Notable that the reasons for failure to meet benchmarks was that the locations were too far away, and that they are moving to the “Micro-Fulfillment Center” approach that Amazon is doing at Whole Foods. This is exactly what everyone predicted when Amazon bought WF - turn it into a grocery FC.
That makes sense to me.
Feels like we’re going to have warehouse scale vending machines in cities, and delivery bots taking them from the warehouse-vending-machine to the customer.
This isn’t what Amazon has done at least here on the Peninsula. My WF orders come from Brisbane just like my overnight shipping and Amazon Fresh orders do. Before that they were coming from San Jose, possibly from a WF there but there are closer WF stores in both directions and none in Brisbane.
Sounds like they just put them in the wrong places.
> Fenyo added that Kroger’s decision to locate the Ocado centers outside of cities turned out to be a key flaw.
> “Ultimately those were hard places to make this model work,” said Fenyo. “You didn’t have enough people ordering, and you had a fair amount of distance to drive to get the orders to them. And so ultimately, these large centers were just not processing enough orders to pay for all that technology investment you had to make.”
“Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway”
In the AI analogy, never underestimate the productivity of a human when dealing with a giant pile of groceries. You can throw all the AI and robots you can at something but sometimes a $20 an hour human picking from stacks of goods and produce simply destroys it in raw economics
I idly wonder if what would actually make sense here is a hybrid model that combines a gigantic fulfillment center with tens of thousands of products located "far" from people, with a large physical footprint and near to road/rail arteries, but with a mid-bandwidth, high-granularity, low-latency physical link to "near" places.
For example, imagine you had an upscaled pneumatic tube system (don't get hung up on the exact implementation, it could be a small gauge train system or conveyer belt: whatever floats your Factorio-addled boat) with a diameter around, say, half a metre to a metre, packed goods into canisters and shot into town where they pop out at local distribution centres for pickup or last-mile delivery.
This is where I thought the Boring Company might be going back before it was obvious it was an anti-public transit gambit.
Possibly the curse of rail systems applies where the maintenance of the track (tube) costs so much that it's cheaper to fly (done delivery) or drive all the way on public roads (current solution). The advantage over rail is that the land footprint is very small: the tract is about a metre wide and can be buried if needed. Perhaps it's just not really different enough to trucking it all into town using semi trailers, which would still be required for large items and especially construction materials.
Then again, even if this hare-brained system were to work, this assumes we actually want to continue to reduce most human commercial interactions to gigantic, remote, anonymous capital-intensive megasystems producing pods that pop out of the ground into robotic vending stations.
Yeah, because arguably the main advantage of Ocado's warehouse is that it's extremely dense: you can pack a lot of storage in a very small area and still access it reasonably efficiently. But this only matters if space is at a premium, like near towns and cities (and for low-margin deliveries, you want your drivers to not have to go very far to your customers).
Good news for blue collar workers, tough for investors. The bean-counting dream of LLM's and Robotics remains, but until ChatGPT is placing your order at Mcdonalds drive-thrus and Amazon is laying off warehouse associates en masse, i'd say that last 5% is still taking 95% of the time.
Having FCs an hour away from your customers, packing groceries into a tiny truck with one or two employees per truck, the trucks alone would never pay for themselves let alone the FCs. This was obvious from the get go and it’s why Walmart has been the only one successfully doing grocery pickup and delivery for 6 years. Every store is an FC and they’re all within 20 minutes of their customers.
There are examples of the warehouse-based model working, but they clearly require both density _and_ mindshare. Its not clear Kroger had either based on the other comments in here. FreshDirect in NYC has been operating since the early 2000s with a fleet of tiny trucks with a couple of employees in them and a giant FC with essentially zero retail footprint.
(As an aside, they also have some of the best meat and produce you can get in the city without going to a farmers market. So many retail grocery stores here lack loading docks, the food handling getting from the truck to the sidewalk to the basement of the store to the shelves is really, really rough especially during the summer months. Skipping that and going warehouse-to-home has advantages)
> Ocado has delivery vans that seem to be in no particular hurry (unlike uber eats etc)
True. the difference is that Uber eats takes an order and immediately aims to deliver it as soon as possible.
Ocado takes a grocery order, stores it and aims to deliver it e.g. "next Wednesday, between 6pm and 7pm", and you can modify the order until midnight on Tuesday.
Given that UK Urban traffic levels are unpredictable, the vans will occasionally be in position before the start time. There also is some time where the guy is in the back of the van finding the current order.
But you're right, as the public interacts with them, Ocado's role is "Deliver groceries in a van" not "automation/robotics". The latter is an implementation detail to us.
Also, several people's groceries in 1 van is more fuel efficient and less traffic congestion than all of those people each driving to a big store and back. So it's a win.
Here in the Houston areas, supermarkets like Kroger/Walmart/HEB/etc always have single floor buildings. Why can't they build multi-floor buildings for storage upstairs and retail walk-in sales on the ground floor? On the above ground floors, they can create an automated or semi-automated system for employees to gather up items for online/delivery orders.
I thought about this a lot with parking spaces, nobody like big, open, tree-less parking lots. Why not just build them up adjacent to the grocery store.
The answer is $. It costs too much and land is cheaper than building up (In most places).
I sometime use Ocado in the UK, and it's 'OK' but it's certainly not at the cheap end of the market.
I more often use a traditional supermarkets home delivery service where it's manually picked; those supermarkets have the advantage of having very little infrastructure overhead in the picking - they mostly use their existing stores and pick at quiet times/over night. Ocado has to run entire warehouses just for this task.
Ocado can only work with packed goods - not weighed vegetables for example - which the hand pickers in store can do, albeit whether they do it well is down to luck and the mood of the picker.
I had a wonderful retro futuristic dream about an automated Costco warehouse a few weeks ago. It was one of the less weird dreams so I still remember it clearly.
Basically, each section is like a closed areas with some windows. Customers order at the computers by the windows and flash their membership cards. Robots glide left and right to move 10 samples to the customer, in an arm with rotating clips. Customers can press a button to rotate the samples, observe them, and place an order by pressing a button. Samples not chosen are temporarily stocked at the window as a “stack”.
In each closed section, there are humans who monitors and maintains the robots, and occasionally fetch samples when robots stop working (hopefully it too often, you know those 9s).
At the exit, a human worker assembles the packages and hand them to the customers with a smile. Customers have a last chance to return unwanted items.
Why was it a retro futuristic dream? Because the customers have the option to go into a bakery to enjoy a cup of coffee/tea, some cake and socialize with fellow customers. All of them looked like the men and women from advertisement from Fallout 4.
What you've re-invented is Keydoozle, from 1937.[1] This was the first automated grocery store. Three stores were opened, but there were enough mechanical problems that it didn't work well.
Also worth mentioning Kroger just lost a multi million dollar lawsuit in Florida after one of their delivery trucks hit a cyclist. I wonder if this has anything to do with it as well.
This is a failure of business model and logistics, not a failure of the robotics.
> Fenyo added that Kroger’s decision to locate the Ocado centers outside of cities turned out to be a key flaw.
They over-spent on automating low-volume FCs. You could draw comparisons to Amdahl's law, they optimised the bit that wasn't the issue, the real issue was delivery distances and times.
Ocado has had good success with the robotics approach in the UK, because the UK is very high density compared to a lot of the US. Plus Ocado put a lot of work into creating good delivery routes, whereas it sounds like that wasn't a component of the automation stack that Kroger bought.
> You didn’t have enough people ordering, and you had a fair amount of distance to drive to get the orders to them
It's clearly not a technology problem, but it was made worse by heavily investing in robotics for locations that already couldn't sustain a fulfillment center.
Weird headline, totally unsupported by the article.
According to the article, there were several strategic blunders, including trying the model outside of cities where lack of density cut against it. Plus the apparent dismissal any value their 2700 retail locations could provide.
As far as I can tell, Kroger didn’t acknowledge anything except a change in strategy.
Whenn I lived in Atlanta the Krogers each had a nickname. There was disco Kroger because of its disco ball. There was murder Kroger that purportedly had been the site of a murder.
Or their online groceries didnt succeed because people can p4ice shop. I am of the belief Kroger does well because they obfuscate their prices and their tags which make it seems like they are sales truely are not.
Unlike Publix if I see a discount I know it is a discount. Every other item in Kroger has a yellow tag and a red price to make you think you are getting a deal when in fact the red number is higher than regular price at normal stores.
I've always said that in the back of my mind, the most successful grocery store would be the 'walls' of the store -- bakery, deli, produce, meats, floral, cheeses, dairy and having a little selection of store brands in the middle where consumers can pick up (and vendors can pay a premium for endcap space, because they're the only non-branded products out there), with the rest of the SKU's behind the walls of the grocery store in a fulfillment only model.
Kroger should have pulled a Wal-Mart and turned to their shrink-heavy stores in urban centers to online fulfillment only -- basically only their delivery drivers can retrieve items for an order, and everything's shopped by an associate (Look south of the MicroCenter in Dallas if you want to see what one looks like: it still has the Murphy USA in the parking lot and is basically an unbranded walmart building with 'driver' and 'associate' entrances -- and then deployed the robotics there: less retail space, more online/fulfillment capacity (have humans grab produce and custom sliced/packed items, robots pick the dry goods), and while you lose some cashier jobs, you'll probably have net improvement in terms of time waiting to be picked.
I work in this exact space (online grocery retailer in Europe). We're profitable and one of the few companies to be so in the sector - many online divisions are losing money and being bankrolled by the parent company with physical stores. Alternatively, burning VC money.
The thing that's wrong with Ocado's technology is that it's ridiculously expensive and tailored for huge FC's (fulfillment centers). The problem with that is that it needs to serve a large population base to be effective and that's hard - in dense metros, the driving times are much longer despite smaller distances. In sparse metros, the distances are just too long. In our experience, the optimal FC size is 5-10K orders/day, maybe up to 20K/day in certain cases, but the core technology should certainly scale down profitably to 3-5K. Ocado solves for scaling up, what needs to be solved is actually scaling down.
There are a lot of logistical challenges outside the FC, especially last mile and you need to see the system as a whole, not just optimize one part to the detriment of all others.
> Speaking during an earnings call, interim Kroger CEO Ron Sargent — who took over in March after McMullen’s sudden departure following an ethics probe — said the company would conduct a “full site-by-site analysis” of the Ocado network.
Sounds like there were some politics involved in the original decision.
All delivery services so far, have been partial emulations of a concierge service including the high cost, but have made the user suffer thru the friction instead of the human concierge.
I don't see a way around this using mere technology. Either the service quality will have to be low, or the cost will have to be prohibitively high, or the people providing the service will have to be very poor.
We have solutions going back centuries that work. If you need more than an occasional concierge service you hire a chef as an employee. At some point it makes more sense to hire a part time chef than to pay enormous extra fees for giant shopping robots. Perhaps the economic decline from AI will make the idea of having house servants more acceptable to the very few people still having an income. Most strategies to replace high labor cost people with technology miss the point of long term permanent economic decline. We don't have a problem of too many financially successful people LOL if anything the problem is a severe lack of them.
Its kind of like cars. You have to unperson poor people who know how to drive if you want success with computer driven cars. This is the same concept but for hiring a servant to cook.
Also its very hard to scale royalty. If only the local equivalent of royalty can afford to pay to make the effort required to shop and cook and drive disappear, its going to be hard to dotcom scale that to billions of customers if there's only a couple multimillionaires who don't care about the price.
Trying to automate food shopping is like trying to go back to full service gas stations. Nobody wants it so its an uphill battle to sell.
Around me, the online grocery shopping and pickup is still pretty popular post-covid. There's always a worker or two going around the store picking up for the online orders, so there's room here for some type of site-local automation (not remote robotic fulfillment centers, i could of told you that would be way too costly comparatively).
However, robotics that can navigate an existing retail grocery store is not here. yet.
49 comments
[ 3.3 ms ] story [ 63.5 ms ] threadThat makes sense to me.
Feels like we’re going to have warehouse scale vending machines in cities, and delivery bots taking them from the warehouse-vending-machine to the customer.
> Fenyo added that Kroger’s decision to locate the Ocado centers outside of cities turned out to be a key flaw.
> “Ultimately those were hard places to make this model work,” said Fenyo. “You didn’t have enough people ordering, and you had a fair amount of distance to drive to get the orders to them. And so ultimately, these large centers were just not processing enough orders to pay for all that technology investment you had to make.”
In the AI analogy, never underestimate the productivity of a human when dealing with a giant pile of groceries. You can throw all the AI and robots you can at something but sometimes a $20 an hour human picking from stacks of goods and produce simply destroys it in raw economics
For example, imagine you had an upscaled pneumatic tube system (don't get hung up on the exact implementation, it could be a small gauge train system or conveyer belt: whatever floats your Factorio-addled boat) with a diameter around, say, half a metre to a metre, packed goods into canisters and shot into town where they pop out at local distribution centres for pickup or last-mile delivery.
This is where I thought the Boring Company might be going back before it was obvious it was an anti-public transit gambit.
Possibly the curse of rail systems applies where the maintenance of the track (tube) costs so much that it's cheaper to fly (done delivery) or drive all the way on public roads (current solution). The advantage over rail is that the land footprint is very small: the tract is about a metre wide and can be buried if needed. Perhaps it's just not really different enough to trucking it all into town using semi trailers, which would still be required for large items and especially construction materials.
Then again, even if this hare-brained system were to work, this assumes we actually want to continue to reduce most human commercial interactions to gigantic, remote, anonymous capital-intensive megasystems producing pods that pop out of the ground into robotic vending stations.
Guess I was right.
(As an aside, they also have some of the best meat and produce you can get in the city without going to a farmers market. So many retail grocery stores here lack loading docks, the food handling getting from the truck to the sidewalk to the basement of the store to the shelves is really, really rough especially during the summer months. Skipping that and going warehouse-to-home has advantages)
I associate it more with delivery vans that seem to be in no particular hurry (unlike uber eats / DPD / UPS etc)
True. the difference is that Uber eats takes an order and immediately aims to deliver it as soon as possible.
Ocado takes a grocery order, stores it and aims to deliver it e.g. "next Wednesday, between 6pm and 7pm", and you can modify the order until midnight on Tuesday.
Given that UK Urban traffic levels are unpredictable, the vans will occasionally be in position before the start time. There also is some time where the guy is in the back of the van finding the current order.
But you're right, as the public interacts with them, Ocado's role is "Deliver groceries in a van" not "automation/robotics". The latter is an implementation detail to us.
Also, several people's groceries in 1 van is more fuel efficient and less traffic congestion than all of those people each driving to a big store and back. So it's a win.
The answer is $. It costs too much and land is cheaper than building up (In most places).
I don't know what success looks like but it's probably fair to say they were over-extended by roughly 30-40%.
https://chainstoreage.com/kroger-pay-350-million-automation-...
I'm also skeptical it'll ever work in America due to the general lack of density.
Not quite. Packed yes, but for many vegetables they have both item count and weight-based packages, e.g. "4 potatoes" vs "1kg potatoes".
I think that strikes the right balance.
Basically, each section is like a closed areas with some windows. Customers order at the computers by the windows and flash their membership cards. Robots glide left and right to move 10 samples to the customer, in an arm with rotating clips. Customers can press a button to rotate the samples, observe them, and place an order by pressing a button. Samples not chosen are temporarily stocked at the window as a “stack”.
In each closed section, there are humans who monitors and maintains the robots, and occasionally fetch samples when robots stop working (hopefully it too often, you know those 9s).
At the exit, a human worker assembles the packages and hand them to the customers with a smile. Customers have a last chance to return unwanted items.
Why was it a retro futuristic dream? Because the customers have the option to go into a bakery to enjoy a cup of coffee/tea, some cake and socialize with fellow customers. All of them looked like the men and women from advertisement from Fallout 4.
I’d like to shop or even help build one of these.
[1] https://rarehistoricalphotos.com/keedoozle-automated-store-p...
https://en.wikipedia.org/wiki/Best_Products
https://m.youtube.com/watch?v=Xc_6wfDYuFU
> Fenyo added that Kroger’s decision to locate the Ocado centers outside of cities turned out to be a key flaw.
They over-spent on automating low-volume FCs. You could draw comparisons to Amdahl's law, they optimised the bit that wasn't the issue, the real issue was delivery distances and times.
Ocado has had good success with the robotics approach in the UK, because the UK is very high density compared to a lot of the US. Plus Ocado put a lot of work into creating good delivery routes, whereas it sounds like that wasn't a component of the automation stack that Kroger bought.
Just the model. They optimized for tax outcomes. They obviously wanted in on the "Amazon race to the bottom" game.
https://www.leesburg-news.com/2025/11/30/kroger-took-incenti...
> the real issue was delivery distances and times.
Which they could care less about. Where else are you going to go?
> You didn’t have enough people ordering, and you had a fair amount of distance to drive to get the orders to them
It's clearly not a technology problem, but it was made worse by heavily investing in robotics for locations that already couldn't sustain a fulfillment center.
According to the article, there were several strategic blunders, including trying the model outside of cities where lack of density cut against it. Plus the apparent dismissal any value their 2700 retail locations could provide.
As far as I can tell, Kroger didn’t acknowledge anything except a change in strategy.
This I guess is Robo Kroger.
... As in the supermarket had a disco ball? Was it spinning?
Unlike Publix if I see a discount I know it is a discount. Every other item in Kroger has a yellow tag and a red price to make you think you are getting a deal when in fact the red number is higher than regular price at normal stores.
Kroger should have pulled a Wal-Mart and turned to their shrink-heavy stores in urban centers to online fulfillment only -- basically only their delivery drivers can retrieve items for an order, and everything's shopped by an associate (Look south of the MicroCenter in Dallas if you want to see what one looks like: it still has the Murphy USA in the parking lot and is basically an unbranded walmart building with 'driver' and 'associate' entrances -- and then deployed the robotics there: less retail space, more online/fulfillment capacity (have humans grab produce and custom sliced/packed items, robots pick the dry goods), and while you lose some cashier jobs, you'll probably have net improvement in terms of time waiting to be picked.
The thing that's wrong with Ocado's technology is that it's ridiculously expensive and tailored for huge FC's (fulfillment centers). The problem with that is that it needs to serve a large population base to be effective and that's hard - in dense metros, the driving times are much longer despite smaller distances. In sparse metros, the distances are just too long. In our experience, the optimal FC size is 5-10K orders/day, maybe up to 20K/day in certain cases, but the core technology should certainly scale down profitably to 3-5K. Ocado solves for scaling up, what needs to be solved is actually scaling down.
There are a lot of logistical challenges outside the FC, especially last mile and you need to see the system as a whole, not just optimize one part to the detriment of all others.
The title is a red herring.
We have done grocery pickup for years but the pickup lanes are almost always empty while dozens of shoppers walk into the store.
To me, shopping for groceries by hand is a waste of time but it clearly has some utility for a lot of people.
I wonder if that inertia is making traditional grocery shopping stickier than it should be and disincentivizing optimization.
I hope consumer tastes will change because there’s no reason for us to all walk into a giant warehouse every week.
Sounds like there were some politics involved in the original decision.
I don't see a way around this using mere technology. Either the service quality will have to be low, or the cost will have to be prohibitively high, or the people providing the service will have to be very poor.
We have solutions going back centuries that work. If you need more than an occasional concierge service you hire a chef as an employee. At some point it makes more sense to hire a part time chef than to pay enormous extra fees for giant shopping robots. Perhaps the economic decline from AI will make the idea of having house servants more acceptable to the very few people still having an income. Most strategies to replace high labor cost people with technology miss the point of long term permanent economic decline. We don't have a problem of too many financially successful people LOL if anything the problem is a severe lack of them.
Its kind of like cars. You have to unperson poor people who know how to drive if you want success with computer driven cars. This is the same concept but for hiring a servant to cook.
Also its very hard to scale royalty. If only the local equivalent of royalty can afford to pay to make the effort required to shop and cook and drive disappear, its going to be hard to dotcom scale that to billions of customers if there's only a couple multimillionaires who don't care about the price.
Trying to automate food shopping is like trying to go back to full service gas stations. Nobody wants it so its an uphill battle to sell.
However, robotics that can navigate an existing retail grocery store is not here. yet.