AI & Tools
CupidBot's Dirty Secret: Claimed 40 Conversions, Actual 2 — A Full ROI Audit
The dashboard says 40 converts. Your bank account says 2. Here's every number CupidBot doesn't want you to run.
Updated Jun 2026 · sourced from 17 YouTube creators and 7 operator groups
Key takeaways
- One operator's CupidBot dashboard claimed 40 conversions from 1,500 convos — real count was 2.
- New per-conversation pricing consumes ~33% of earnings when real CR sits under 1%.
- Dashboard-reported 12%+ CRs are contradicted by widespread operator reports of 1–2% actual.
- The bot has been caught revealing it's Claude/AI to fans and spamming already-blocked users.
- CR has been described as steadily declining for months across multiple operator groups.
Somebody in an operator group ran the numbers last month. CupidBot's dashboard was glowing: 40 conversions from roughly 1,500 conversations.
They logged into OnlyFans to check the actual subscriber bump.
It was 2.
That is not a rounding error. That is a 95% gap between the metric the tool surfaces and the metric that pays your rent.
And the 40-versus-2 example isn't a lone anomaly — it's the sharpest data point in a pattern that operators across multiple groups have been describing for months.
This is the full audit.
What CupidBot Actually Claims
The numbers CupidBot publishes in its own marketing and in-dashboard stats are striking. One group has been tracking and sharing platform-level averages for months, and the figures are consistent: Snapchat at 12–13.7%, Telegram at 11.77–12.93%, Discord at 9.96–13.58%, Instagram at ~13%, even Grindr at 13.19% (figures attributed to January 2026 data). (habibi, Nov 2024)
Those numbers would be genuinely compelling if they were real.
They are not real conversion rates — not in the way any operator should care about.
The Counting Trick Inside the Dashboard
Here is how the inflation happens, according to operators in one group (corroborated separately, late 2025 into early 2026):
- CupidBot marks CTA-link clicks as conversions — someone taps the link, counts as a convert, whether or not they ever subscribed.
- It also marks interactions with other Cupid accounts as conversions, polluting the sample with circular traffic.
- On top of that, tier-2 and tier-3 country adds — users who fake engagement or have near-zero intent to pay — are logged as potential converts, making the denominator look better than it is.
The Cupid Conversation Filter AI is supposed to solve the last problem, with one group claiming it "quadruples conversion" by filtering low-quality traffic. That claim comes from a single source and has no independent corroboration.
Treat it as one unverified data point.
What is broadly corroborated — across a single but prolific group tracking this across multiple reports from December 2025 through June 2026 — is that real CR is running at 1–2%, not 12%. The 12.52% Discord figure is described directly as "misleading" by operators who went and checked their actual subscriber counts.
Forty claimed. Two real.
That's the math.
The Pricing Trap at Sub-1% CR
This is where it gets expensive fast.
CupidBot charges per conversation, not per conversion — a structure that made more sense when the tool was cheaper and conversions were (allegedly) higher. One group noted in early 2026 that Cupid removed its weekly plans, the 40k-conversation tier now runs $0.07 per conversation, and you have to buy in 100k bulk to hit the old $0.06 rate.
Do the math at 1% CR:
- 1,000 conversations = $70 in bot costs
- ~10 actual subscribers at whatever your sub price is
- At $10/month that's $100 gross, $70 cost — 70% of revenue eaten by the tool
Operators in one group, posting as recently as June 2026, put it more bluntly: at under 1% conversion, new CupidBot pricing takes roughly a third of your earnings. That figure assumes a somewhat better CR than the 2-out-of-1,500 extreme — but the direction is unambiguous.
The per-conversation billing model is rational for the vendor. It is punishing for the operator when the dashboard CR and the real CR diverge by a factor of ten.
The Declining CR Trend
This isn't static. Multiple operators, across what appears to be one very active group tracked from late 2025 through mid-2026, describe conversion rates as steadily dropping for months.
Not a one-week anomaly. Not a platform update blip.
A directional trend.
That trend matters for projections. If you modeled your ROI on the December 2025 dashboard numbers and you're billing today, you're already behind.
Nobody in the evidence base offers a clean explanation for why the drop is happening. Possible causes floating in operator discussion: platform antibot improvements, audience fatigue with the funnel copy, or the quality issues below.
The Bot Is Outing Itself
This one is bad.
As of around May 2026, operators in one group began reporting that CupidBot was:
- Repeating words mid-conversation, creating obviously robotic text patterns.
- Refusing roleplay requests — the core interaction type that drives OFM revenue.
- Telling fans it is Claude / an AI — directly, in the chat window, to the person you're trying to convert.
Point three is a conversion killer by definition. Subscription willingness drops sharply the moment a mark realizes they're talking to a bot. (Bjorn Olsen, Apr 2026)
The entire premise of the tool relies on the conversation feeling human enough to move someone toward a paid subscription.
A bot that volunteers its own AI identity mid-funnel isn't just underperforming. It's actively working against you.
The Blocked-User and Platform Problems
Two more documented failure modes worth flagging:
Blocked users. Operators have reported CupidBot sending messages to users who have already blocked the account. This doesn't just waste conversation credits — it risks platform flags and further bans on accounts you've invested in.
Grindr and Discord. Both platforms are cited in the evidence as underperforming relative to their dashboard numbers. Discord's "real"
CR of 1–2% against an advertised 12.52% has been called out explicitly. Grindr AI bot figures (13.19%) appear to be January 2026 vintage — potentially dated, and sourced from the same group that tracks these numbers rather than independently verified.
Where Operators Actually Disagree (Both Sides)
Fairness demands surfacing the conflict in the evidence, not just the skeptics.
The bull case: A subset of operators in the same group that identified the 40-versus-2 problem also posts dashboard screenshots showing double-digit CRs. Some of these may reflect genuine high-performing accounts, better traffic sources, or specific platform/niche combinations.
The Snapchat numbers (10.57%–13.7% across multiple reports spanning December 2025 to March 2026) appear with unusual consistency, suggesting Snapchat may genuinely outperform other platforms even on real CR — though no operator in the evidence confirms a Snapchat real-versus-dashboard comparison the way the 40/2 case does.
The bear case: The 40/2 data point, the "1–2% real" characterization of Discord specifically, and the "steadily dropping" trend language all come from the same group source — meaning corroboration is wide within that group but the group itself could be biased, wrong, or containing disguised competitive talk. One group is not six groups.
Bottom line on the disagreement: the preponderance of evidence leans heavily toward the real CR being materially below dashboard CR. But the degree of gap may vary by platform, traffic quality, and operator.
Snapchat appears to be the least-bad platform in the evidence. Discord and Instagram have the most skeptical operator commentary.
The Comparison That Matters
For context on what a non-inflated AI funnel bot looks like: one vetted creator reports Capital AI hitting 10–14% real conversion rates on Snapchat and Telegram funnels to OnlyFans, and credits the higher rate as compounding meaningfully against blackhat infrastructure costs. (Dr. Hadi Talks, May 2026) (Dr. Hadi Talks, May 2026)
That's a vetted, on-record claim about a competing tool — not anonymous group chatter. It doesn't prove Capital AI delivers on that number consistently, but it establishes a benchmark that makes CupidBot's real 1–2% look worse, not better.
What the Numbers Actually Say You Should Do
Run your own attribution. Not the CupidBot dashboard — your actual OnlyFans subscriber count before and after a controlled run.
Exclude any period where you ran other traffic simultaneously. The 40-versus-2 operator did this.
It took about five minutes and changed their entire cost model.
If your real CR is above 5%, the per-conversation pricing is probably defensible. Below 3%, you are likely funding Cupid's margins more than your own.
Below 1%, the June 2026 operator math holds: roughly one-third of gross revenue goes to the tool.
The platform-specific evidence suggests Snapchat is your best bet if you continue using CupidBot — multiple report windows across several months show the most consistent numbers there, and the feature set (watermark removal, auto-accept, story scheduling) adds operational value beyond just conversation volume.
Avoid Discord until you've verified your own real CR. The gap between advertised and actual is the most directly documented in the evidence for that platform.
On the AI-outing issue: this is non-negotiable. If your bot is telling fans it's Claude, pause the account and check your conversation logs before running another dollar of traffic through it.
One conversation like that doesn't just lose a sub — it can get screenshotted and distributed. (Only Hustlas, Apr 2026)
The Bottom Line
CupidBot is not a scam in the legal sense. It's a tool with a measurement problem, a pricing model that punishes low CR, a documented and worsening quality issue, and a gap between its marketing numbers and operator reality that — at its most extreme — is a 20-to-1 overstatement.
The dashboard is not lying to you maliciously. It's counting things that aren't conversions and calling them conversions.
That distinction matters less than you think when your expenses are real and your subs are imaginary.
Track your own numbers. Price accordingly.
And maybe stop taking any tool's dashboard at face value — including this one.
Sources
On the record (YouTube creators):
- habibi — Onlyfans Reddit Strategy OCT 2025**, Nov 2024. Watch ↗
- Only Hustlas — How to Get Unlimited Free Traffic For Your OnlyFans, Apr 2026. Watch ↗
- Bjorn Olsen — $30,072 Per Month From ONE AI Model Fanvue Using Reddit (AI OFM), Apr 2026. Watch ↗
- Dr. Hadi Talks — Reddit OFM Blackhat 2026 Method (Full Guide), May 2026. Watch ↗
Community intelligence: 88 operator claims aggregated from 7 separate private OFM groups (Dec 2025–Jun 2026), corroboration counted across groups. Group identities are withheld to protect sources; browse the underlying intel in the Community Intel Wiki.