No Golden September: Comparing China's Carmaker KPI Gaps, the Trade-Offs and What to Do Next

By James Hartley, IT Insights — sector desk

Verdict before numbers. In 2026 the seasonal engine that the Chinese car industry has cranked for a decade stopped turning, and the question "who is on track?" changed meaning altogether. A completion percentage now reports three separate things at once: how the target was written, how much of the volume left the country, and whether the cars were sold or merely parked at dealers. Read that way, October's scorecards are not a ranking of effort — they are a diagnostic of which business models still convert units into cash.

A line circulating inside the industry says it more efficiently than any spreadsheet: "The peak season didn't peak, and that is the most dangerous part." The reason the fear is arithmetic rather than mood becomes obvious once the calendar is laid out. Carmaker sales plans were designed around one pull-through in September and October plus a final sprint at year-end; when the first wave never grew, the base for the second wave vanished with it. With two months left, the loss is no longer recoverable by trying harder.

Roughly 1.6 million passenger cars were retailed in October, down 1.1% month on month. September itself came in at 1.618 million, 28% below the year-ago month, and for two consecutive months the annual decline has hovered around 28%. The CPCA — the passenger-car committee housed inside China's automobile dealers' association — forecasts full-year 2026 retail of 18.5 million units, a 22% fall and the first double-digit annual decline the Chinese passenger-car market has ever recorded. As one chart of the same period, previously compiled by Chaodian Lab, showed, the pressure is concentrated in exactly the two months that were supposed to pay the bills.

The decline is also uneven, which matters for how you should read every figure below. CPCA data published in October put new-energy retail penetration at 67.0% in September, an all-time high, with wholesale penetration at 65.4%, while fuel-only retail fell 43% year on year. The association's weekly tracker for 1–25 October showed 1.145 million units retailed, down 26%, and year-to-date retail of 14.479 million, down 22%. One drivetrain is being drained; the other has simply stopped growing fast.

Scoreboard: turn every shortfall into a monthly run-rate

Percentage completion is the least useful number on the page. The useful one is the monthly rate a company must now hold for the remaining two months, compared against the best month it has ever produced. Only Zeekr clears the industry's own rule of thumb — reaching 80% or more of the annual figure within the first 10 months, so that the last two months leave room to manoeuvre. This year it is the only maker able to close its books early: 320,887 units between January and October against a 300,000 target, 107% done.

Carmaker	Jan–Oct 2026	Annual target	Still to find	Required per month, Nov + Dec	Reality check
Zeekr	320,887	300,000	—	—	107%; the only account that can be closed early
Leapmotor	77.27% complete (second place)	1,000,000	227,300	113,700	Its October record was 106,127 — the required rate is above its best ever month
Geely	2,545,700 (73.8%)	3,450,000	904,300	452,000	Only just broke 310,000 in a month for the first time; a near-46% jump is needed
Chery Group	78.2% complete	—	—	—	Most respectable of the traditional makers, but exports were more than two-thirds of October volume
BYD	490,200 sold in October alone, +11% year on year	5,000,000	1,378,300	689,000	October exports 170,000, more than double last year
SAIC	3,577,100	5,000,000	1,422,900	711,000	The largest absolute hole on the board
Great Wall Motor	1,040,700	1,800,000	759,300	379,700	October sales were 120,400

Completion percentages were compiled by Chaodian Lab; the run-rate column is IT Insights' own division of the same published numbers. Doing that division once is enough to see the result: the target line is not within range for almost anyone. October retail fell close to 30% against a year earlier, which is the first time "Golden September, Silver October" has disappeared in this particular way — and with the calendar holding only two more months, finishing the KPI is a distant dream for most companies. Missing a target, in other words, is not shameful this year; it is the statistical mean.

Under the headline numbers: real orders versus inherited momentum

The new-force column splits the same way once you ask where each unit went.

• XPeng: 50,237 deliveries in October, a monthly record, of which 13,326 went overseas — up 166%. The growth came from a new export space, not from clawing share out of a shrinking domestic pool.
• Xiaomi: more than 50,000 units after 30,000 in August and 40,000 in September — three steps in three months. The first 50,000 month came from the Pengcheng series' first full delivery month, not from a year-end push.
• Leapmotor now sits at TOP 1 among the new-force makers. Of the six companies that were still brawling in the same tier last year, the leader this month is 3.7 times the size of the one in last place.
• On the other side: NIO at 34,246, Li Auto at 29,025, and the Harmony Intelligent Mobility Alliance at 33,537 — down 50.8% year on year, and now falling for a fifth consecutive month.
• Legacy joint-venture brands are drifting down with it: Dongfeng Honda slipped again, GAC Toyota lost nearly 25%, and Japanese brands' combined share of China has already fallen below 11%.

Why the maths was never going to close

Look back at the target list published at the start of 2026 and it reads like a memo written by an optimist: SAIC 5 million, BYD 5 million, Changan 3.3 million, Leapmotor 1 million, Harmony Intelligent Mobility Alliance 1 million to 1.3 million. In a year whose retail total is forecast to fall by a double digit, most of those numbers embed growth in the tens of percent, in several cases 30% or more.

The manufacturers were not blind to a falling market. An industry analyst's judgement is that the difficulty was well understood, but the anchor was never terminal demand — it was not falling behind. Once penetration passes half and incremental volume turns into a fight over stock, the target stops being a navigation device and becomes a posture addressed to capital markets and suppliers. "Setting a KPI" itself, not any particular number, is what failed in 2026.

• Mainstream targets were set in the band "grow 20% to 50% on 2025" — the standard playbook of the past decade, when the market was rising and a stretch goal was a way to seize share.
• The market went the other way, down 22%. Between a target built on +20% growth and a market shrinking 22% sits a gap of 40 to 70 percentage points.
• Work one example through: a company that sold 100,000 vehicles in 2025 and sets 120,000 for 2026 while the market falls from about 23.7 million to 18.5 million must lift its share from 4.2% to 6.5% — a 55% improvement.
• That is not something "sell a bit more" solves. It means prising an extra 55% of share out of competitors' hands with only the fourth quarter's two months on the clock. Renaming the target now cannot fill the hole: the denominator of the target and the numerator of the market are moving in opposite directions.

The three tools for forcing a number were taken away

1. The tax lever. On 1 January 2026 the new-energy purchase-tax exemption switched to a half-rate collection, with the per-vehicle tax cut capped at RMB 15,000 instead of RMB 30,000. State tax-bureau guidance confirms the halving runs to the end of 2027 and — new for 2026 — that plug-in and range-extended models must clear a 100 km electric range to qualify, with the relief tied to the date on the sales invoice. The late-2025 rush that this deadline triggered pulled early-2026 demand forward, and the usual year-end tailwind never showed.
2. The subsidy lever. The national trade-in scheme moved from flat-rate vouchers to a percentage of the new car's price, with the top case (scrapping an old car for a new EV) capped at RMB 20,000. The marginal push is weaker, and it is now conditional in ways that trip buyers up.
3. The price lever. In February the automotive pricing-compliance guidelines took effect, defining below-cost selling as a major compliance risk. Finishing a KPI used to have one last resort — cut price, dump inventory into the channel, book the wholesale number. In 2026 that resort costs more than most players can pay.

The channel answered first. New cars now carry an average gross margin of -21.4%, nearly 80% of dealers sell below their own purchase price, and the inventory coefficient stood at 1.58 — which, against the dealers' association's own scale where 0.8 to 1.2 is healthy and anything above 1.5 is an alert level, is a red light rather than a statistic. More than 1,800 authorised dealer outlets closed or handed back their franchises in the first half of 2026, an average of 10 a day, the fastest network retreat in nearly five years. In that condition, forcing the wholesale number up in the last two months destroys dealer cash flow and whatever brand value is left.

The income statement says the same thing: the industry's sales profit margin was 3.3% for January to September 2026, and only 1.4% in September alone, the weakest reading in nearly four years. Changan earns RMB 619 to 811 per vehicle; Leapmotor earns RMB 589. Selling one car now yields less than a decent meal.

Volume, profit or export: three target models side by side

Model	Who is running it	What it optimises	What it trades away	Where it works	Where it breaks
Volume-first, target unchanged	SAIC (5m), BYD (5m), Changan (3.3m), Harmony Intelligent Mobility Alliance (1m–1.3m)	Rank, scale, share narrative	Channel cash, price discipline, residual value	Expanding markets where one push moves the whole curve	A market falling 22% with an inventory coefficient above the alert line
Profit-first reset	Great Wall Motor; Dongfeng	Earnings and cash	League-table optics	Shrinking market, 3.3% sector margin	Only credible if the restated number is actually enforced
Export-led second curve	Chery Group, BYD, Geely, XPeng	Volume outside the domestic squeeze	Dependence on tariffs, freight, localisation	Brands that already have overseas channels	When export share is used to disguise domestic weakness
Cost-and-rhythm leader	Leapmotor, Xiaomi	Real orders at controlled cost	Thin per-vehicle profit (RMB 589)	Price-sensitive mass segments	If a model transition misses by one quarter

Great Wall cut its sales assessment target from 2.49 million units to 1.8 million, while leaving its net-profit bar of no less than RMB 10 billion untouched. Dongfeng reset its full-year number from 3.25 million to 2.6 million. These moves are usually read as surrender. The more accurate reading: when volume no longer converts into profit and cash flow, keeping a sales KPI at the centre of management is a failure of governance. Sacrificing units to protect the earnings line is a choice, and in 2026 a rational one.

The flattering completion rates deserve the same separation of water from substance. Leapmotor's 77.27% rests on extreme cost control and product cadence: RMB 589 a car, but real orders for real products, not wholesale inflated by channel stuffing. Chery's 78.2% looks strong until you open the structure — October exports of 201,000 units, the third time within the year that a single month cleared 200,000, and exports accounting for the great majority of its volume. BYD shipped 170,000 abroad, up 103%, and Geely 107,000, up 158%. At industry level, customs-linked data cited by the commerce ministry's trade-remedies portal showed China's vehicle exports rising 53.8% by volume in January to May 2026, with new-energy exports up 78.5% and making 71.6% of the total. That growth is not a by-product of hitting targets; the second curve is underwriting the domestic grind.

Which is the real dividing line in the January–October completion table: "who is still trading real products for real demand" versus "who is still propping numbers up with inertia".

Five misconceptions worth discarding

1. "A low completion rate means a lazy team." It usually means a target set 40 to 70 percentage points above the market's direction. Only one maker crossed 80% after ten months.
2. "Missing the target is bad news for buyers." It can be bad news for service: with about 10 dealer outlets closing a day in the first half of 2026, continuity of warranty and maintenance is a purchase criterion, not an afterthought.
3. "End-of-year discounts are pure savings." The dealers' association and Jingzhengu's monthly residual-value work puts three-year retention for new-energy cars around 43.35%, against roughly 46% for fuel cars — the discount you receive today is partly prepaid out of resale value tomorrow.
4. "Exports are a footnote." For one major group they were more than two-thirds of a single month.
5. "Cutting the target is admitting defeat." Two of the largest groups restated numbers while holding profit floors. That is discipline, not retreat.

What to actually do: three short checklists

If you are buying before year-end

1. Compare the two subsidy routes before you decide. Scrapping and re-buying pays a higher percentage than a private trade-in but excludes scrapping a new-energy car to buy a petrol one, and only one benefit can be claimed per person in 2026.
2. Check the model is on the tax-relief catalogue and — for plug-in or range-extended cars — clears the 100 km electric-range bar; relief follows the invoice date, not the order date.
3. Discount room is real but so is dealer distress: ask about delivery timing, and prefer a dealer with a service history you can transfer.
4. Price the car on three-year resale, not on sticker saving. A steep markdown on a fast-iterating EV is a shorter holding period, not a bargain.
5. Ignore the "last days of the season" framing. There is no scarcity left in the channel; you are negotiating against an inventory coefficient of 1.58.

If you are judging a company

• Convert the residual gap into a required monthly run-rate and compare it with that maker's best-ever month.
• Split domestic from export volume before you praise a completion rate.
• Look for a restated target paired with an unchanged profit condition — that combination signals management that is steering, not performing.
• Treat a wholesale-to-retail gap and an inventory coefficient above 1.5 as a quality-of-earnings warning.

If you run a dealership

• Defend cash conversion over month-end wholesale volume; accepting a stock dump to keep a rebate can cost more than the rebate pays.
• Push trade-in leads first — the subsidy is percentage-based now, so the higher-value qualifying case is the one worth chasing.
• Track your own inventory depth against the 1.2 healthy ceiling rather than against the network average.

FAQ

Did "Golden September, Silver October" actually die?

What died is its function as a scoring event. With roughly 1.6 million units retailed in October, down 1.1% on September, and both September and October running about 28% below last year, the season no longer delivers the surge that targets were priced on. Demand moved into a steadier, more promotional rhythm; the calendar window lost its scarcity.

Will makers cut prices to close the gap anyway?

Less room than before. Below-cost sales are a defined compliance risk since February, average new-car gross margin is -21.4%, and September's sector profit of 1.4% leaves nothing to spend. The realistic version of a "year-end push" is bundled finance and trade-in support, not a headline price cut.

Which names look genuinely solid rather than lucky?

Zeekr, because 107% against a 300,000 plan is the only early close this year. Leapmotor, because its completion came from product cadence and cost control at RMB 589 a vehicle. Great Wall, because it protected a net-profit floor of RMB 10 billion instead of defending a unit number. Each is a different answer, which is the point.

Is the 18.5 million forecast a reason to delay a purchase?

It is a reason to price differently, not to wait indefinitely. A market falling to 18.5 million units means supply is plentiful and negotiation is easier, but the same fall is what erodes residual values and squeezes dealers. The CPCA's own split — penetration at a record 67.0% while fuel-car retail drops 43% — says the used-market risk is concentrated in whichever drivetrain you are buying.

Does this change anything for carmakers outside China?

Read it as an early warning. Two of the three tools that historically rescued a soft year — tax relief and discretionary price cuts — were withdrawn at the same time, and the season-based target model was the first casualty. Any market that reaches penetration past half and still plans volume on "one surge plus a year-end sprint" is using the same map.

The overdue correction

The signal from 2026 is not "set smaller numbers". When most of an industry cannot reach what it wrote in January, the problem is that target-setting has detached from operating reality. With sector margin at 3.3%, cash decides survival long before rank does. Take the sales number down from the pedestal and put profit, cash flow, value per vehicle and real demand back in its place. That is not regression — it is releasing the industry from its worship of scale, and starting a correction that is overdue.

Data corrections or tips for the IT Insights sector desk: contact@itinsights.com.