China-UK Freight Rates Have Swung From +75% to -19.5% in Three Months. Stop Building Q4 Costs on Last Week's Number

China-UK freight rates have moved almost 100 percentage points in three months. They went up 75% in June, up a further 36% in July, then fell 14-19.5% in August. A 40ft container that cost around $1,440 in May was running over $3,000 by July, and is now sitting between roughly $4,095 and $5,005 as of September.

The short answer: if your Q4 landed cost is built on a freight quote from more than a few weeks ago, it is already wrong. Rates on this lane have not settled into a new normal. They have moved three times in three months, in both directions, and the sensible response is not to guess the next move. It is to build a landed cost model that gets rechecked every time a shipment is booked, not once during budgeting and then filed away. Brands who treat freight as a fixed line in their P&L are the ones who get caught out. This matters most right now, because Q4 stock builds are exactly when a stale number does the most damage.

What Has Actually Happened to China-UK Freight Rates Since June

The pattern is worth setting out in full, because no single reading in it tells the real story.

• June 2026: Sea freight rates rose approximately 75% month on month. A 40ft container running around $1,440 in May was quoted at $2,520 to $3,080 by late June, with Southampton capacity already tightening.
• July 2026: Rates climbed a further 36% on top of that. Surcharges intensified and FCL capacity on the main UK-bound services stayed tight through the month.
• August 2026: Rates fell 14-19.5% as the July surcharge period ended. The Container Freight Index (CCFI) was down 33.2% year on year over the same period, a sign that the broader trend is softer than the summer spike suggested.
• September 2026: Current quotes are running $4,095 to $5,005 for a 40ft container and $2,615 to $3,196 for a 20ft container, described by forwarders as unchanged from August. That is still a wide band for the same lane, and forwarders are recommending brands treat any quote as valid for two to three weeks, not longer.

We wrote about the June spike at the time, when the 75% jump first hit brands building EU inventory. The number in that piece is now three readings out of date, which is exactly the point of writing this one differently.

Why This Is Not Really a Freight Story

Freight is the most visible line because it moves in headline-grabbing jumps. It is not the only cost line doing this, and brands that only watch freight are still exposed everywhere else.

Amazon referral fees are a good example, and they move in both directions. Most launch models assume a flat rate. On one power-tool launch, a flat 7% assumption came in at an actual 15% once the correct category classification was applied, and the gap absorbed most of the benefit of a higher-priced bundle. But a flat assumption misses reductions just as often as it misses increases. Several categories have been recently discounted to 7% for example. A brand that never checks which tier its own category sits in can be planning for the wrong rate in either direction, not just the expensive one.

3PL invoices behave the same way. A brand will quote its "fulfilment cost" off the pick-and-ship line on the rate card, then get a full invoice that also carries storage, FBA inbound transfer, fuel surcharge and integration fees. On one engagement, the headline pick-and-ship line was £293.28. The full invoice was £559.55, roughly 45% higher once every real charge was included.

Neither of these is a freight problem. Both are the same failure: pricing a cost once, from a partial or dated number, and not checking it again. Our piece on where Amazon launch P&L costs actually hide covers the referral fee and 3PL gap in more detail.

There is a currency layer underneath the freight number too. China-UK freight is quoted in US dollars. Even if the dollar rate itself holds still, a move in GBP/USD changes what that rate costs a UK-based brand once it is converted, in either direction, completely separately from anything happening on the shipping lane itself. A brand only watching the freight quote is still exposed on the exchange rate behind it.

What Landed Cost Monitoring Actually Looks Like

Monitoring does not mean predicting the next freight move. Nobody can do that reliably, and a forecast dressed up as a plan is still a guess.

What works in practice is a landed cost model that gets rechecked at fixed trigger points rather than once a year:

• Every time a shipment is booked. The freight line in your model should reflect the quote you actually received for that shipment, not the quote from the last one.
• Every quarter, at minimum, for the other moving lines. Referral fee category classifications, 3PL rate cards and duty rates do not move as visibly as freight, but they do move, and a quarterly check catches drift before it compounds.
• Before every major PO run. If a purchase order is being placed for Q4 or beyond, the landed cost behind it should be built from current numbers, not the number used for the last order.
• With a currency view, not just a rate view. If freight and factory costs are dollar-denominated, tracking the freight quote alone only covers half the exposure. A simple hedging strategy, such as a forward contract that locks the exchange rate for a known shipment schedule, protects the plan from a currency swing instead of reacting to it after the invoice lands.

This is a discipline, not a system. It costs almost nothing to run and it is the difference between a Q4 plan built on evidence and one built on a number that happened to be true in June.

What a Missed Number Costs

The clearest example of what happens when a moving input gets treated as fixed is not a freight story at all. It is a timing failure.

On a homeware brand we worked with, purchase orders that needed to be placed before Chinese New Year factory closures were not raised in time. Lead times moved against the brand once factories reopened, and priority ranges that should have landed in June or July slipped to mid-September, with the rest pushed to the end of October. The run-rate exposure was around 7,650 units, approximately £249,000 in retail billing and approximately £209,000 in net revenue, before a separate fixed retailer order and a spot deal were even factored in.

That was not a bad forecast. It was a fixed date treated as flexible, the same mistake as treating a moving freight number as settled. November and December carry the majority of second-half demand for a business like this, which is exactly why a slipped October landing can still save the year and a slipped December one cannot.

Frequently Asked Questions

How often should a brand recheck its landed cost model?

Recheck the freight line every time a shipment is booked, since that is the most volatile input. Recheck referral fees, 3PL rate cards and duty rates at least quarterly. Anything feeding a Q4 or peak-season PO should be rebuilt from current numbers immediately before the order is placed, not carried over from the last one.

Why do China-UK freight rates swing so much in a few months?

Capacity on the China-UK lane tightens and loosens with surcharge periods, seasonal demand and carrier scheduling changes, and a relatively small shift in available capacity can move rates by double digits within weeks. The August fall of 14-19.5% followed directly from the July surcharge period ending, which shows how quickly the direction can reverse.

Besides freight, what other costs move without brands noticing?

Amazon referral fees are commonly assumed at a flat rate, but many categories run a tiered structure that can move the real rate several points in either direction. 3PL invoices carry storage, inbound transfer and integration charges beyond the quoted pick-and-ship rate. GBP/USD movement changes the sterling cost of any dollar-denominated freight or factory price, and EPR and packaging compliance obligations change periodically and are easy to set up once and never revisit.

Should a brand build a freight rate buffer into its Q4 pricing?

A fixed buffer is a reasonable stopgap, but it is a blunt tool against a lane that has moved almost 100 percentage points in three months. A rechecked, current landed cost model at the point of ordering gives a more accurate number than any buffer built on last quarter's rate.

Freight rates on this lane are not going to settle into a predictable pattern any time soon, and neither will referral fees, 3PL charges or compliance obligations. We work with brands who need their numbers checked properly before a Q4 stock build, not after it goes wrong. Book a free 30-minute call if your landed cost model needs a second look.