, they cluster at 70%+. (Bigger account, more bots - that part will not surprise anyone.)
But once you bucket by account type, the picture gets more interesting. At similar follower-count sizes, B2B SaaS attracts much less bot traffic than solo-founder personal-brand accounts. A 23K-follower B2B SaaS account (ZoomInfo) in our sample showed 25.9% bot rate. A 66K-follower community/events account showed 23.6%. A similar-sized solo founder account showed 18.1% but climbed to 58.5% once we looked at one of the 100K+ founder accounts in the same category.
The pattern is consistent across our 22-account sample: 17.6% bot rate across 5 B2B SaaS accounts, vs 54.2% bot rate across 7 solo and small-team software founder accounts, despite the B2B SaaS group averaging much smaller account sizes (around 8K followers each on average) than the founder group (around 211K each).
Why B2B SaaS attracts fewer bots: the credibility-cover mechanism
Bots optimise for credibility cover. The fake personas they maintain have to pass casual inspection by anyone scrolling through a follower list.
"Aspiring solo founder following a prominent operator" is one of the cheapest personas to fake on X. A generic startup-adjacent bio passes for organic. The follow choice signals nothing unusual. The profile blends in with the broader build-in-public scene. (You can probably picture three accounts in your timeline like this right now.)
"Sales rep evaluating a B2B sales tool" is a harder fake. The persona has no camouflage payoff - bot armies do not follow each other from a Lusha account, so there is no internal-engagement boost to mining that audience. The follow does not pass for organic on inspection (why would a throwaway account care about Apollo?). And the profile does not blend with the broader X scene the way a startup-curious bio does.
So bots concentrate where the cover is good (personal brand, community, political accounts) and avoid where it is not (B2B sales tools).
What this means if you are mining competitor follower lists for outbound
If you are sourcing leads from a competitor or strong-signal account on X (per our usual playbook), the practical implications are:
1. Calibrate the bot-filter overhead by competitor category. Mining a B2B SaaS competitor: expect around 17% noise to filter out before classification. Mining a prominent solo founder: expect around 54%. The cleaning cost is category-dependent, and the difference is large enough to matter if you are running this at volume.
2. Account size is a heuristic, not destiny. A 23K-follower B2B SaaS in our sample (ZoomInfo) had a cleaner follower base than a 66K-follower community account at 23.6% bot rate. Do not assume follower count alone tells you what to expect when the categories differ.
3. The "bigger account = more bots" rule of thumb holds at the extremes but breaks down in the middle. Below 10K and above 1M followers, account size is a reasonable predictor. Between 10K and 1M, account type matters more than size.
This is also why CTGO bot-filters before classification rather than letting bots through to the LLM. Cheap heuristic detection up front saves expensive GPT cycles on profiles that will never convert to leads anyway. The category-by-category bot rate is what determines how much GPT spend a given tracked account avoids.
Methodology and caveats
The numbers above come from CTGO's classification pipeline running across 22 tracked X accounts, observing 17,182 distinct X follower profiles between 2026-03-18 and 2026-05-24. Bot detection is heuristic-based and reports false positives and false negatives in both directions. Accuracy is consistent across accounts (which is what lets us compare category-to-category) but any specific number should be read with a ±5% tolerance.
Sample limitations to know about before citing these:
- 22 accounts is a small sample. The higher-confidence findings (B2B SaaS at 17.6%, solo founder at 54.2%, tech/marketing media at 49.2%) rest on multiple accounts each. The single-account category (community/events at 23.6%) is illustrative only.
- Frontier sizes are capped at roughly 500 to 1,000 profiles per account on most syncs. We are seeing a slice of each account's full follower base, weighted toward recent followers - where bots tend to concentrate.
- We classify only what our pipeline has actually run on. At the time of the data pull, 5,118 profiles in our frontiers had not yet been classified, so the bot rate is calibrated against profiles where detection had already run.
Treat these as directional, not definitive. The category gaps are large enough that they survive normal measurement noise, but the precise numbers will shift as the sample grows. We will refresh this dataset quarterly.