Quick Commerce Delivery - Grocery On Demand: Delivering the Potential

Quick commerce delivery is put under the microscope as Colin Peacock speaks with Professor Daniel Corsten about why on-demand, store-picked grocery orders are so difficult to run profitably.

They separate true immediate delivery of small baskets from scheduled online grocery and explain how labour, picking and delivery costs can quickly overwhelm margins.

They also explore how on-shelf availability and substitutions add time, cost and the risk of losing the whole basket, before closing with practical ideas around replenishment, smarter substitutions and better data sharing.

Cleaned transcript

Setting the scene for quick commerce delivery

COLIN:

In this podcast, we are in conversation with Professor Daniel Corsten, and we are exploring the operational challenges around the quick delivery commerce proposition for grocery retail.

In our travels, we’ve come across this world of quick commerce in grocery – or “grocery on demand”, G.O.D. as some say – helped and enabled by third-party platforms such as Deliveroo, DoorDash, Instacart, and I think you’ve also worked with one called Everli.

We’re also seeing retailers doing their own versions, like Tesco with Woosh, and other examples around the world from companies such as Loblaws and others.

It’s a really interesting area, not least because it really puts a fine point on what’s been close to our heart: when you’re picking from the shelf, on-shelf availability suddenly becomes really important – something you really want to get right.

We’ve had these platforms join our meetings because of this. They’re looking to work and collaborate with retailers who own the shelves – and are often doing the picking – to make sure that the products are actually there.

Because there is a world of pain when the products aren’t there, especially with the quick commerce proposition.

So I really wanted to set up this conversation with you, Daniel, to share – and hear from you – your reflections on what we’ve been hearing in our meetings over the last 12 to 18 months, what some of the implications are for retailers and for the platforms, and where some of the ideas for improvement might lie.

What counts as quick commerce delivery

DANIEL:

Yes, absolutely. Before we start talking about quick commerce, let’s first clearly define what we mean – and what we don’t mean.

In the context of the seminar that we had, we’re looking at quick commerce that is picked in stores, and delivered from stores.

That’s very important, because there are versions of quick commerce that pick from dark stores – stores that do not allow customers to enter – which are really mini-fulfilment centre warehouses. We’re not looking at that. They involve different economics.

We also don’t look at warehouse-based picking. So it’s very important to know what’s out of scope.

The second thing we’re not looking at is also important, because everything tends to get mixed up.

When we talk about quick commerce here, we mean on-demand commerce: I order, and it gets delivered immediately.

“Immediately” here is very different to the original form of online ordering, where you and I, Colin, would order the weekend shopping on a Saturday and want it delivered on Wednesday evening.

There, we’d book a slot, we’d have lots of time to wait, we’d plan ahead and make provisions. Then a few big boxes would arrive, and we’d make sure that somebody was at home.

Why is this distinction important? Because the economics are different.

As soon as I have large baskets and I can schedule delivery, I can get economies of scope. I can get some batching. So the economics are very different.

Why small baskets make quick commerce delivery so hard

DANIEL:

What we’re talking about here is Colin sitting on his couch in summer, totally hot and sweating.

He suddenly realises that although there were five bottles of beer in the fridge, now they’re all gone. He wonders why, but he needs new beer – and he also needs a bag of crisps.

He’s a little too tired to drive – and thank God he’s not driving, because he’s realised this – so he calls some service and wants that beer delivered within the next 15 or 30 minutes.

It’s exactly another five cans of beer and two bags of crisps, for a total of maybe 10 bucks. That’s not a very expensive purchase.

The problem is the person who will bring this will not be able to do another job at the same time.

They will drive to you, drop the stuff, and come back. There is no way to achieve economies of scale on that trip.

This is why this model is so interesting – and problematic.

Every ride is specific to a single order, and those orders can be very small. The economics of this business model are terrible.

That’s why it’s come under scrutiny. Over the last two or three years, there’s been a frenzy of mergers and acquisitions in this space, because many players realised they weren’t viable.

The economics didn’t add up. They were making losses. Venture capitalists were willing to fund that, but only to a degree.

So they gobbled each other up, hoping that there would eventually be one, two or three players left who could then turn a profit. That’s the interesting backdrop.

So that’s the first key point: what quick commerce we’re actually talking about and why it matters.

Is quick commerce delivery viable, and where

COLIN:

So let’s get to the point that’s probably the most interesting: is it viable, yes or no?

I guess one distinction is who does the picking, right?

DANIEL:

There is something more fundamental, though. It does have to do with picking, but let me take one step back.

The question is: Is quick commerce delivery viable? And if yes, when and where?

This is interesting because, first of all, there are places where it is viable – but these are not the places where you and I live. These are locations with very, very low manual labour costs.

The problem is the cost of labour. It’s also, to some degree, the cost of the product.

Many of these platforms don’t have the same volume discounts as a Tesco or Walmart. They pay more to branded suppliers for the crisps and the beer, so there’s a disadvantage on the purchasing side.

There’s also a massive disadvantage on the manual labour cost side.

Remember, the breakthrough and revolution of supermarkets was that the whole job of picking was effectively outsourced to you and me. We go to the supermarket and pick for ourselves. We’re the warehouse workers.

More than that, we often deliver the products home ourselves. So we’re warehouse workers, and we’re logistics service providers. That’s why supermarkets can be so cheap.

The problem with quick commerce delivery is that all these additional costs – picking and delivery – come back in.

So you have two extra costs to account for: the picking cost, and the delivery cost.

Retailers and platforms have partly found ways to roll over some of the delivery cost – through delivery fees, minimum order values and so on – and that covers some of the delivery.

But what nobody is clearly covering is the cost of picking.

The data show that the cost of what you and I do for free is substantial.

If you roll this over, you realise that for an ordinary job of, say, 20 items, we easily spend 20 to 30 minutes in the supermarket picking – running around from A to B and finding products.

This cost is one of the big costs that makes quick commerce loss-making.

In countries in the Middle East, Africa and parts of Asia, where manual labour costs are much lower, it turns out you can actually run this profitably or close to profitably, because the cost of picking is so much lower.

Who pays for the work behind quick commerce delivery

COLIN:

We had an academic, Santiago Gallino, illustrate this in his Harvard Business Review paper.

He articulated exactly this question: where is this cost being picked up?

I think he was a bit disgruntled that he was busy doing all his shopping, while people like me, sitting there with five beers and two packs of crisps, were being subsidised by him – or that prices were a bit higher because of people like me.

DANIEL:

He has a point. That’s an interesting angle, without giving away too much.

Santiago Gallino, professor of Wharton, has a big family. He has to do big shopping trips, often with a big car.

He realised that he and his wife are spending all this time doing the shopping, and yet he’s paying the same prices as others who are just like you, Colin, waiting at home for the stuff to arrive.

So he basically asks: Why should I – the person who does the warehousing work by picking and delivering my own orders – pay the same as someone who gets it delivered?

Why should the people who do the work subsidise those at home?

He wonders whether there should be some incentive for doing one or the other. That was one of the questions we had.

There is a reason why retailers might accept losses on online picking for a while.

It’s an interesting group of people: particularly young customers.

Retailers may want to acquire them when they’re young and maybe a little foolish.

Over time, they grow into families and continue shopping with the retailer. Their purchases and baskets get bigger, and eventually the retailer hopes to make money from them.

So there are some strategic “kinks” in the model.

We also had some really deep work from other attendees, such as William West from West Monroe, who looked in detail at all the data around picking times.

That work shows just how big a problem it is.

On-shelf availability, substitutions and hidden picking costs

COLIN:

Let’s dive into that a bit. We saw the breakdown of costs in a really interesting model presented by West Monroe.

Using your 8 percent number for the likely proportion of items that are out of stock, we saw that maybe a quarter, or even a third, of all the costs associated with picking can be down to the problem of: it’s not there on the shelf when I’m there to pick it.

What were some of the ways you were learning about in terms of interventions to reduce that problem of needing substitutes?

DANIEL:

This is quite simple at a high level.

The first solution is: if you’re always in stock, you don’t have to substitute. Full stock, no substitution.

But being always in stock requires a very sophisticated replenishment system and higher safety stocks.

That can be a problem, particularly for very volatile products with high demand variability. So that’s not always an option.

The second issue is what’s happening right now in many operations.

You may order a product, and on the website it says it’s available. But because the picking is done from the store, by the time the picker comes to pick it, it’s no longer there.

At that point, the picker may call you – if they’re lucky and you answer.

That’s an enormous amount of wasted time and effort for the picker, but some retailers still do this.

Smarter approaches include asking the customer upfront: If that product you want isn’t available, which one would you like instead?

Customers can pre-register substitute items.

Very few people actually take the time to do this, unless they have a strong reason – for example, a child with a severe allergy, where they really don’t want the non-vegan variety or a product with a particular ingredient.

So we worked on another interesting alternative: an algorithm that predicts, based on your previous purchases, what your personalised substitution recommendation should be.

I think we’ll see more of that in future.

But again, the best way is to avoid substitutions altogether, because of their cost.

From our research last year, remember, we found that for every product you have to substitute, for every dollar of margin that you have on that product, you’re spending about 2.6 dollars in the actual cost of the substitution.

That includes trading up to a more expensive substitute, the picker searching for it and finding it, and the risk that even then it might be the wrong one.

Basket loss, minimum order values and quick commerce delivery

COLIN:

And going back to that beer and two packs of crisps example: I may not reach the minimum order value with just that, so I might add three bars of chocolate.

If the bar of chocolate is in stock, but the beer is out of stock, I might say, “I don’t want the order if there’s no beer.”

That’s where you can lose the whole basket because of one out-of-stock item.

So again, it highlights how important it is.

I think you’re right: we talked a lot about how platforms and retailers can work together.

There was a lot of momentum behind questions like: How much more data sharing can go on between delivery platforms and retailers?

How much can they see in real time what’s actually there on the shelf?

That then enables some of those practices around substitutions you described, and options like “don’t send the order if this key item isn’t there”.

Data sharing and operational roles in quick commerce delivery

DANIEL:

Yes, there are definitely ways to share more data. But the data needs to be good. Sharing bad data isn’t very helpful.

In some situations, particularly in the UK, external platforms are not really welcome to pick in the stores. Instead, a store associate does the picking.

They have access to the inventory system. The external party then essentially becomes just a delivery platform.

That’s one way to improve the situation.

Another, simpler option is to share location information. More and more retailers are doing that.

It sounds trivial, and technically it can be solved, but there are challenges: the location information may not be very good, and it may not be precise enough.

So there are some obvious solutions being explored, and more retailers are working to get them into operation.

I’m hopeful that some of these problems will be solved.

I’m a bit less hopeful that we can fully solve on-shelf availability.

But that’s good news for us in the ECR community, because it means we can continue with this working group.

Technology signals and the on-shelf availability problem

COLIN:

Absolutely – it’s the gift that keeps on giving, right?

Seriously, I think there could also be a role for computer vision.

We’ve talked about where cameras could play a role, either as a signal or via robots or other technologies. Potentially, this is another area where they could be useful.

As you say, quick commerce delivery is bringing on-shelf availability right back up the priority list.

If you want to grow this quick commerce business, it’s pretty clear that you need to reduce the cost of picking by reducing out-of-stocks and the need for substitutions.

Closing thoughts on quick commerce delivery

COLIN:

Excellent. We’ll leave it there.

Thank you very much again for giving up your time, Daniel. We’ll meet again.

In a real quick commerce situation, from the moment you first mentioned that you’d like to have a beer, by now you should have it.

So my question is: did anybody ring on the door?

DANIEL:

Not yet, not yet. Maybe they’re out of stock.

COLIN:

Okie dokie. Thank you so much. Bye-bye.