Every SMM panel claims to be the best, the fastest, the most reliable, the most established. The score we attach to each panel exists because none of those claims are testable on their own. The score is the result of seven independent signals, each computed automatically, each weighted by how predictive it has proven to be of long-term reliability.
This post walks through every signal, what we measure, where the data comes from, and how the weights got their numbers. Nothing here is hand-tuned. The weights were back-fit against 18 months of panel outcomes — panels that vanished, panels that held, panels that quietly reduced refill enforcement six months after launch.
Signal 1: Domain age
Pulled from WHOIS at the time we index a panel and refreshed monthly. Panels under 6 months get a floor; panels over 24 months get a ceiling. The shape of the curve is non-linear because the survival rate between months 3 and 9 is the steepest cliff in this industry.
Domain age alone does not save a panel. We have caught two-year-old domains that were quietly resold to new operators mid-life. To catch that, we also fingerprint the panel's API response shape and Stripe/NOWPayments wallet rotation. If the wallet changes without an announced operator handover, the domain age score is reset to the wallet age.
Signal 2: API uptime
We ping every panel's documented API endpoint on a rotating schedule and log the response. The metric is not just up/down — it is the rolling 30-day p95 latency plus the failure rate. A panel whose API answers but takes 12 seconds to confirm an order is treated as half-failed, because at scale that latency is a refund machine.
Signal 3: Refill enforcement
This is the hardest signal to measure and the most predictive one we have. We seed each indexed panel with small test orders and then re-check the delivered count 30 and 60 days later. If the panel claims a 30-day refill, we open a refill ticket the moment the standing count drops below the claimed threshold and we log whether the refill arrives, how long it takes, and whether the panel asked for additional proof.
Refill enforcement is the single biggest contributor to whether a panel score moves up or down month-over-month. A panel can be five years old, with great latency, and still lose 30 points in a quarter if they quietly stop honoring refills.
Signal 4: Claim authority
Who answers your support ticket? Is it the same handle on Telegram, on email, and on the dashboard? Does that handle have control over the panel's MX records and the payment wallet? We map this through public-facing data plus what panels disclose in their owner verification process.
Panels with diffuse claim authority — three different handles, none with verified domain control — score lower because the dispute path is unclear. When the panel disappears, you need one accountable human, not a Telegram group.
Signal 5: Review fingerprinting
Anyone can buy 200 five-star reviews. The fingerprint catches it. We cluster reviews by writing style, by posting time-of-day, by IP geography (where available), and by which other panels each reviewer has rated. Clusters that look algorithmic get their weight reduced to near zero.
The reverse is also true. A panel with 40 organic-looking negative reviews from accounts with diverse review histories outweighs 400 manufactured five-stars.
Signal 6: Price stability
We snapshot the panel's full price list every 24 hours. A panel that swings prices by more than 25% on a flagship SKU inside a 30-day window is flagged. Usually the swing is the panel chasing a wholesale supplier change. Occasionally it is the panel trying to liquidate inventory before going dark.
Stability is rewarded, not penalized — boring price curves correlate strongly with operators who plan to be around next year.
Signal 7: Dispute trail
Every contact-form submission, every chargeback report sent to us, every reply from the panel to a public review — all of it is anchored to a panel's permanent dispute trail. The trail is what catches the slow-motion failures: refill denials inching up, support replies inching out, payout wallet rotating without notice.
How the seven signals combine
Each signal is normalized to a 0–100 sub-score. The final panel score is a weighted geometric mean — geometric, not arithmetic, because we want a panel that fails badly on any one signal to be punished, not averaged into safety.
The default weights, in descending order: refill enforcement (0.28), API uptime (0.18), dispute trail (0.16), domain + wallet age (0.14), claim authority (0.10), price stability (0.08), review fingerprint (0.06). Weights are reviewed quarterly. We publish the next revision before it goes live.
None of this is magic. It is a deliberate refusal to trust anything the panel itself says about its own quality. Every signal in the score comes from data we can collect without the panel's cooperation. That is the only way to keep the score honest as panels learn what we are measuring.