The old fisherman, X5, cast his net for the third time… or
Sketches of a Workable E-commerce Model in Food Retail

Like we can’t stop Mr. Kobzon as he sings, there is no stopping the X5 Comrades in their eagerness to keep chewing cactus on an industrial scale. Meaning, to keep dumping new tens of millions of greenbacks, which are going up in price, into obvious screwjobs.

Against the bullshit that constitutes 80% of the article referred to above, about X5 Retail Group venturing into e-commerce once again, the last paragraph stands out as something quite reasonable. Characteristically, the speaker in it has nothing to do with X5.

It does not take a Sherlock Holmes to guess what will come of their endeavour. Our pre-emptive condolences go to the old chaps from Alfa Bank. Maybe veterans do stay young at heart (like in that Soviet song), but to step on the same rake for the third time… In short, we send rays of compassion to Mr. Fridman.

And now let us look at the food retail e-commerce venture in general.

We do not know a single successful (= profitable) undertaking of this kind; online stores of foie gras, elite cheeses and teas, liquor, etc. do not count.

For a long time we believed (and repeatedly wrote) that the task had no solution. As Mr. Trout said, you can’t get there from here.

But now we have a vision of a business model that has some chances of being viable. Here are some sketches of it.

We mean a hybrid of a network of ‘next door’ shops and a big distribution warehouse on the outskirts.

‘Next door’ shops’ assortment matrix is quite ordinary: a basic range of staple foods meeting one’s daily needs. Including perishable goods, of course. The pricing policy is also the usual one for this format.

All the rest is stored on the hub’s premises. Assortment breadth is limited by the resources, ambitions and common sense. The pricing policy is that of a discounter or even super-discounter.

Visitors to the shops are cultivated: the traffic, natural for ‘next door’ shops, gets marketing messages that suggest the behaviour pattern of ‘what we don’t have at the shop, you’ll order online tomorrow morning and collect as you return from work in the afternoon. Forget Auchan, our prices are even lower.’ This just technical, but the chances of success are splendid in this case.

Incidentally, the prices will really be lower than Auchan’s, as the typical consumer will compare the hub’s (discount and super-discount) prices – and what is already lying at the shop isn’t worth comparing. What do they have there? Salt and yoghurt – come and buy. And the hub’s costs will be fundamentally lower than Auchan’s, so the latter will eventually lose.

And you needn’t go to Auchan or Metro and stand in lines there. Everything is next to your home. Or can even be delivered to your door.

Logistics layout: A placed order is compiled simultaneously as two sets, in your hub and a store you need (cross-docking)

1. In the case of delivery. The delivery vehicle loaded with half-assembled portions of the orders arrives at the shop (each shop is the centre of a delivery sector), collects the orders’ complementary portions there (packed in standard bags with bar codes for ease of processing) and takes them to addresses in the sector as usual.

2. Pick up. The delivery vehicle leaves the hub’s portions of the orders at the shop (packed in standard marked containers just the same way) where the buyer collects them as he/she visits the shop. It can be a hands-off process, e.g. the buyer goes to the corner full of bags, chooses his/her own by the number, and goes to the counter; the cashier beeps on the bags’ bar codes, and the amount is registered automatically. Still another technical thing.

This is how we work around the fundamental problem of costly processing of cheap goods. Both cheap and perishable goods are processed cheaply in a (nearly – see below) traditional convenient store format.

Besides, your customers can enjoy combination of superb pricing (moreover, the one for those goods when such prices do actually increase sales rather than constrain your profitability) and convenience.

If we look a short way behind, the Utkonos (Platypus) company was after something similar as they were trying to arrange sales right at their online store outlets. But as those duck-nosed Comrades were reasoning from the contrary and bluntly trying to reduce their losses from keeping the pick-up points (like the Russian Post tries to sell postcards and soap), nothing good came of it. On the other hand, it’s their birthmark; see what a real platypus looks like.

In the home delivery case, the drivers can be liberated from receiving cash – for their security, quicker execution and lower costs; the orders will be prepaid on the website. But it will not restrict demand in this case, as the people ordering food online for home delivery have plastic cards and can use them.

The above scheme works smoothly only if the whole network operates in a single information space with online transactions. (As opposed to the ordinary prehistoric scheme when the shops are equipped with POS terminals or local 1C workstations, the main software runs at the headquarters and the whole thing is ‘synchronized’ at night, not without faults, of course.)

Where the single info space condition is met, virtually unlimited functionality – from ordinary (=antiquated) food retail’s perspective – becomes available, including, but not limited to:

• shop assortment being replenished ‘on the fly’. All the sales and inventories at each shops are visible in a single system online. For example, you can send replenishment vehicles from the distribution warehouse every three hours, loaded just before dispatch taking into account each shop’s inventory at the moment. Up-to-the-minute orders to suppliers can be generated just as promptly. The need for working capital is reduced by a big ratio. In brackets, we should say that if you use a minimum reasonable approach, your working capital will go negative (and this is NOT mere words);
• efficient use of electronic price tags at the shops. Unlimited freedom of pricing; you can alter your prices every five minutes, at all or selected shops, and apply business logic of any complexity: e.g. raise the price automatically if the goods item is running out. Reduce your pricing automatically if some inventories are nearing their expiry date. If your supplier’s purchase prices have changed (in automatically imported price lists), the retail prices at your shops will change immediately. In particular, this eliminates the monkey business of re-writing and arranging the price tags, with its resultant faults.
• An MLM-style bonus loyalty system. Actually, an adapted version of the XXL Bonus that has performed perfectly well at Ulmart. In the case we describe, its expected efficiency is higher by an order, for the gross margin is higher, so much more interesting bonuses can be granted – as food is bought somewhat more often than notebook PCs. This results in greater consumer involvement, and a convenient store’s focus on limited geography will multiply the efficiency of communication between your target audience members (e.g. your lady next door joining the next MLM level);
• elements of artificial intelligence in ‘actions’ design based on the analysis of dependencies (known as ‘big data’), e.g. if you buy XXX beer you get HHH breadsticks at half the price (at the vendor’s expense, to be more accurate); in respect of a specific buyer (on your birthday, get your personal 30% discount for XXX sparkling wine); or as any arbitrary combinations no matter how complex (a 20% discount for charcoal, vodka, and meat if bought together on a summer Friday evening as part of a total purchase worth RUR 7000 or more). If it is raining outside (according to data imported from weather sites), a 5% discount for umbrellas can be granted. Et cetera, et cetera. The actions are designed at the head office – both by the system robot and by the category managers, and broadcast online at the shops. No shop personnel’s involvement is needed at all; the business logic runs automatically on the cash registers, and the buyers are informed via LCD panels. Voice announcements can generally be auto-generated, too;
  
• a development of the foregoing. All those ‘actions’ can be generated for each buyer individually, ‘on the fly’. If Vasily buys cigarettes regularly, why not offer him three lighters at the price of two, once in a couple of weeks? The rules of forming such simple ‘actions’ can be described in the system’s business logic (based on links among related goods which can, again, be identified through regression analysis of the sales data, and subject to margin sacrifice limits) and fully auto-generated.
• more intellectual things for which human brain is indispensable for now. Like the classical story from big data tradesters’ stockpot: if Martha has bought no towels for more than two months, let’s try to offer her some maternity wear.

Now a few concluding remarks about ‘actions’. These can be communicated to the customer in the classical fashion – contextually, as the order is booked on the website, or in a hi-tech manner: at the shop itself. The discount card is equipped with a RFID chip, so the system gets a signal when the customer comes in. We know already at which hour the specific person called at the shop. Then the client passes e.g. into the beer section. The RFID scanner reads his location, the system generates a contextual marketing offer ‘on the fly’ and sends it in a SMS/MMS message (which is not elegant) and/or displays it on the LCD ad screen installed in that section. Those same RFID scanners give us a full picture of the buyers’ movement around the shop broken down by time and linked to the final receipt. A good merchandiser will sell his soul for that. Fantastic? Not at all. Moreover, those technical joys now cost quite reasonable money.

What we have next:

• A soaring mind
• Catharsis

In conclusion we must note that, besides a single information space at the enterprise, the above requires a highly (and even super-) productive and reliable IT system.