Actionable Analysis of Account-Specific Data
We have heard it said, and all the speakers this morning have supported this idea, that “the devil is in the details, but the dollars are in the details too.” For our presentation this morning, we want to show you how publishers can look at some very particular details — those provided by Barnes & Noble through BookScan about sales and inventory throughout their supply chain — in what we call an “actionable” way.
The reports we are about to see change and enhance the B&N data from what the publishers get in three very important ways:
1. They archive the data, so we are looking not just at this week’s numbers, but at relevant numbers from last week and, with some views, data that goes back as much as a year.
2. They calculate key metrics, some very simple and some a little complicated, to give a relational picture of what the sales mean.
3. They present only the data that is needed for the purposes intended by a particular report to avoid distracting a decision-maker with irrelevant information.
We have found that the right reports make competent business analysts out of reps whose great strength had always been that they were good “presenters”, not good numbers people. And those who were already good with analysis can save time if they are working with the right reports. They find what they need to know faster and, perhaps more importantly, have a vastly simplified task presenting “evidence” to a buyer because they can present a summarized report of the inventory performance for every book.
This morning, let’s look at two particular analytical challenges facing every B&N rep and how our reports help reps tackle them. One is finding books that are moving, but under the radar; selling relatively well for the investment allotted to them but, because perhaps they were distributed lightly, not achieving numbers that will get them noticed for increased distribution. And the other is responding to the opportunity of a backlist inventory review for any store section.
The first one is to quickly spot the titles that are selling well, but flying “under the radar”. There are two qualifications we need to define there. How do we measure “selling well”? And what’s “under the radar”?
Here you see a sample, from a very large publisher, of our weekly “Flash report.” We have anonymized the titles and authors for obvious reasons. The Flash report, which is largely, but not entirely, based on a single week’s data, identifies the titles that sold the most in relation to their Superstore inventory! The cutoff we use for “hot movers” for most publishers is 10%; we trap all books that sold more than 10% of their Superstore on-hand last week.
A quick rule of thumb here — each percentage point of sales against inventory in a week translates into about half-a-turn a year. So books that sell 10% of their on-hand in a week are moving at about 5 turns a year. I’ll be glad to go over this calculation in more detail later if anybody asks.
So that you will be able to see what happens when you view by everybody’s favorite metric — sales last week — we did our first sort of this publisher’s Flash report by that number, and then anonymized the report by giving each book a number corresponding to its ranking by sales last week. This slide is sorted by Superstore on-hand. You’ll see that it doesn’t throw off the title numbering, which is by sales, very much. It isn’t precisely the same, but the point is that how high books rank in sales is largely determined by where they rank in inventory.
Let’s quickly go over what each of these columns means.
The first data column is Superstore unit sales for the week in question, which for this report began on December 6, 2003. Then: year-to-date sales in units, and then the total number of copies modeled in the superstores, which is a B&N term for computer-generated reordering. Then the Superstore on-hand and the Superstore on-order at the end of that week, followed by the Superstores sales for the prior year. That’s followed by the Dot Com’s sales of the title for the year and then Distribution Center quantity on-hand and on-order at the week’s end. This is a total for B&N’s two retail distribution centers, one in New Jersey and one in Nevada.
The next two columns are the key Flash metric columns: the percentage of the Superstore on-hand that sold for this week, plus that Flash metric for the prior week. And the last column is the percentage of the Superstore on-hand in relation to the model at the end of the week.
Just a moment’s explanation of the term “model”. B&N buyers can place a “model” quantity for any book in any store. When they have done that, the reordering of the book up to the model quantity is automatic; it is done by computer without any need for human intercession. The model number on the report is the sum of all the store models for that particular title.
Now, of course, the most important books are near the top when we sort by inventory. In general, books that have more than 2000 copies on hand at B&N are watched by hawk eyes: buyer hawk eyes and rep hawk eyes as well as store manager hawk eyes! They don’t need much help from us. So let’s drop below the radar.
What we call the “opportunity band” are the titles that have more than 50 or 100 copies on-hand at the Superstores, but fewer than 1500. Because this big publisher has so many titles above the opportunity band to worry about, we’ll just look at the titles with 100 to 1500 copies.
Here’s what that group looks like, sorted by on-hand quantity. You can see how the books rank in sales by looking at the numbers we used to substitute for the titles. A weakness in the data set we get today, which we work around, is that you don’t know the number of stores in which activity took place. If there is a large sale at one store — caused perhaps by an autographing, or a class reading the book, or some other local activity — we might see a big percentage sell-through that would be misleading. That’s why we include last week’s Flash number as well. Anomalies can happen two weeks in a row, but they don’t very often.
So since we know that all the books here sold more than 10% of their on-hand this week, or they wouldn’t be here, let’s sort the report by descending order of the Flash metric for last week.
There is a slew of opportunity here, but in the interests of time, I’ll point out only a few highlights. Both the second and third lines, title 89 and title 57 — two biographies — are quite striking to me. Title 89 was published more than two years before this report, but its sell-through numbers are fabulous on a small, but statistically significant base. It also sold significantly in the previous calendar year. It looks for all the world like this book was de-stocked too soon. Title 57 is more recent, not quite a year old. It has an even more significant base of distribution, but still an average of about 1 copy per store, and even heftier sell-through numbers. Neither of these books are modelled; both of them might be worthy of it.
By the way, you’ll also notice that Title 57 has no stock on-hand or on-order at the DC. This can happen with books that aren’t modeled, although less frequently than ever. If it is modeled above 50 copies, it will get into the DC automatically, and that number will soon be reduced.
Now, of course, if either of these is the story of the person who killed the first Thanksgiving turkey, their sales pop could be explained by timing. The rep will know that; and the rep knowledge is, of course, a key piece here. But, absent something like that, these are books that are selling at impressive velocity but below the inventory level that a buyer would normally have time to examine, particularly on books that aren’t terribly recent.
Let’s look at one more pair on this report: lines 8 and 12, titles 56 and 70. Skip over to the last column there and you’ll see that they both have low percentages on-hand in relation to their models, 47% and 65% respectively. If models are right, being restocked from the DC and the DC is properly stocked, that number shouldn’t fall below 85%.
Title 56, you see, has only 1 on-hand at the DC, but 220 on order. Meanwhile, the Superstores have over 1,000 copies on order. Title 70 has has 73 in the DC and 192 on order there, while the stores have orders for 298. In both these cases, it would seem that the model reorders from the stores are going straight to the publisher, which is certainly not the fastest and surest way to get the book restocked. It is also possible, in both cases, that the publisher is short of stock. But, again, the rep would know that. Whatever the case, sales are being lost.
The second challenge we believe our analytical tools help with considerably is in conducting a “backlist review” of any section. This is a precious opportunity that reps get only occasionally with buyers; most no more than once a year for a section. We have the feeling that our tools are enabling our clients to get more frequent reviews, because they are effective, but I honestly have no way to prove that empirically.
To do what is necessary here, the historical data is critical. When you’re looking for hot movers, a week or two’s sales tell the story. Backlist doesn’t move that fast. On a Flash report, you’ll see books that for a week or two are turning at annual rates of five, ten, or more, on signficant levels of inventory. That doesn’t happen very often over longer periods of time.
Here we look at a different publisher at a different time. This is a somewhat smaller one, and the time of this report is the week beginning September 22nd. Once again, we’ve masked the titles and authors and we have also anonymized the store section.
First we sort the entire list of this publisher’s books for on-hand inventory. Let me explain what the headings mean here.
The first six are Superstore sales this week, year-to-date, the Superstore model total, and the Superstore on-hand and on-order, and the Superstore sales last year. Then we come to three stock turn columns: calculated on sales and inventory over the past 4 weeks, the past 13 weeks, and the past 52 weeks. The next column is the Flash metric for the previous week — the percentage of superstore on-hand that sold. Then we have dot com sales year-to-date and that total expressed as a percentage of Superstore sales year-to-date. To clarify, if Superstores sold 100 and dot com sold 50, this would say 50%. The next two columns are mall store sales and those sales expressed as a percentage of superstore sales. The last two columns are the DC on-hand and on-order as of the end of the week.
First of all, we want to look at just one category, so here’s a slide that takes out just the titles in Category A. But, for the same reasons we did with the Flash, to stick to “actionable” titles, we’re going to cut all the titles that have fewer than 100 copies on-hand. After we’ve done that, we sort the books by 52-week stock turn, descending, and this is where that gets us.
Now what we want is books that should have their quantities, or models, increased. Once again, we have a spectacular number of titles that could justify increased distribution. Let’s look at a couple. The very top title on the list, number 168, is turning at an extraordinary velocity; selling 30% or so of its stock a week. I wish we had the pub month here; we don’t. But I’d suspect this book to be less than 13 weeks old because the turn is so high and because the 13- and 52-week stock turn numbers are identical. Nonetheless, this book is certainly more than four weeks old, and at these sales levels, going out of stock in many stores every single day.
The other one I’d call attention to here would be row 6, title 744. You see it has great stock turn numbers, but a model of only about 300 and even fewer in the stores. The low on-hand relative to the model plus DC on-order quantity here would indicate that this is a book where there is a stock shortage at the publisher. The rep has to get the model up, but also has to manage the situation at the publisher so that’s done in conjunction with available stock. If a buyer sees low sales with a high model, they might not do the investigation necessary to find out how much of the blame for that was due to stock-outs.
One other things worth noticing here is illustrated by row 10, title number 187. This book has a high 52-week turn, but is clearly trending down. Sales in relation to inventory are slowing, which we can see because the 13- and 4-week turn numbers are lower. This, therefore, is NOT a candidate for more distribution, even though it has a good 1-year performance.
What a rep would do, then, is highlight all the titles found this way that were candidates for inventory increases and there could be quite a few.
Now we need to look for books that are failing. To do that, we sort for 52-week stock turn, but the other way, ascending.
And now we find the books that are OVERstocked, books with extremely low turns. Particularly those with high models should get immediate attention. So line 14, title number 580, stands out. Turns are low, and a model of 276 is wasted on this book. If you continued to scroll down, you’d find more candidates for reduction.
Now we’ll find more candidates for reduction by sorting in descending order by model. Look at line 13, book number 15. A model of more than 1300 on a book that is turning about one time a year is a bad use of Barnes & Noble’s money. This publisher has much better places for them to put those dollars, as we’ve demonstrated.
If we sort for the models the other way, we can find some that should be raised. We are showing you that sort here for rows 47-60. Look at row 55, title number 216. It has had a strong performance for a year, yet has no model. One would want to know exactly what the book is, of course, but these numbers would suggest that a model could well be appropriate.
There are two other tools here.
This is a slide sorted by dotcom YTD sales. If you look at line 5, title number 291, you see it has a decent model and decent superstore turn. It also has sold 15% of what it sold in stores through dot com. The “average” number there would be more like 10%. This could be a candidate for more superstore distribution based on that indicator, although again it would depend on the book. Would impulse sales go up with more exposure? Or is it more likely to be a targeted book purchase, like a course adoption? The rep would know.
And in the same way, we can sort by the mall sales column. The “normal” percentage there would be more like 3-to5 percent. This publisher has many books selling much better in the mall stores, relatively, than in the Superstores. Is that appropriate? Probably not on ALL these titles. At least some of them would sell more in superstores if there were more copies there.
This overall exercise with the Stock Turn Report leads us to find books where there is evidence distribution should be increased and others where stock should be reduced. Simply cutting and pasting these rows into a separate spreadsheet gives the rep a tool to use with a buyer, and avoids requiring the buyer to go through hundreds, or even thousands, of titles in the section. That’s what we mean by “actionable”. We believe that tools like this help reps help buyers in a constructive way to improve inventory performance to everybody’s advantage.
Two years of experience with our techniques at Barnes & Noble have demonstrated that by archiving data snapshots, looking at them over time, relentlessly indexing sales to inventory, and focusing on books underneath the very top layer, sales reps can provide really useful support to buyers faced with literally millions of decisions about hundreds of thousands of titles in hundreds of stores. Each account does, indeed, have its own unique “supply chain”. The better publishers understand how each account works, and put forth the effort to apply their analytical energy in an account-specific way, the more sales they can achieve for every dollar invested in inventory. And the happier their customers will be.