Ocado automated the picking and not the driving. Its accounts show one cost line standing still while the other climbs

Two public documents price the same company's machine and human labour side by side. The first is Ocado Retail's annual report for the 70 weeks to 6 April 2025, audited by Deloitte and signed on 2 September 2025. The second is Ocado Group's unaudited interim results for the 26 weeks to 31 May 2026, published on 16 July. Both are free to read. Every figure below is stated in them, with the basis named each time, because the reporting periods are a trap: the annual report's growth figures use an unaudited 52-week comparison that the company publishes precisely so its awkward 70-week statutory period can be compared fairly. We do not mix the two.

Inside them sits a natural experiment. Ocado Retail runs two labour-heavy operations at once: picking orders in warehouses, which it has automated, and driving them to doorsteps, which it has not. Same company, same orders, same periods. On the 52-week basis, orders in the year to 6 April 2025 grew 15.2 per cent. Fulfilment costs, which the document says primarily comprise labour and consumables, grew 3.7 per cent to £187.0m. Service delivery costs, which it says comprise labour, fuel and fleet, grew 22.7 per cent to £337.0m. Warehouse productivity rose 11.9 per cent to 225 units per hour on that report's measure; items delivered per van fell 1.2 per cent to 939. In the half year to 31 May 2026 the pattern held: orders up 12.8 per cent, warehouse costs up 1.7 per cent to £99.9m, delivery costs up 14.9 per cent to £211.8m. And for a household the same table already reads as prices: Ocado's average item price rose 2.1 per cent in that half, against grocery inflation of 3.9 per cent on Nielsen's measure, which the company cites.

Say the caveats aloud before the conclusion. The company does not draw this contrast; it is our reading of lines it prints side by side, and warehouse work and delivery work are different jobs under different constraints, so the comparison is illustrative rather than controlled. A cost line is not a headcount: no audited headcount for the warehouse workforce appears in these filings, and we are not borrowing one from unaudited trackers. And the people themselves sit in a third set of books: the interims describe Ocado Logistics, a Group business, operating the warehouses and the vans for both Ocado Retail and Morrisons and recharging the costs plus a management fee of about 4 per cent. The hands and the wheels are employed beside the numbers, not inside them.

The company gives five reasons for that 22.7 per cent delivery rise, and all five belong here: order growth, wage inflation, delivery inefficiencies, additional insurance costs, and network reconfiguration after closing Hatfield. The absence of automation is not on its list; that connection is ours. In the latest half it names two more forces on the human legs: the National Living Wage rise and higher employer National Insurance contributions. So some of what looks like machines winning is Parliament pricing human work higher, a point our own Contrarian made on 19 June, and on the delivery line these filings support him.

Now the tell in what moved the wrong way for a tidy story. Delivery productivity improved in the latest half: drops per standardised eight-hour shift rose 6.1 per cent to 22.5, which the company attributes to routing software and denser rounds. The machine is not only replacing the picker; the software is squeezing more from the driver it cannot yet replace. One profit-and-loss account, three states of work: automated, augmented, and priced up by statute.

Where does this touch a British household first? Probably not as anyone's P45, on this evidence, but at the till and in the pension. Prices first: an operation whose warehouse productivity rose 11 per cent while its item prices rose slower than the market's is showing one route by which cheaper machine labour reaches a shopping bill. Margins took some of it too: Ocado Retail reported adjusted earnings of £72.9m in the half and a pre-tax profit of £11.6m against a £17.1m loss a year earlier. How the gains from machine labour split between prices, wages and profits is the distributional question of the coming decade, and grocery filings like these are where it will be answered first. The pension channel is exposure, stated as information only: Ocado Group is a London-listed company, so index funds and workplace pensions commonly hold it, meaning many households own a sliver of this machinery without knowing. That is a fact about ownership, not a view on the shares, and this desk gives none.

Scale, held honestly: under two per cent of the nation's food shopping runs through these warehouses. What to watch next: whether the £15.6m of half-yearly fees still paid for the closed Hatfield site run off, and what the numbers show when Asda's Ocado-powered service goes live next year and this machinery starts running under a second British brand.