European On-demand Grocery Delivery – Dark Store Model & Operations Analysis

  • Private Equity
  • Consumer
  • Europe
Premium Unlocked


Former senior executive at Gorillas Technologies GmbH


  • Dark store penetration prospects for q-commerce (quick commerce) players by major regions and location dynamics
  • Inventory turnover and SKU management vs micro-fulfilment and warehouse models
  • Competitive landscape and dark store roll-out by key players such as Glovo (ETR: DHER), Getir and Gopuff
  • CAPEX per store and profitability potential



How do you define a dark store in the context of the European q-commerce [quick commerce] and on-demand delivery market?

Specialist (SP): A dark store, it’s actually a shop where customers cannot get inside to physically buy products but they have to shop online, most likely through an application. On the application, they’re going to see the full catalogue, full assortment, they’re going to select products, add to them to the basket, prepaid orders, so through credit cards, PayPal or other payment methods. Then they place the order. The nearby dark store, so based on their geolocation, receives the order, and then the order gets fulfilled immediately inside a dark store and then assigned to riders for final delivery to customers. If we identify dark stores in a quick commerce environment, then we have to think about on-demand deliveries, which range between 10 minutes, 30 minutes, up to an hour. Anything else, it’s scheduled deliveries. I would not define it dark store. It can be called dark store, but it’s not quick commerce, it’s more like online shopping with pre-fixed time slots that customers have to book. That’s the dark store definition, I will say, on-demand service in a matter of few minutes, and then dark store, it’s a shop where you cannot go inside as a customer, not physically get in. We did some tests, some experiments for click and collect, so the only option for customers to pre-order anyway on the application, prepay everything and come to collect items themselves.

Back to top
How is a dark store different from a micro-fulfilment centre or warehouses that marketplaces are beginning to use?

SP: The main difference is about layout. Dark store, as I said, is a shop where customers, they don’t get in, so it’s not optimised for product utilisation like a normal shop, working shop, but it’s optimised as a routing inside to prepare a shopping bag with grocery products. Every product has a zone, so specific letters, numbers or codes which identify a specific area in the shop, and there is a routing, so when the customer selects products and adds it on the basket and the order, it’s received to be fulfilled, it’s sorted out in terms of the zoning assigned in dark store. The shelves and fridges, they have a specific zone. Each SKU is connected to a zone, and then their ranking is done based on product dimension and weight. Same as a standard customer would have to get products at the cash counter in a supermarket and put them inside shopping bag to get these products at home, then these items, they will be collected during the picking process with the same priority. Heavy items, they will be first, like bottles, water, beverages. Anything that’s heavy and big will go on the bottom space of shopping paper bags, and then across the circuit, across the routing, all items, they will come later and, at the end, you will have soft items, vegetables, eggs, everything else that can get smashed if it stays at the bottom.

Just to clarify, dark store has a definition, you can call it mini-fulfilment centre, we call it warehouse. It’s about naming, but what’s its main difference between dark store on a quick commerce operation and a big warehouse for e-commerce operation is that there’s a specific route inside the location and you don’t have shelves, which they don’t have a specific, let’s say, picking circuit. That’s because the assortment, it’s consolidated on a few thousand SKUs. You will never have dark stores probably 10,000-20,000 SKUs. According to the business models and trade-off, the number is going to be limited, and then that’s how you’re going to have routing that allows you to go and see all products in a matter of minutes.

Back to top
How many SKUs are typically housed within one dark store? Can you outline a dark store’s size, given it only holds a certain number of SKUs?

SP: For example, in European model, most of the players, they set their assortment between 1,000 and 2,000 SKUs, so then you have to consider the main complexity in terms of logistics supply chain is to stock up a few amount of SKUs, so low amount of SKUs, up to 2,000, in very few units because everything has to be high-rotation, move in and out quickly because you don’t want to invest too much in real estate and to rent big spaces located in the city centre. Then as big is the space, more expensive, but also you waste more time to do the picking because it’s not that dense, items are far away, and then it’s not convenient for fulfilling orders quickly. Let’s say the picking area, the shop itself has up to 2,000 SKUs maximum, and then they are stored in very few units, let’s say maximum 50-100, according also to dimension, but average from a minimum of carton boxes, so 12-24 items for each SKU, up to available space, some small items, they can even be stored in 100 or 200 pieces in the same shelf if possible. Then the picking area itself will be roughly 150-200 square metres maximum, including shelves, fridges and refrigerators, and then you might have a so-called back-stock area. It’s buffer stock where you go to restock items when the shelves are empty, there’s high rotation during the day and they need to be stocked up from existing items and SKUs already inside dark stores. Then you will receive daily or every second day a replenishment from suppliers or eventually from a central warehouse expansion set-up.

Back to top
How much of the 1,000-2,000 SKU range comes from fresh food such as vegetables vs more FMCG-related products vs personal care? How are dark stores determining the SKU mix that can ensure a short inventory turnover period?

SP: Fresh products, like ultra-fresh fruit and veggies, bakery, these kind of products, are the ones that are targeting customers, the most loyal customers, because if they buy fresh products, it means they buy them every day and they always come back and they also spend money on the other products, other parts of the assortment. I would say in terms of SKUs, 50% has to be fresh, fresh from fruit and veggies, then daily products, even meat and fish, so at least you have a basic assortment, standard products everybody can easily find on supermarkets. You probably don’t want to go deep in terms of selection of these kinds of products because there’s a big trade-off as much more options you provide, of course, you might attract new customers or provide a broader selection, so better catalogue, but still, you highly increase opportunities to create wastage. If ultra-fresh, fresh products, they don’t rotate quickly, then, of course, they deteriorate, they expire and then you have to throw them away. Wastage has to be minimised, of course. So-called ambient products, they have a long-term expiration date, between six months, 18 months. It’s not a problem there, but with fresh products, if you don’t rotate fast enough, you always need to have a minimum available inventory, minimum stock available, you generate wastage. Then you could do some tests, so test a specific kind of meat, fish or specific seasonal vegetables, but then if they don’t rotate, there’s no point to have 100 different kinds of fruits SKUs because, of course, you’re going to sell the most popular ones, like bananas, most than the others.

You need to make sure you have a short selection and always stock available, and it’s also crucial to do very high standard quality check, especially on fruits and veggies, because customers, they put their trust on this service of somebody selecting those kinds of items, those kinds of products as they would do by themselves when they’re at supermarkets. If I order and I receive fruit and veggies which are not fresh, they’re rotting, deteriorated, next time I will probably not order again, so it’s fundamental to keep very high quality standards and also this creates opportunity to have waste.

Back to top
How well-penetrated are dark stores in the key European markets, so Italy, Spain, Germany and France? How have certain players approached this model? Is it consistent across all players in different markets or are there any nuances?

SP: Overall, I can say penetration at the moment is very low, considering the major cities in Europe, for example, Gorillas and the other players, they cover a majority of city population in the city centres, the very dense areas, so talking about potentially millions of customers. Considering the number of orders we daily do, there’s a lot more that can be done. Overall, I will say 1% maximum could be the penetration in cities more structured like Berlin and Amsterdam. In Italy, we are about 0.5%. London, I guess it will be something in between 1% and 0.5%. There’s a lot more potential to be unlocked and their customer acquisition is still early stage because targeting few specific groups of customers and not going so mainstream in some markets, you then have to really crack the mainstream consumers at the moment.

Back to top
How are players such as Getir, Gopuff, Glovo and Gorillas approaching the dark store model differently from each other and what are their penetration rates?

SP: Honestly, I believe penetration is quite similar, sometimes some players, they push, they sponsor some more marketing investments or they push on discounts, aggressive vouchers to get promotion and to acquire new customers. Overall, from the market I oversee, it’s pretty much similar, the amount of customers, because they are interchangeable, so one player gets into the market first, they create the market, then consumers, they eventually try similar services because they already know how it works.

Third Bridge (TB): Is the dark store model consistent across all players or are there any differences?

SP: Yes, the main differences are related to densifications of areas and then assortment rates, so some players, they want to focus on very short selections, so focus on 1,000 SKUs or even less, super essential products, so they target some emergency shopping habits and mostly beverages or these kinds of products that people buy. It’s not a full shopping experience, full basket, but there are only a few target SKUs, and so they have minimum lower basket. They have lower basket because less available options, and they have to target on very high volume and deliveries very close to their location to make some good economics. Other players are targeting a full shopping grocery experience, so they target maybe customers that they buy once a week and they get the whole catalogue, all products, and it’s something that should substitute the big shopping at supermarkets that people do on the weekend, for example. You need way more SKUs, up to 5,000-7,000, to provide a full range of products so people will have full options, full choices, their basket size will increase, but, of course, orders tend to become heavier and you need to probably have extended delivery areas to catch more customers so that they’re going to buy the full experience.

It’s a trade-off between distance, assortment. Distance comes also with densification, so understanding how many dark stores you need to open to target a specific population. Then as soon as orders, they become heavy, it’s also a differentiation on delivery time, of course, and the vehicle you’re going to use to do these deliveries, because having only electric bikes doesn’t allow to carry a lot of kilos, so you need cargo bikes, motorbikes, electric vans eventually. Then it’s also a trade-off between the service if you can stay in a range of on-demand deliveries, so 10-30 minutes, or then you go above and you have time slots to be booked from customers, but that’s a very different thing in my opinion. Then some players are doing some sort of different models, maybe we can consider food delivery players because they have a higher penetration in terms that they can leverage their huge customer base because they already have a lot of customers buying regularly from their restaurant marketplace. Those players, they might have a more interesting penetration at the moment, but some of them, like Glovo, they apply a considerable mark-up on products, so they come as a little bit more expensive, and they’re not yet to substitute the full shopping experience, like once a week for all grocery products. It’s more emergency shopping habits.

Back to top
A key point you mentioned is how players are targeting certain geographies to determine the number of dark stores they’re supposed to have. In a large city such as London, the under-10-minute proposition means that a lot of dark stores are necessary. How does a player determine the number of dark stores for one serviceable area, and how does it define a serviceable consumer base or area?

SP: It depends on the market unit economics. Roughly, an estimation, if you target a specific delivery area then you can have some separate topics about if you want to target high-income district, family districts more or, I don’t know, somewhere where more students live in or you have offices. It really depends on the strategy. Overall, you should target to open a dark store that at least can serve from 50,000-100,000 potential customers. In that case, if you have 1% penetration, which is still low but something you can expect at the beginning, you can range between 500-1,000 orders daily, for example, or every second day, and so you can make some estimation or you can forecast on profitability. It’s a game of volumes, of course, because you need to optimise cost of employees in the warehouse, then delivery cost, it can be viable, depends on which employment model do you use, but also then you have to cover inventory and logistics costs, which are pretty much fixed to guarantee replenishment and stock availability.

You need between 500 to 1,000 orders daily. That’s how you target how many dark stores to open in one city or if makes sense to open a small, medium or second-tier, third-tier city, if allows this kind of density, and also you have to consider delivery area, distance. In my opinion, from my experience, if you do deliveries on average range between 500 metres to one kilometre, you can get very good level of delivery productivity. If you target areas where you go two, three, four kilometres on average distance, then it’s going to be very complicated to get some optimisation in terms of delivery cost, and also you’re going to provide a delivery time which is not optimal to customers, so have them wait longer to group orders together and to get some profitability, but then it’s not going to be that quick.

Back to top
You said a player needs at least 500-1,000 orders per day to justify having a dark store. Considering the number of existing players, is 500-1,000 currently too high and can players hit this number per dark store?

SP: It can be kind of ambitious, considering cost of acquisition and competition. Of course, this can be considered a premium service because you end up paying delivery fee, which might be symbolic, of course, do not cover all costs for the company, but still it’s another price you’re going to pay, and products, maybe they have a mark-up, so they’re a bit more expensive than offline options to do shopping in the neighbourhood. What also has to be considered, it’s average order value, so really depends on how much revenue overall you make, how much gross percentage value you generate, and then average order value comes with a trade-off also of delivery cost. Higher amount of money spent from customers on a single order, it probably means that order has, if a higher value, higher amount of items, so more complexity, more heavy weight to be delivered, and also maybe you need different kinds of vehicles, you need to deliver it slowly and so on. The trade-off is between orders, average order value and the time you have to consider to deliver the same amount of orders.

Back to top
Are the dark stores mainly in commercial or residential properties? How are players evaluating and sourcing ideal locations to offer the value proposition?

SP: Mainly, research starts with a location, so target location in terms of geolocation, to make sure you are in between the neighbourhood or the district where you want to be, so you’re close to the customers you want to target. You start your real estate research around that. Also considering delivery areas match between dark stores. It depends also from the company technology, so, for example, at Gorillas you set a delivery area and then it can never overlap with the other ones. If some companies are able to do overlapping delivery areas or dynamic adjustments based on volume, timing, emergency or even understanding if one customer is localised between two areas and then the orders would be submitted to the store opening the application, customer will see dark store, maybe not the closest one but the one with higher amount of inventory or most selection to make sure he places an order. This is technology that has to be discussed by the company.

Then, going back to the selection, so first you identify the district where you want to be, then you need to, of course, consider real estate rental cost, minimise risk in terms of yearly years of rent contract, if possible, but, of course, as more you commit, more you’re going to negotiate the price down. Then, overall, on a monthly, yearly P&L of a single shop, even if you pay high amount of rental, it’s not as relevant as labour cost at the end, but that’s another topic. Then you need to understand how many square metres you need and that’s according to the number of SKUs you’re going to have in stock and if you want to have a high amount of buffer stock to daily do replenishment. That’s something that has to be strategically understood. Range of square metres according to how many products you want to sell. It depends then on local regulation, if you need additional space for employees if they need to dress up, change their clothes, if you need to store bags inside, if you have any other equipment you need to store, I don’t know, some sort of transpallet to move things around, if you need the buffer stock for paper bags. Any additional space, it’s something that has to be considered according to local regulation or business needs.

Then, ideally, you need to really be careful about accessibility. Accessibility for trucks, for example, to load products on a daily replenishment, on a weekly replenishment, so you need to understand if your supply chain transportation comes with pallets, roads, which parking space you need or loading area for trucks and so on, and also accessibility for riders to get in and out. Riders or cargo bikes or motorbikes, or if you have electric motorbikes, you need to find also, maybe, charging stations nearby. All these kinds of details have to be considered according to the business model or the strategy needed. Of course, for accessibility, you need to look for locations which are at the ground floor, so easy access, no underground or no upper floors because, of course, everything gets more complicated to load, in and out, products.

Back to top
How do dark stores receive the products from suppliers? How is the order made from the dark store to the supplier? What does that supply chain look like?

SP: I can explain what we did as a set-up in Italy, then there are other models, so can be direct deliveries from suppliers or central warehouse. Ideally, you have a buying team that centralises all purchasing activities, so buying team selects the catalogue, they decide which products, which categories, sub-categories you want to have on the application. They do their own consideration, own selection of how many fresh products, food products, non-food products, beverages and all the other categories according to their targets in terms of category mix, margin they want to achieve, budget and everything. As soon as they selected products, they start to identify suppliers and they leverage their own network, they get suppliers, so directly calling brands, distributors, wholesalers, everybody that can get the products they’re looking for. Then, ideally, all catalogues nationwide should be 90% standardised, especially if you want to have a central warehouse so you can aggregate all products on a specific location, and you can have some localised products to target specific needs of local customers. For example, bakery products, it’s an exception because in every city you have a different bakery supplier which comes every day with these ultra-fresh products and they do direct deliveries to every single dark store.

For everything else, what we arrange, it’s a central warehouse that collects all orders from all suppliers, all categories, so so-called ambient products that are not fresh and they don’t need to be refrigerated, fresh products as well. Then you can do stock-up of everything that doesn’t have a very short expiration date in this central warehouse. For fruit and veggies, we arrange it as a cross-stocking, so on-demand requests from warehouses which we’re going to deep dive soon. All products coming in the main warehouse and then aggregated to have a single drop for a single dark store, so receiving every second day in this case, but can be done daily or once, twice a week, according to business needs or product deterioration. We arrange it every second day, so three times a week, to make sure products are not deteriorating and always available. You have a single drop for fresh, ambient products, refrigerated products, beverage and so on, so that simplifies inbound activities because you can have additional staff on specific moments of the day and they know, for example, every second day, in the morning, they do inbound activities. In the morning you have less orders to fulfil, and then you can use the same amount of people to do inbound, stock up products, you do strategically before the weekend, for example, or before the day you know when you’re going to sell more than the average of weekdays and so on, so you can have the stock ready for peak times.

Then how the quantities are decided, we build up a replenishment forecast tool, so based on available quantities in the central warehouse, each single dark store, we forecast it based on historical sales from a few weeks before, how many units were supposed to be replenished every second day. Then forecast distributed with the local dark store team to provide feedback, so to understand if everything was correct from a number point of view and also to understand if quantities were correct in terms of space needed to stock them up. If the system is not so sophisticated in terms of dimension, because products, they might change over time so you might have different products, different packaging, then you change your catalogue, so it’s not super precise in terms of dimension, you need to provide some physical feedback as soon as then the order is validated. All warehouses across the network, they provide feedback, and then that feedback goes to suppliers and orders are merged and then delivered by suppliers in the central warehouse. That’s how we arrange it.

Some countries, some other players, they merge the category selection and supplier negotiation, but then they also negotiate with suppliers to have direct deliveries to every single dark store. It might be one dark store receives, every day, 20 different suppliers with different products every time of the day and you need always somebody ready to receive them to quality check, count, any inbound activities. That can be a little bit messy, and, of course, cost of transportation will affect your procurement cost if you don’t collect it and merge it together and do some optimisation.

Back to top
Once a customer places an order, can you outline the process by which the order reaches the dark store and is then executed? What is that supply chain?

SP: The customer places an order, selects products from the application, pay credit card, order gets received on a nearby dark store, then we use some scanners, so scanners, they have a sound alarm or they see priority list. You see a new order coming in, then the picker, which is the employee assigned to fulfilment activities, sees the order, sees the list of items, which are sorted according to the routing inside the dark store, and then accepts the order, it means order received, and then he starts to do the picking. He gets a shopping cart and then paper bags and he starts to find items on shelves and refrigerators, gets them from every single location, scans them to make sure they are correct, so we use product barcodes to make sure they are unique and we don’t do picking mistakes, and then he also has alerts if you need to pick, multiple times, the same item and so on. Then he collects all items on the shopping list. When the order is ready, shopping bags, paper bags are full, then he closes the order.

It will have to be assigned to a rider at this stage, so then the rider itself can scan the shopping bags, which they have a unique QR code which links their account, so rider account, to the customer order number. When they scan, they accept and they are ready to deliver the specific order. They can scan one or multiple orders if they need to group them together if they come more or less at the same time and orders, they go in the same direction, also if backpack space allows to carry more than one order. They can scan up to three, usually. More, it’s complicated. They assign to themselves this order, they see all customer details, so name, address. They can call customers through triangulation system, so they don’t get the phone number but they can directly contact, on the phone, customers. They get navigation details and then they can ride to the customer, ring the bell, go to the doorstep and deliver the order to the customer and go back, or deliver another order nearby.

Back to top
Where are the major bottlenecks within the dark store supply chain that are restricting the overall timing? The 10-minute proposition requires a high amount of efficiency and automation, so how well-automated is the process?

SP: I have to say, picking process itself has never been a bottleneck because it takes 1-2 minutes to prepare orders. On average, they have 10-15 item, so it doesn’t require a lot of time. The main bottleneck to keep the promise of 10 minutes is to have riders available when you have peak times. In this business model, peak times, they come, from our experience, lunchtime and most likely evening times, so you have a few hours during the day with low amount of orders or stable amount of orders and then 2-3 hours in the evening when you have a lot of orders. To make sure all customers are fulfilled in the same amount of time, you need to really be precise on scheduling, so scheduling the right amount of people according to how many orders you expect to deliver. To make sure 10 minutes, you always need somebody available in dark store ready to go when you have orders, and then if you know from one hour to another you’re going to have double, triple amounts of orders, you need to do the right staffing at the right time to make sure you fulfil the promise.

Back to top
What is staffing like within a dark store? What are the peak hours where you typically need the most staff, and are they flexible, so part-time, or are they full-time employees?

SP: It depends on the operational hours and how many days a week you operate, and then you can arrange contract hours according to business needs. For example, what we need in Italy, operating seven days a week, 08:00-00:00, so 16 hours a day, we needed basically two shifts to cover the whole day. Mainly, you need supervisors, so you can have managers or supervisors, somebody responsible for the location which coordinates the other employees, and you always need one person at least to do picking to prepare the orders. Then you can have additional staff on specific time slots when you need to do inbound activity, because you need extra hands to basically move boxes, count and put stuff into shelves and fridges, especially when you do refrigerated or fresh products, because they cannot stay outside a refrigerator when the truck unloads them. Then they need to be stocked up quite quickly, otherwise they deteriorate. That’s the basic, let’s say, warehouse employee staffing, but you might have an additional picker in peak times in the evenings or a specific day. A weekday when you know people are ordering more, then you need extra support or some overlap between shifts. You can arrange it through full-time, part-time contract, according to business hours.

The most complex thing, it’s riders. That’s something we learn over time, it’s way more convenient to have only part-time contracts, so you can do shifts from a minimum of 3-4 hours up to six maximum. Full-time doesn’t make sense because you will not have a stable demand of orders through the whole day unless you do some sophisticated strategies to, I don’t know, push discounts on a certain hourly slot or depends on your assortment. If you focus on ready meals and you are in an office area, you’re going to have more orders at lunchtime. If you don’t focus on these products, maybe the curve is more flat, but, of course, a lot of orders, they come at the end of the day after 18:00 when people leave the offices and they’re ready to shop or they just are on the way home or they’re already home and they order some grocery shopping.

To understand how to properly staff riders you need to know the numbers, KPIs and metrics of productivity or productivity you want to target. If you know one rider is able to do four deliveries in one hour, according to how many orders you expect in each time slot, you’re going to staff the amount of riders you need to reach that metric or get historical data that will cover that kind of specific volume.

Then it depends on staffing employee models, so if you have directly employed riders, then you might know you have some inefficiencies because they might get sick, they might not show up, they can be on holiday, so you need to plan a little bit of overstaffing. Of course, with directly employed riders, you have more control, you can train them, you can have loyal employees, they’re better performing, maybe more motivated, and they perform better quality service to face customers overall. If you want to stay super flexible, you get freelancers, and maybe they’re not directly located in dark stores but they need to be called to do pickup, so it’s going to take a little bit longer. Of course, in this case, you can control cost because you might pay by delivery or you can also outsource to third-party, the logistic companies, where they can provide rider-as-a-service for a few hours or they can provide you delivery service for one single drop or, I don’t know, you can negotiate for multiple packages. It really depends, the staffing, according to the employment model and has to be forecast on order volumes and productivity targets you want to achieve.

Back to top
How do you assess players’ inventory management software and forecasting software or abilities? Can they do accurate forecasting or is there a lot of mismatch in the market?

SP: Forecasting is not super complicated, you don’t have, usually, huge spikes, big surprises. Grocery shopping, as a shopping habit itself, it’s stable. As soon as you acquire customers, they like the service, they come back regularly, so it’s easy to plan, unless you do massive marketing activities to push acquisition or heavy discounts so your orders double from one day to another, or, I don’t know, when you have tough weather conditions, you might have an order spike, but then it’s something that can be stabilised over time.

Ideally, here the sophistication, in terms of technology, knowing the amount of orders more or less and forecast you want to have as a growing trajectory, technology will help on the automation to assign orders to riders and group them together without having somebody to do (inaudible 44.14) interpretations. A software that tells you if you can group one, two, three orders together, which one you have to deliver first, or a software that can do a forecast and understand that if one order arrived now, at 12:00, you don’t go out immediately to deliver it because he knows, on average, in a range between 12 and five minutes, you’re going to usually get 2-3 orders and you might be able to group them together. Instead of going and doing hub-and-spoke system, so one delivery at a time and back, if you wait a few minutes to receive a second order, which most likely goes in the same direction, you can serve two customers on a smaller range. Also, in this case, you need technology to dynamically inform and update customers on expected time of arrival, so something providing a range and updating over time to make sure you provide full transparency and visibility to customers about where is their order and when they’re going to get it in a matter of a few minutes.

Then other technology can be added for inventory forecast, replenishment, as much more sophistication you get and precision on replenishment, so inventory and forecast to place orders to suppliers, so you get a lot of optimisation. Especially if you can include some also expiration date track record, because what you don’t sell and deteriorate goes out of the inventory, but you need to know that you threw it away and you didn’t sell it. The system will tell you, “Don’t buy this because you’re not selling it, it’s just wastage,” so you get this kind of sophistication and they have a lot to get through the economics.

Back to top
How are players dealing with wastage? Is there a benchmark maximum wastage amount per week or month that is acceptable, and how do you assess players’ abilities to minimise wastage in the current environment?

SP: Of course, you want to target zero, but it’s fairly impossible. Even big supermarkets, they range between minimum level of 2-3%. It depends also if you do food product manipulation, so if you do some packaging or you need to cut and prepare food, which is something that it’s more like a sophisticated service, more premium service, it’s something that generates waste and it’s complicated. To minimise it, everybody should target something between 2-3% at the end, on a daily, weekly, monthly basis at the end, otherwise the model is not sustainable. To minimise it, it’s mostly a strategy related to the assortment. Then you need to choose if you want to have a higher selection of fresh products, because fresh products are the only ones, mostly, 99%, generating wastage. Also ambient products, if the package gets broken, deteriorated, you cannot sell it, but mostly likely you throw away items that are fresh and they expired or that they deteriorate because they’re not properly stored in the right conditions. To minimise waste you need to make sure you have right conditions in-store, so right temperature or humidity level, make sure everything is preserved, even lighting. Dark store structure helps to better preserve items because you don’t need to have customer visualisation, you don’t need to put lights or, I don’t know, display products to be hand-picked by customers, but you can do some optimisations in terms of temperature, humidity and lighting to make sure they have a longer lifetime. Then you need to really be careful in terms of selecting fresh products, if you select seasonal products that are not very high rotation, you do a lot of wastage, so you need to make sure it’s a product that sells. You can do some tests, some trials or remove from the inventory as quick as possible stuff that doesn’t sell and generates waste. Those are some quick adjustments that can be done to minimise it.

One last thing to minimise waste, if, in the central warehouse, you do pre-quality check and you return items to suppliers, so you make sure all products that get delivered to dark stores are already preselected.

Back to top
Can you outline the quality control process that’s being established across the supply chain and value chain? How are players ensuring there is a degree of quality control and assurance?

SP: If you have a logistic model with a central warehouse where all orders, they get aggregated and suppliers, they deliver to this single location, first quality check can be done there, so checking quantities, matching invoices and all documents, and also sample quality check can be done, so you do rejections, return in case quality standards are not compliant. Then when items are prepared to be delivered to dark stores, it’s a second step, because you have to get single carton boxes, single boxes’ items to be prepared to be delivered, and then a third step, it’s done on the inbound process. When pallets or rolls are received in a single dark store, inbound process, make sure they are counted, properly scanned and stored on shelves and fridges. You unpack all boxes, all packaging, it’s thrown away on a single item, then you’re going to do quality check, not one by one, but mostly it’s very accurate because you don’t have 10,000 units of products every day. You have 2,000 units and then, as soon as you open boxes, you can do a very deep quality check. It’s done pretty much manually by the operators. Scanning to make sure unique codes are the ones supposed to be received, so matching shipping level with items physically received, it’s a quality check to make sure inventory numbers are correct and invoiced and they get paid for what has been ordered. Quality check itself for quality standards, it’s done manually by the operators according to training and knowledge of products.

Back to top
Are players purchasing the land for a dark store or is it leased out? What are a dark store’s initial set-up costs?

SP: As soon as you rent a location which is suitable to open a dark store, you need to probably do some restoration, so make sure the location can be suitable for these kinds of activities. When it gets empty, most likely you will have to upgrade the electrical system, if the one existing, it’s not that new or not powerful, because you need a lot of power for refrigerators and air conditioning. New electrical system, then set up of air conditioning. Most of the locations, they don’t have a proper air conditioning system because then, as soon as you have refrigerators, you need to keep a certain temperature inside the store, but also when you have refrigerators, they pull out hot, warm air from the back side, so then you need to have, all the time on, air conditioning. In some countries, in Italy, for example, it’s mandatory by law. Some countries, they don’t require it, for example, Netherlands, where the weather is not that hot. More longer wintertime, long winter months, but summertime they are not required, but it can get quite hot in certain locations, so I strongly advise to have controlled temperature with air conditioning. We try to not do construction, tearing down walls, everything that becomes more complicated in terms of timing, but also you need concession, documentation if you change the official layout of a single location.

Usually, everything can be done, restoration, construction if needed, installation of fridges and refrigerators in a range between four to five weeks max. On average it costs, everything, like also to buy CAPEX, so CAPEX you call refrigerator, air conditioning and shelves and installation, something between EUR 200,000 and EUR 250,000. We decided to buy refrigerators, some other players, they can decide to lease them. Then it comes probably a little bit higher price provided by better flexibility.

Back to top
Can you outline the OPEX associated with running a dark store, such as labour, electricity and utilities?

SP: Numbers are changing now. Utilities, usually it’s not a big deal, it doesn’t make a big difference overall in the P&L, we’re talking about a few thousand euros on a monthly basis for electricity cost, and you have a fixed cost for wifi connection, that’s even less than electricity. You need to consider monthly rent for vehicles. Some players even decide to buy vehicles and have technical assistance also hired from the company to make sure they’re always available. We decided to lease them and make a contract with maintenance fully included, that’s something, comes a monthly cost for warehouses. Then electricity bills, what else? Payment fees I won’t consider as a dark store cost because it’s like a platform fee, so credit card fees and so on. Then the big chunk of costs, it’s mainly salaries, so riders and warehouse staff that does fulfilment, inbound activities, so managers, supervisors, associate pickers responsible for that location. You have consumable products like paper bags you use to pack the orders or, I don’t know, some other stationery, these kinds of things, but it’s not very evident.

Back to top
Your comments suggest this is an efficient model with a great runway, but what are the core difficulties or issues around operating a dark store model, especially given the density required?

SP: As I said, it’s a trade-off between catalogue, as soon as you find the right catalogue and make sure customers, they buy regularly, it’s something that’s not impossible, not complicated, because customers, they really get spoiled and addicted by the quick delivery services, so you need to make sure to be consistent in terms of timing. As soon as you identify customer target and the right range, the complexity is to make sure you reach delivery profitability because all the other costs of which, you pretty much fix. Of course, there’s complexity because you buy inventory, you hold it, you have inventory risk and so on, but, as soon as you find enough rotation, it makes sense to buy outright your inventory and make sure it’s safe to be sold. Then you need to have the right amount of volumes and you need to get profitability through the delivery fleet. It means number of hours, number of deliveries done in one working hour, so you know the cost of a working hour and then you know much does it cost to fulfil a specific order, and then you’re going to see if it makes sense according to your average order value. That’s the only way to make it sustainable, to make sure there’s a break-even at some point, and then it ranges, of course, between number of orders, average order value and how many orders you can deliver in one hour with a single rider or if you do outsource or other flexible models.

Back to top
Which major markets are the most challenging for the dark store model to succeed in, and in which markets is it more viable?

SP: From the experience in Europe so far, more challenging markets are, I have to say, unfortunately, southern Europe, so Italy and Spain, because purchase power is less developed than Nordics or more developed markets in Europe. A little bit less digitalised, also less used to buy online these kinds of products. Then it really depends, I will not say markets but cities itself, because every single shop, every single dark store has to be sustainable itself, so it makes sense to open a single city. I will focus on a city level and not country level because you can’t even open one new city without having the service available in the whole country, it just doesn’t make sense. As soon as you make sustainable one city or a single shop, to open a single shop on a huge city probably doesn’t make sense because you don’t have synergies, you don’t have network. You need to cover and reach a certain amount of customers with multiple numbers of dark stores in the same city also, because there you get optimisation with logistic distribution from a central warehouse eventually.

I would say the most complex markets are the ones without density, so, I don’t know, if you think about the Netherlands, if you go on second-, third-tier cities, maybe they don’t have enough density and delivery area, delivery distance has to be too high, you need to go too far away, you don’t get optimisation in terms of delivery cost. Then, yes, some markets are more competitive because, of course, you have a lot of players, so they might be challenging, could be London, London lots of players are focusing on this kind of business, but, of course, more competition means that also it’s a more attractive market overall. Then, yes, it’s about density, if it justifies opening dark stores, and then also one thing to be considered, it’s pricing. Some markets in Europe, they have very similar pricing between quick commerce, offline, so, from a customer point of view, it doesn’t make any difference, or it’s a very small difference to pay a symbolic delivery fee and get exactly the same products at the same price delivered to you instead of going to do shopping offline by yourself. In Italy, for example, even Spain, similar, products listed in the app are slightly more expensive, there’s a small mark-up. Customers that are very price-sensitive, very conscious about what they buy on a daily basis, they will perceive it as a premium service, so more luxury habit, because, at the end, you spend more, even, overall, 20-30% more than going offline, so they will use it once in a while, they will not substitute the daily shopping experience. That’s also a trade-off. It depends on purchase power, negotiation, the buying team’s opportunity to get margin negotiation on a specific country, on a local market.

Back to top
The information, material and content contained in this transcript (“Content”) is for information purposes only and does not constitute advice of any type or a trade recommendation and should not form the basis of any investment decision.This transcript has been edited by Third Bridge for ease of reading. Third Bridge Group Limited and its affiliates (together “Third Bridge”) make no representation and accept no liability for the Contentor for any errors, omissions or inaccuracies in respect of it. The views of the specialist expressed in the Content are those of the specialist and they are not endorsed by, nor do they represent the opinion of, Third Bridge. Third Bridge reserves all copyright, intellectual and other property rights in the Content. Any modification, reformatting, copying, displaying, distributing, transmitting, publishing, licensing, creating derivative works from, transferring or selling any Content is strictly prohibited