Yunda, STO and YTO Split on September Pricing as Parcel ASPs Move Opposite Ways

The Analysis

Wednesday, 16:20, and the roller shutter is halfway down. This is the picture I cannot get out of my head this month. A guy shipping auto-parts samples out of Yiwu has sixty parcels stacked against the wall, maybe 0.8 kilogram each, a few hundred RMB of declared value apiece, all of them due at an export consolidation warehouse in Ningbo before a Friday 14:00 booking cut. Miss the warehouse and the parcels miss the container. Miss the container and the purchase order behind them slides a week, and the buyer in Hamburg starts asking whether the factory is serious. He is not asking me about rates. What he asked, standing there in a doorway in the October heat, was whether Thursday's pickup is going to behave the way September's pickup did. That is the honest question buried inside this month's express numbers, and almost nobody writing about them is asking it.

Here is what the three September operating releases actually said. Yunda put revenue up 7.88 percent at RMB 4.587 billion, parcel volume up 1.37 percent at 2.139 billion pieces, and revenue per parcel up 5.94 percent at RMB 2.14. STO logged revenue up 18.05 percent at RMB 5.469 billion, volume up 15.43 percent at 2.524 billion, revenue per parcel up 2.36 percent at RMB 2.17. YTO reported revenue up 5.86 percent at RMB 6.138 billion, volume up 12.33 percent at 2.951 billion, and revenue per parcel down 5.76 percent at RMB 2.08. Three companies, one month, one approaching peak season, and per-parcel revenue moving plus 5.94, plus 2.36 and minus 5.76.

Set those three single-parcel lines side by side and something shows up that no single company's release carries: this is the first time going into a peak that the large Tongda networks have pushed that number in genuinely opposite directions. Every commentary I have seen draws the easy conclusion, which is that the price war has ended and discipline is back. I read the trio differently, and so should anybody with parcels by the door. My job is not to explain an index. My job is to know what happens to sixty cartons between 16:20 and tomorrow's cutoff, and the clock does not care about anybody's excuses.

Start with the arithmetic nobody prints underneath their own headline. If Yunda's per-parcel revenue rose 5.94 percent to land at RMB 2.14, then on the same basis a year earlier it sat near 2.14 divided by 1.0594, call it RMB 2.02. Run the same step backwards on the other two. STO's 2.17 divided by 1.0236 is about RMB 2.12. YTO's 2.08 divided by 0.9424 is about RMB 2.21. A year ago the ladder read YTO on top near 2.21, STO in the middle near 2.12, Yunda at the bottom near 2.02. This September the ladder reads STO 2.17, Yunda 2.14, YTO 2.08. The top rung and the bottom rung traded places. Yunda added roughly twelve fen a parcel, STO added roughly five, and YTO handed back roughly thirteen. None of that is in the releases. It takes sixty seconds with a phone calculator and it changes what you do with the number.

Then look at the spread rather than the arrows, because the spread tells a second story that fights the first one. Twelve months ago the distance between the dearest of the three and the cheapest was nineteen fen, about 9.4 percent of the cheapest ticket. This September it is nine fen, about 4.3 percent. The networks are not diverging into two camps. They are converging on a band around RMB 2.1 from three different starting points, and they arrived there in September, on the eve of the single largest shipping week of the Chinese calendar. If you price your fourth quarter off the divergence you are pricing off a gap that has already halved. If you price off convergence you are pricing off a band that may not survive mid-November. That is the bind, and it is why the national average is useless to the person standing by the roller shutter.

Now the part I would bet money on being misread, and here I am going to be blunt about it. Revenue per parcel is total revenue divided by parcel count. It is not the unit price printed on your own contract, and it never has been. Two variables move that ratio and only one of them is what you pay. Suppose a network's September book contained heavier parcels than its September last year, meaning fewer sub-half-kilo e-commerce envelopes and more pieces in the one-to-three kilogram bracket, with not one cent of list price changed anywhere.

Run it with explicit assumptions so you can check my arithmetic: assume last September the mix was 70 percent light pieces billed at RMB 1.80 each and 30 percent heavier pieces billed at RMB 2.60 each, which blends to 0.7 times 1.80 plus 0.3 times 2.60, or RMB 2.04. Assume this September the mix drifted to 60 and 40 while both prices held exactly where they were. The new blend is 0.6 times 1.80 plus 0.4 times 2.60, or RMB 2.12.

That is a 3.9 percent rise in reported revenue per parcel produced entirely by the weight of the boxes, with an identical price card. Yunda's move was 5.94 percent. STO's was 2.36 percent. A meaningful slice of what the market is reading as a return of pricing discipline could be nothing more than a shift in what went through the scanners.

Which one is it. That question, not the headline, is what decides how you negotiate next week, and there are three mechanisms that could each produce this pattern. One is genuine list-price movement, where a network with flat volume chooses to defend yield. Yunda fits that description on its own numbers, since holding per-parcel revenue up nearly 6 percent while adding only 1.37 percent in parcels is the profile of somebody choosing price over share. A second is mix, described above, which can lift the reported ratio without touching a single rate card.

STO is the interesting case here, growing parcels 15.43 percent while per-parcel revenue rose 2.36 percent: adding a third of a billion pieces in a month usually drags the average down unless something in the product mix pulled it up, so I would not assume that 2.36 percent landed on anybody's invoice. The third is straight share buying. YTO pushed volume up 12.33 percent and let per-parcel revenue fall 5.76 percent, and I have no difficulty believing that number is real money off the table, because cutting yield that hard while growing double digits is exactly what buying season capacity looks like. All three can be true at once for three different companies, which is why "the price war is over" is a sentence nobody with parcels should repeat to their boss without checking their own settlement file first.

Let me put real money against it, because feelings do not survive a budget meeting. Assume an exporter shipping 30,000 parcels a month at an average 0.8 kilogram, currently settled at RMB 2.05 each, which is inside the band I described. Assume that in the Q4 negotiation the account moves with Yunda's reported direction, up 5.94 percent, landing near RMB 2.17. That is RMB 0.12 a parcel, or RMB 3,600 a month, or RMB 43,200 across twelve months. Assume instead it moves toward YTO's direction, down 5.76 percent to about RMB 1.93, saving RMB 3,600 a month. The swing between the two worlds is RMB 7,200 a month on 30,000 parcels.

Scale it and it stops being noise: assume 120,000 parcels a month and the same two scenarios produce a spread of RMB 28,800 a month, RMB 345,600 a year. At 500 parcels a month the entire question is worth RMB 60 a month and you should stop reading and go pack boxes. That is the honest layering. What a network-wide average per-parcel number is worth to you scales with your own volume, and the read is very different for a factory shipping samples than for an e-commerce seller shipping orders.

The layer nobody quantifies is exposure to the nine-fen spread by lane rather than by network. The reported figures blend every province each network touches, including provinces where the price floor has been held for years and provinces where it never existed. The number that touches you is set at the branch that picks you up, against the corridor you ship on. Ship out of Yiwu toward Ningbo, or Guangzhou toward Shenzhen, or Hangzhou toward Shanghai, and you are in the corridors where discounting is local, personal, and driven by whoever owns the trucks on that street.

In those corridors it is entirely possible for your effective price to fall in the same month your network's reported average rises, because the national average went up in three provinces you do not ship from. This is why the instruction everyone repeats about the fourth quarter, negotiate corridors rather than the national average, is correct but incomplete: negotiate corridors, and negotiate corridor weight brackets, because the bracket is where the mix effect lives.

Timing is where this stops being interesting and starts being urgent. Here is the schedule as I would run it for the man at the door, every figure an assumption you can replace with your own scanner data. Assume a pickup cutoff at 17:30 on day zero. Assume the branch collects and consolidates until about 18:30. Assume outbound sorting runs 19:00 to 20:30, the trunk truck departs at 21:00, the destination hub sorts between 02:00 and 03:30, the receiving branch gets the load at 05:00, delivery loads out at 08:00, and the parcels are delivered between 12:00 and 17:00 on day one. Total elapsed time, roughly twenty to twenty-four hours door to door. Now assume peak doubles the daily volume at that branch from 4,000 pieces to 8,000 pieces while sorting headcount and machine capacity stay where they are.

I am not going to pretend I can prove the next numbers, but the shape is not in doubt: assume outbound sorting runs sixty minutes long, assume the trunk departure slips thirty minutes waiting on the last cages, assume the destination sort backlog adds ninety minutes in the small hours, assume delivery therefore loads out at 09:30 instead of 08:00 and finishes at 20:30 instead of 17:00. You are now roughly three and a half hours later than September, on the same lane, with the same promise outstanding.

Which hour breaks first. Not the highway. Highways do not fill up, they just run late. What breaks first is the ninety-minute window between outbound sorting and trunk departure, because it is the only link in that chain with a hard gate: anything not sorted by 21:00 waits for the next departure, and the next departure might be two hours away or four. A parcel checked in at 20:55 and one checked in at 21:05 do not arrive two hours apart in your customer's hands. They can arrive on different days. Everything downstream inherits that jump, with no exception event anywhere in the tracking feed, because a missed departure is not an exception, it is just a late scan.

This is the single most useful hour in the whole week to understand, and it is why I tell every account manager to treat the pickup cutoff as a promise to themselves rather than a suggestion from the carrier. Assume a branch whose effective cutoff slides from 17:30 to 16:45 in the last week of October, which is normal, and your whole day moves forty-five minutes earlier with it.

Now the question that matters for the sixty parcels. Does any of this actually break his promise. His promise was seventy-two hours to the warehouse, made on Wednesday afternoon against a Friday 14:00 booking cut. A three-and-a-half-hour slip is survivable, nine hours is survivable, and even a full day is survivable if nothing else goes wrong. What kills him is not hours at all. What kills him is the departure count: if the parcel misses the last trunk departure of a day it loses near enough twenty-four hours, and in peak there is no spare truck to catch up with. So the correct reading is that hours decide whether today's shipment goes today, and days decide whether you breach. Any remedy aimed at shaving minutes is wasted effort next to anything that protects a departure.

Triage follows directly, and this is where everyone wastes money. Assume those sixty parcels split into three groups. Group A is twelve pieces that must be inside the warehouse before Friday 14:00 because they back a purchase order worth more than USD 30,000. Group B is twenty-six pieces due Monday. Group C is twenty-two replacement samples with no binding deadline. Short haulage capacity does not mean all sixty are equally at risk, and treating them equally is how you spend eighteen RMB protecting something worth eighteen RMB. Assume that moving group A onto the earliest possible departure costs RMB 1.5 more a parcel, say RMB 18 for the twelve pieces.

Assume the consequence of missing that purchase order is one emergency air re-send at RMB 4,800, plus a buyer who now asks for air freight on future samples. Eighteen against forty-eight hundred is not a decision, it is a reflex. Meanwhile group C can sit two extra days for free, and if you are paying to expedite group C in the last week of October you have lost the plot completely. This is what "get the parcel back first" means in practice: rank by deadline proximity multiplied by what is at stake, then spend only on the top slice.

The counterargument deserves its hour in court, because this fork may not be a fork at all. If September was a settlement-rhythm artifact, a mix shift, or the mechanical result of which networks booked which kinds of parcels that month, then the three numbers will partly reverse in the October releases and anybody who rebuilt a Q4 contract around them will have locked the wrong direction. The condition that would flip my read is narrow and testable. Take your own twelve weeks of invoices. Compute blended price as total billed freight divided by total parcels, and compute average parcel weight over the same weeks, then compare the two series.

If weight held flat and blended price moved with the network's reported direction, you are looking at a real price change and you should negotiate on that basis. If weight moved and price held, you are looking at your own mix, and the headline belongs to somebody else's business. If both moved, you have a story nobody else has, and you should bring it to the table before your account manager brings theirs.

That brings me to the trap everyone walks into in the last two weeks of October, which is switching networks to chase the cheaper number. Assume a mid-size account needs three to five days to move: new waybills printed, system integration tested, pickup route reassigned, and exceptions handed to people who do not know the account yet. Assume there is a real on-time dip during that window, maybe a few points on your delivery-success rate for a week.

Now put those five days anywhere between late October and the week after Double Eleven and you have scheduled that dip to land exactly on your most valuable shipping days. If the numbers say switch, switch in December, and bridge November with volume splits rather than new contracts. A second trap sits inside the rebate tier rather than the headline rate: many schedules pay back per piece against a monthly volume band, and losing one tier can cost several fen a piece, which is more than the nine-fen spread you were chasing.

Assume you sit in the second of three tiers and a soft month drops you to the first at a five-fen penalty; on 30,000 parcels that is RMB 1,500, and it quietly eats most of the discount you just negotiated. Pull your own contract, find the tier table, and put your real monthly minimum into it before anybody quotes you a rate.

The third trap is choosing the cheapest network when the cheapest network is also the one growing fastest. STO added 15.43 percent in parcels during September, which is more than double anybody else's growth. Additional flow lands on concrete: sorting floors, dock doors, and trunks on specific corridors, and the corridors that absorb growth first are the ones where transit time slips first. That is not a reason to avoid anybody. It is a reason to ask one narrow question before you sign anything: what was your transit time, in hours, on my top three destination provinces, in the last two weeks of October, measured from scan-in to delivered. Ask it of all three networks. Whoever answers with a number rather than a brochure is whoever deserves the volume, and this is exactly the kind of cost-to-serve question Grace Yang picks apart better than I do. My lane is the clock, and the clock says you have about ten days before pretending nothing happened stops being an option.

Tonight, before anything else, three things. Pull every parcel due against a customer deadline in the next fourteen days and split them by departure date rather than by pickup date, so you can see which ones have only one truck left in their life. Move your own internal cutoff sixty minutes earlier than the carrier's published one for the rest of October, and tell your floor why, because the reasons I hear for missing a cutoff are always the same and always useless. And write down the two numbers that will tell you whether this month's split was price or mix: your own blended billed price per parcel, and your own average parcel weight, week by week, for twelve weeks. Everything else follows from those two columns and you can do it before Friday.

↑ Back to the key points

— By Nadia Pryce