Trust and delivery, measured
What we delivered, what we refunded, and when we read it
Every figure below is read from our own order ledger when this page loads, for 30 days. None of it is a promise about your order: it is what happened to the orders already placed.
Read from the live ledger on 2026-11-20 01:21 UTC.
Per supplier, the same window
Every order a supplier accepted in the window, including our own test orders (14 of them): they are real deliveries on the same route a customer order takes. A zero is shown as a zero.
| Supplier | Orders | Full | Partial | Refunded | Auto-refund | Median start, min | Last order |
|---|---|---|---|---|---|---|---|
| smmxpanel | 13 | 8 | 0 | 5 | 0 | 3.2 | 2026-11-17 |
| smmcost | 4 | 0 | 0 | 4 | 0 | — | 2026-11-14 |
| smmgen | 4 | 3 | 0 | 1 | 0 | 2.2 | 2026-11-12 |
A percentage is printed only when at least 20 orders are settled. Below that the counts stand alone: a rate from two orders is noise.
The refund rule, and what it did
A paid order that has not started at the supplier 180 minutes after payment is refunded in full to the wallet by the order worker, with no ticket. The worker checks every 2 minutes. The rule in full →
3 orders have been refunded by this rule since 2026-11-06, $12.06 in total — every order counted, including our own test orders.
Availability, from our own request log
Page requests recorded in the window: 1,482,547, across 14 days with traffic.
Answered with a server error: 0 — 0% of the requests we recorded.
External probe on the same host: 70 paths checked at 2026-11-20 01:07:31 UTC, 0 confirmed server errors.
This is our own server's log, not an external monitor with a public history. A day without traffic is a day without a reading, and a request that never reached the server is not in it.
What these numbers cannot tell you
- Whether your order will arrive. These are past orders, not a guarantee of the next one.
- Whether the platform keeps the delivery: followers can drop and views can be corrected after they arrive.
- Which supplier is better on a product the sample is thin on: a family with two orders cannot rank two panels.
- Anything about buyers we do not have yet: an empty window prints no rate, and we will not fill it with a number.
The questions this page should answer
How is the delivery rate calculated?
Delivered in full divided by settled orders in the same window. Settled means completed, partial or refunded. A refund stays in the denominator: an order that came back is a delivery that did not happen, not a row to drop.
Do you include your own test orders?
In the per-supplier table, yes, because they are real orders on the same route a customer order takes, and excluding them would make a thin sample look thinner. The count of customer orders and the per-supplier table are labelled separately and never added together.
Why is there no five-star rating or a badge here?
Because there are no published reviews from buyers yet, and a rating with nothing behind it is the exact thing this page exists to replace. When real reviews exist they go to the reviews page, not into a widget on a trust page.
When were these numbers read?
When this page loaded. The stamp at the top is the moment of the read, and every table column is read in the same query. A cached copy is never older than a minute.
What happens if an order never starts?
The order worker refunds it to the wallet in full once the deadline on the refund page passes, without a ticket. The count of orders that rule has refunded is printed above.
Every probe row → The order receipts → The refund rule → The price dataset →
The window closes at the last order we recorded, 2026-11-17. If a figure here and the page it links to disagree, the linked page is the evidence.