A more positive approach to returns • 16 July 2008 • The SnowBlog

A more positive approach to returns

It's all very well me going on about how awful returns are. But so far I haven't offered a solution. Here is what I would do if I ran a bookshop chain. (Warning - I haven't written the post yet but it's bound to be long.) (I've written it now - it is.) A lot of what follows comes down to learning from other retail sectors. The history of stock control and forecasting is long and (if you're me) interesting. In the 1980s, many of today's leading retailers were starting their careers at the Burton Group, Mars and P&G, where they learned what at the time were cutting edge supply chain and merchandising processes. Several such processes, which have been standard in high street retail for many years now, concern forecasting and stock control. To forecast how much stock they'll need, supply chain managers group products according to their sales profile: 1) core stock - products which stay on the shelf week in, week out. Think Stephen King's backlist, the Dummies series, OUP's Very Short Introductions, Maeve Binchy, etc. 2) new products - those without a sales history. 3) promotional products - those which will be introduced in store and sold within a short period of time Further layers of complexity get introduced when it comes to the stores themselves: 1) Store size and shape. Not all stores within one retailer are not the same size and shape. Retailers who grow through acquisition are especially susceptible to having stores, as well as fixtures and fittings, of different shapes and sizes, since the old shelving tends to get used - it's jolly expensive to refit 100 stores in one go with standard shelves. 2) Store geography. A store in Torbay has a very different customer profile to a store in Covent Garden. I was extolling the virtues of our horror novels to Matthew from The Torbay Bookshop yesterday who gently told me that their core customer base is approaching or in retirement age - not obvious lovers of American vampire novels. 3) Position in town. There are two big DIY stores in Morden, South London, but one is on an out of town retail estate and one is in the town centre. They serve different shopping needs for different customers. So you can have neither the same range nor forecast for all products in, say, a 30,000 sq ft store in Covent Garden and a 7000 sq ft store in Carmarthen. That's fine, though, because you develop ranges. Ranges are typically planned and communicated to stores via planograms - a drawing of the shelf fixture, with the products indicated. Here's a 12ft, 8ft and 4ft planogram for some power tools. plano.jpg The 4ft planogram contains only the lines with the highest sales and margin forecasts. It makes sense that if you have restricted space, you only include products that are going to give the highest return. On the 8ft planogram, you can include more products with more modest forecasts. On the largest planogram, you have more space to include the full range of products. (As an aside, in our previous roles, Rob and I lost count of the number of times a buyer said "oh, I need to include product X for 'range credibility'", by which they mean you need untested, sexy products that the buyer really likes in the range to give shoppers a sense that the retailer is an expert in this field. This has been proven again and again to be nonsense. There've been many studies which prove that the fewer the products in a range, the greater the perceived choice by the shopper. Too many, and it's brain overload - you just can't process all the different choices. This is where packaging and genre clues come in, but that's a cover design discussion for another day (or indeed, a previous day, as we've blogged about this before.)) So if planograms are based on projected/forecast sales and margin, how do retailers come up with those forecasts? And how do those forecasts translate into actual orders? Forecasting With lines that have a sales history, it's easy. You look at what's sold before, adjust for availability and seasonality and assume that the following 12 months will look pretty similar. Adjusting for availability means if sales were 1000 units per month except in July when they were 50 units, you check that you actually had some stock in July. If you were out of stock, that's a clue that July's sales would normally have been 1000. Adjusting for seasonality means that if the sales graph is a nice bell curve peaking in May and dropping to zero in November, you're probably looking at a barbeque or a summer dress. You usually have to adjust for Christmas and Easter bank holiday, as well. Buyers will often try to adjust for made up events like The Diana Effect (the UK marked its mourning for their princess by not shopping for a couple of days), The World Cup Effect (everyone was drinking too much to shop) or The Weather (the catch-all excuse to get buyers out of trouble if sales are poor) but you shouldn't listen to them. For lines that are new, and promotional, you might think that there's no data available so it's finger in the air time - but no. Contrary to what manufacturers (and I'll include authors in that) believe, their product (book) is not unique. There is always a similar product out there that retailers can learn from. I remember a meeting at Superdrug when I was Men's Toiletries assistant buyer. Gilette were launching the Mach 3 and were adamant there was nothing else like it. Well, correct, there wasn't, but we used the Sensor Excel launch sales profile with an increase of 7% and were exactly right - we had 92% availability and only 8% stock outs. That's notable in itself - you don't want to have 100% availability, and you do want to sell out of stock. If you have 100%, it means you have far too much stock. If you sell through, you've not got a load of stock left over to sell at discount or - ah, we come to the point - return to sender. Fashion retailers manage far more new lines than book retailers. Take Zara - they seem to have new stock in every week, which they sell through. They are vertically integrated - they manufacture their own goods - so they can't rely on returns. Instead they hedge their bets by delaying manufacture till they see what sales are like. Say they list a wrapover dress. They will cut the cloth in various sizes and colours, make up a certain number of each, send them into store and see which sizes and colours sell the best. They'll rapidly make up the waiting stock and ship out only the bestsellers. Orders So once a forecast is agreed, how are the orders calculated? Various algorithms can be used, but here's one normal approach. Orders are divided into two different types: store and warehouse. (Some retailers use direct to store deliveries, but this is very old fashioned. Oddly, Borders are just closing their warehouse and moving to store direct delivery. I can't say I quite understand this.) To calculate warehouse orders for an ongoing line, the analyst looks at the average weekly sale for the last 4 weeks, then at the number of week's cover they'd like, then at the agreed lead time with the supplier. So if the average weekly sale for a line per store is 13 units, and there are 100 stores ranging the line, and the stock level is 2300 units, you have 1.7 week's cover. If your lead time is 5 days, you might want to place an order for 400 units, to take you up to 2 week's cover which is your target. Any more and you have quite a lot of stock - and that's bad for the retailer's balance sheet (especially if it's coming up to year end). Once the stock is in the warehouse, you can pay attention to store orders. Many lines are on autoreplen, or automatic replenishment. This means that a computer program does a quick calculation based on the average weekly sale in store, the presentation quantity (how many units you want to have in store at all times to make the fixture look nice - usually two units), the required weekly sales cover to be held in store (usually 1.5 weeks if a store has two deliveries a week) and the weekly sales forecast as entered by the stock analyst. The analyst will scan these automatically generated orders to see that nothing's too amiss (for instance if a product is coming off promotion the analyst will pay special attention to future orders - you don't want to base them on promotional sales). For promotional and new lines - say, for a Christmas range - it's very important to stagger how stock gets into stores. Usually the team will agree how the stock allocation will be weighted - say, 40:20:20 (Sept Oct Nov) with the final 20% left on hand at the supplier to call in if sales are on track. There is also a planned sell through calculation. Most retailers aim to sell through (in other words, sell out) of their Christmas range by mid December, which gives them time to prepare the clearance paperwork to send to stores. They are perfectly happy to have some lines that have 90-95% sell through by early December. This is a different mindset to a lot of book retailers. The key to managing stock levels like this is communication with the supplier. Retailer and supplier together agree what the lead time is, what the supplier's reserve stock levels should be, what the sales forecast at store and warehouse level should be and what the exit plan is. Perhaps the book retailers do this detailed category management planning with the large publishers, but it doesn't happen for smaller suppliers. We receive orders rounded to the nearest thousand or hundred - always a warning sign. Let's be clear - this isn't the buyer's problem (although, of course, it impacts on their days). This is an infrastructural problem - seems to me that book retailers are missing a couple of functions that have been normal for ages in other retail sectors. Those are a pricing function (because in our industry it's the manufacturer that sets the retail price), and a collaborative planning, forecasting and replenishment function (CPFR). I also get the sense that the IT systems of book retailers are less than fully integrated across all functions (which Borders appear to be addressing at the moment). This isn't that unusual - systems often grow organically, with one chap in the IT department frantically lashing together the old IBM AS400 mainframe that they bought in the 90s and adding a data warehouse here, a bespoke planogram tool there until it's a bit of a mess. That chap then usually gets a really good job offer from PriceWaterhouseCoopers or somewhere, which means that next time a bit of the system starts to unravel the company goes into meltdown. Usually happens around year end, when you start to hear heroic stories on the grapevine of people sleeping under their desks for days on end to manually haul the business back into repair. To say that I am glad I'm out of this picture is something of an understatement. Summary I started this by saying 'this is what I'd do'. To summarise, then, I would: 1) Introduce standard range planning systems, to include planograms mapped to sales and margin per linear foot forecasts, including range planning for seasonal promotions, not just core stock. 2) Ensure that an autoreplen system was working well and had the correct assumptions in it - like supplier lead time and agreed weeks' cover - and that all products were assigned a sales profile based on either their own historical sales or historical sales of a similar title. 3) Establish CPFR workgroups with suppliers to manage the product lifecycle. This would be to get the right level of availability, to keep the supplier appraised of how much stock they should be holding for the retailer to call off, the agreed lead times, and the promotional plans for the next 12 months. I would also revisit the convention of having pile-it-high tables of stock at the front of store. I would make these tables browseable but not like bargain basement tables. To satisfy shoppers' need to rummage, I'd have dump bins, but would reserve the tables to be dressed - think about the tables in Laura Ashley's Home stores, which give the shopper a beautifully merchandised taster that leads them into the rest of the store. This would return the book to icon, special product status rather than the old Tesco pile it high model from the 60s. I am also going to pinch one of Rob's ideas now, and propose that my chain would have one or two Clearance stores - like Next have. I would ship all overstocks to these stores and discount them heavily. There's a Books Etc store in Bicester Village like this, near me - but the stock seem to be damages and special purchases rather than overstocks. This would result in overstocks generating a bit of money, rather than being returned to the supplier for zero gain. Finally... I would love to talk to other independent publishers about this sort of thing. It could even be the case that we might want to form a bit of a group - Independents For Forecasting would probably be better than Independents Against Returns (purely from the acronym point of view, IFF is better than IAR). As I say, perhaps the large retailers and publishers do get together to plan their stock properly - but since we hear tales of high returns from publishers large and small, there does seem to be room for improvement. If you'd be interested, say so in the comments. I'd also love to hear if you have any further things to add to how we can together manage stock control - I am aware that this post only glosses over the surface of the subject and there are lots of omissions (I haven't even described category management). Come, one and all!


The SnowBlog is one of the oldest publishing blogs, started in 2003, and it's been through various content management systems over the years. A 2005 techno-blunder meant we lost the early years, but the archives you're reading now go all the way back to 2005.

Many of the older posts in our blog archive suffer from link rot. Apologies if you see missing links and images: let us know if you'd like us to find any in particular.

Read more from the SnowBlog...

« Sliding Doors
Garbled »