Agency Desk service levels
The service-level framework, before the service-level numbers
This page defines how RyanFulfil measures its agency operations: the four clocks that run behind your account, what starts and stops each one, and when each one pauses. Each clock carries a target counted in working days on China time. What it does not publish yet is a credit schedule or an update cadence by severity, because RyanFulfil publishes those only after the founding pilot has actually measured them.
A definition you can plan around is worth more than a number nobody has hit yet. The structure on this page is fixed and you can build a client promise on it today. The targets are in the table below. The credit schedule and the update cadence per severity are being logged over the 90-day founding pilot and will be published here, on this page, once they are real. Delivery transit is not one of those clocks: it stays a route-specific estimate, never a guarantee.
- Four defined operational clocks
- One figure measured and published
- Targets in working days; credits after the pilot
- Delivery stays an estimate, not a guarantee
This expands the response-time note and the onboarding-speed answer summarised on the Agency Desk overview. The seller-side measurement it draws on is on the main FAQ.
The rule that governs this page
Why this page defines the clocks but does not print the hours
The Agency Desk overview answers “how long does onboarding take?” with “it depends, and we would rather say that than quote a number the pilot has not measured.” This page is where that answer is worked out in full, so you can see exactly what is being measured and what each clock commits to.
A framework is a definition. A target is a number.
A framework says what a clock measures — when it starts, when it stops, when it pauses and what makes it urgent. A target says how fast that clock should run. The framework can be honest on day one because it describes how we work. A target can only be honest once it has been measured against real partner orders, in real conditions, including a Chinese New Year. We are publishing the first now and withholding the second until it exists.
A target published before it is measured is a marketing number.
Quoting an SLA the operation has never been held to is the same mistake as a guaranteed delivery date: it reads well in a proposal and fails at exactly the wrong moment. The site’s standing rule is that nothing is promised that has not been tested. Service-level targets sit squarely under that rule, so they wait for the pilot rather than being written to look reassuring.
What you actually get during the pilot
Each of the four clocks below is logged from the day you go live. You get the definition now, so your client promise is built on a real structure rather than a guess. You get the four targets below, each counted in working days on China time, and the one figure already measured. As the credit schedule and the update cadence stabilise against pilot data they are published on this page first, before they appear in any proposal or agreement. When a number arrives here, it is a number we have hit, not one we hope to.
The exception to the no-numbers rule
The one figure we have actually measured
There is exactly one service-level number on this page, because it is the only one that has been measured rather than promised.
Median first reply: 23 minutes, in working hours
Across client requests that arrived inside working hours, the median first reply was 23 minutes. That is the middle of the distribution, not a ceiling, and it is a reply — not a resolution.
The half of that number that matters when you are the one waiting: about one in five in-hours messages still took more than half a day; a message arriving at 10pm China time is answered the next morning, not the same evening; and a first reply is not a resolved problem. The full measurement, and what it does not cover, is on the main FAQ.
Why this is a baseline, not an SLA
A median describes what usually happens; a target commits to what will happen every time, and the gap between them is the one-in-five tail above. This figure is the starting point for two of the clocks below — quote acknowledgement and issue acknowledgement both begin with a first reply — but it is not itself the target for either. The published target for both is one working day, set well above the median on purpose: it is the ceiling we commit to, and the pilot distribution, tail included, is what we measure against it.
A first reply is not the finish line
Acknowledging a quote request in minutes says nothing about how long the quote takes. Acknowledging a lost parcel says nothing about how long the carrier investigation runs. That is precisely why the framework splits acknowledgement from completion into separate clocks, so a fast reply can never stand in for slow work.
The structure
Four clocks, from a brief to a resolved issue
These are the four things a partner actually waits on. Each is measured separately, because bundling them hides the slow one behind the fast one. Read the sequence first, then the definition table below it.
- 1. Quote acknowledgement Starts when you submit a product, route and volume brief. Stops when we confirm receipt and either accept the brief or name exactly what is missing. This is the clock the 23-minute median feeds into. RyanFulfil
- 2. Quote completion Starts when the brief is complete enough to price. Stops when you have the net quote and its written assumptions in hand. It counts only the time we hold a complete brief, which is why a vague brief costs you the pause, not us the clock. RyanFulfil
- 3. Order release to dispatch Starts when an order is released to fulfilment, meaning paid or confirmed, with stock and a valid address. Stops when the parcel is handed to the carrier and a tracking number is returned to the connected store. This is processing time only. Carrier transit after it is a route estimate, not a clock. RyanFulfil
- 4. Issue acknowledgement and next update Starts when you raise a fulfilment issue. Stops, for acknowledgement, when we confirm it and open an investigation, then runs as a committed “next update by” cadence until it is resolved. You still take the first line with the end customer; we return the evidence. RyanFulfil
| Quote acknowledgement | Quote completion | Order release to dispatch | Issue acknowledgement + next update | |
|---|---|---|---|---|
| Start event | You submit a product, route and volume brief. | The brief is complete enough to price. | An order is released to fulfilment: paid or confirmed, with stock and a valid address. | You raise a fulfilment issue through the partner channel. |
| Stop event | We confirm receipt and accept the brief or name what is missing. | You receive the net quote and its written assumptions. | The parcel is handed to the carrier and tracking returns to the store. | We acknowledge and open an investigation, then update on a committed cadence to resolution. |
| Business hours and timezone | Counts working hours only, CST (UTC+8). | Counts working hours only, CST (UTC+8). | Counts working hours only, CST (UTC+8). | Acknowledgement counts working hours, CST (UTC+8); a critical issue is handled outside them. |
| Missing-info pause | Paused until the brief detail we asked for arrives. | Paused while awaiting a spec, variant or approval only you can give. | Paused on a bad address, a stock hold or an unconfirmed payment. | Paused while awaiting evidence, an approval or a decision from you. |
| Peak-season exception | Extended per a declared peak window. | Extended per a declared peak window. | Extended or suspended per a declared peak window. | Acknowledgement holds; resolution stretches with the carriers. |
| Severity applies | No. Standard handling. | No. Expedite only by agreement. | No. Expedite only by agreement. | Yes. Severity sets the update cadence. |
| Credit applies | No. | No. | Possible once the target is set: a controllable clock. | Possible where the cause is controllable, never for a carrier outcome. |
| Published target | Within one working day of the request reaching the desk. Measured so far: median first reply 23 minutes in working hours, August 2026. | Within one working day of a complete brief. | Within one working day when the stock is already in the Guangzhou warehouse. Factory-sourced orders follow the lead time stated in the quote. | Within one working day of the report reaching the desk, with the severity set. The update cadence by severity follows the pilot. |
The issue clock is the seller-side workflow you already know, reached through your account. The status meanings and the evidence we return are documented at the order issue centre, and it is the same material partners work from.
Two conventions that decide every number
Working hours, and the pause that stops the clock
Before any target means anything, two rules have to be fixed: what time counts, and what time does not. Both are the same across all four clocks, so a partner is never guessing which convention a given number was measured under.
Working hours are China hours
Every clock counts working hours on China Standard Time (UTC+8), because that is where the Guangzhou desk and the suppliers are. A request that lands outside working hours starts its clock at the next working-hour boundary. That is the whole reason a 10pm China-time message is answered the next morning rather than the same evening, and it is why the honest half of the 23-minute figure exists. When your client is in another timezone, this is the arithmetic to build their expectation on, not a promise that the desk is awake at their midnight.
The clock pauses when it is waiting on you
Any clock that is waiting on something only you can supply — a missing variant, a sample approval, a corrected address, a confirmed payment — is paused, and resumes when the information arrives. The pause is logged, so when you ask where the time went, the answer is a record rather than an argument. It is not a loophole: we still acknowledge and tell you what is missing. It means a completion target is measured against the time we held a workable brief, not the days a brief sat half-finished.
A worked example, without a single hour in it
You submit a quote brief that names the product and the variants but not the destination country. The quote-acknowledgement clock starts, and stops when we reply that we have it and need the country. The quote-completion clock does not start yet, because the brief is not workable. You send the country the next day; completion starts then. When the quote lands, the completion time on record is the stretch we actually held a complete brief, and the gap while we waited on the country shows as a pause against your side, not a slow quote against ours. Same order, two clocks, no argument about which delay belonged to whom.
What changes when it is urgent
Severity: defined by impact, not by a stopwatch
Severity attaches to the fourth clock, the issue clock. It is defined by what is at stake, not by a promised number of hours, because the honest lever during an incident is how hard we push and how often we come back to you — not a countdown we may not control. The update cadence for each level is one of the targets being measured over the pilot.
| Level | What it covers | A typical example | What tightens | Credit |
|---|---|---|---|---|
| Critical | Safety, compliance, or a stranded or mis-routed shipment where waiting makes the outcome worse. | A shipment held at customs; a product-safety signal on a live line. | Handled outside working hours, on the fastest cadence, and may trigger the emergency direct-contact exception in the partner terms. | Considered where the cause is controllable. |
| Major | A live order or batch blocked, or a defect affecting a run that has already shipped. | A wrong item across a shipped batch; a sustained address-correction backlog. | Prioritised investigation and a committed update cadence until it is resolved. | Considered per the schedule set at pilot. |
| Standard | Routine questions, single-parcel queries and status checks. | A where-is-my-parcel query on one order. | Nothing beyond the normal issue clock. | Not applicable. |
Severity does not move the client-contact line
A critical severity can let us act fast; it does not let us go around you. The one place direct contact with your client is permitted is a defined emergency — a safety, compliance or stranded-shipment case where waiting would make things worse — and that exception is written into the partner terms rather than decided in the moment. The full set of controls is on the client-protection page.
The exception written down in advance
The peak-season and Chinese New Year exception
Chinese New Year is the single most predictable way a first year of fulfilment goes wrong, and the honest response is to plan for it rather than absorb it. So it is an exception in the framework, declared before it starts, not discovered in February.
What actually moves
Factory closures and carrier congestion move the whole timeline, not one clock. During a declared peak window the dispatch clock is extended or suspended, and the two quote clocks are extended, per the service policy. The issue clock keeps its acknowledgement but resolution stretches with the carriers, because a congested network is outside anyone’s control and pretending otherwise is how partners get blindsided.
How you hear about it
A peak window is declared in advance through the partner operations channel, with the dates and what changes, so your client-facing checkout copy and delivery promises can be adjusted before the orders come in rather than after they are late. The planning material — cut-off timing, pre-stock decisions, how to word the change to a client — is part of the partner kit built with the founding cohort.
Delivery estimates already widen at peak
Separate from these clocks, the route transit estimates on the maintained destination table widen during peak, and a quote reflects the window for the route at the time it is issued. Delivery was never a clock and never a guarantee, so peak does not break a promise there — it changes an estimate, which is exactly what an estimate is for. The per-route working ranges are on the shipping-times page, and what each carrier scan means is on tracking status.
When a miss costs us, and when it cannot
Where a credit applies, and where it honestly cannot
A service credit only means something if it attaches to a clock we control. The rule for which clocks can carry one is fixed now; the schedule — what triggers a credit and how large it is — is set with the founding cohort and published here once the targets it hangs on exist.
A credit can apply
Against a controllable clock, once its target is measured and set.
- Order release to dispatchOur processing time is the clearest controllable case.
- A controllable issue-clock missWhere the cause sat with our side, not the carrier’s.
- A credit is taken against the fulfilment feeNot a payout, and not a promise about the client’s revenue.
A credit cannot apply
By definition, not as a loophole.
- Carrier transit and delivery dateAn estimate, never a guarantee. Paying a credit on it would be pretending we control it.
- Any clock while it is paused for missing infoThe clock was waiting on your side, so it was not running.
- Anything inside a declared peak windowThe exception is announced in advance for exactly this reason.
- Sales, product-market fit or profitOutside the service entirely, on every path.
Why the amount is not on this page
A credit schedule is only meaningful once the target it protects has been measured. Publishing a credit against a target we have not set would be inventing both halves of a promise. So the trigger rule is here, honestly, and the figures follow the pilot — on this page, before they appear in any agreement.
Where this sits against the paperwork
The framework here, the schedule in your agreement
This page is the working framework, and it is the one to plan and sell from. It is written to match what you sign, not to soften it.
What governs if the two ever differ
The binding service-level schedule lives in the agency service policy attached to your partner agreement. Where this page and the signed agreement ever differ, the signed agreement governs and this page is the copy that is wrong. Tell us, and we will fix the page — we would rather the public version be right than convenient.
How the numbers arrive
The targets above are published. As the update cadence and the credit schedule stabilise over the 90-day pilot they are published on this page first. The definitions do not change when those numbers land; the clocks you plan around today are the clocks the targets will be measured against. Nothing here is promised that has not been tested, which is the same rule the rest of the site runs on.
Founding partner pilot
Build your client promise on the clocks, not on a hopeful number.
Apply with your services, client count, target markets and expected volume. We will map your first account onto these clocks, tell you which path fits, and be straight about what the pilot has and has not measured yet.
Delivery windows are route-specific estimates, not guarantees, and every service-level target is counted in working days on China time; the credit schedule is published only after the founding pilot measures it. The one figure measured to date is a 23-minute median first reply in working hours, with the honest tail set out above.