For fashion and sports retailer Blacks, matching supply to demand
is a fine art and, says Julia Vowler, the critical key to making
that match successful is its datawarehouse
The world of retail is harsh and unforgiving. When it comes to
products that consumers want, if you haven't got them on the shelf
you can't sell them, and if you can't sell them there's no point
stocking them. But what can you sell - or not - and when and where
and why? And have you got it to sell in the right places at the
right time in the right quantities to maximise sales and
profits?
Matching supply to demand is a fine art and for fashion and
sportswear retailer Blacks, an increasingly critical key to making
that match ever more successful is its data-warehouse. Here, the
crucial link between what comes in and what goes out is made,
keeping the supply chain as close to customer demand as
possible.
"It's a juggling act," admits Blacks' management information
systems manager John Horner.
The company wants to keep stock held at its 400 retail outlets
to a minimum so that the maximum amount of space can be given over
to the sales area for the customers to browse, yet the last thing
it wants to do is run out of stock that it could happily sell that
day because it's sitting in a distribution warehouse waiting to be
delivered to the shop.
Even though staff are trained to cross-sell where possible, if
the first requested item is out of stock, running out of anything,
says Horner, "is a potential missed sale". An opportunity to make a
sale is gone, possibly forever.
Tightening the loop between stock walking out of the shop and
being replenished round the back is the datawarehouse, taking in
feeds from the company's operational transactions systems.
"The main data it takes in is firstly the daily electronic point
of sales (Epos) right down to the level of the sales receipt
identification so that we know what people are buying," says
Horner, "and next the distribution centre stock data. All the tills
data is backloaded onto the server at head office so we can do the
stock and distribution management."
Taking the two data sources together gives Blacks an accurate,
timely view of stock flow in and out in great detail.
"On a daily basis we can see what is in stock, and in what sizes
and colours and so on, and at the end of the day we know exactly
what is in each branch," says Horner.
The trick, obviously, is to stock exactly what will sell out,
but no more and no less. But how to know what that magic number is?
Like scientists and historians, retailers look to what has already
happened to find a clue as to what will happen next time. By
interrogating the Epos and supply chain data in the datawarehouse,
Blacks can watch the patterns of ebb and flow forming.
"Every weekend we analyse which branches ran out so we can look
at replenishment levels and build a model," says Horner.
"It's difficult to model," he admits, and the priority is to
select the company's flagship stores and ensure that they, at
least, do not run out of stock.
Each store has a stock maximum for every product sold. "For
example, a shop in Newcastle will have three pairs of trainers, and
if it sells two it receives two more," says Horner.
However, he adds that every store gets a daily delivery from
Monday to Friday. What Blacks wants to avoid is a situation in
which a branch is out of stock after the Friday delivery and before
the Monday one, as it could be selling that product over the
weekend.
"If a store sells three pairs of that trainer on Thursday it
won't get another three until the Friday afternoon, and if it sells
out by the end of Friday it won't have any for the weekend," he
says.
However, he says that if they interrogate the datawarehouse and
then build up sales and stock analysis the company should be able
to build a replenishment model for each branch. Moreover, the
system itself "should know when the branch needs restocking".
At this point, the warehouse moves into that much-desired realm
of operational datawarehouse - not just supplying information for
off-line analysis, but actually feeding back into the transaction
systems and triggering the replenishment systems automatically.
Even so, Horner points out that the warehouse is only refreshed
nightly at the close of business when the transaction systems are
flushed. "We don't know what we've sold until the night," he
says.
Changing over to real-time online Epos feeds could, he believes,
bring disadvantages as well as the advantage of having a real-time
picture of how stock is moving. "What would we do if we lost the
connection?" asks Horner.
He says the whole idea of having automatic stock replenishment
is very attractive from a business point of view.
"Our business people are all for it because it frees up time. We
have about 200,000 lines of products, taking into account different
styles, colours and sizes, in our distribution centre."
Perming 200,000 variants by the hundreds of outlets to ensure
each store has its maximum stock level for each product line makes
the task very time-consuming - and tedious.
"It takes ages and it's error prone," says Horner.
It's also not the most interesting job in the world, so it has
an impact on staff retention. Automating the process would enable
staff "to concentrate on analysis", Horner says.
He believes datawarehousing has a role to play in improving
staff morale.
"I think one of the main benefits that is not perhaps widely
appreciated is that this sort of system can help your user base to
do their jobs better and this can be a great morale booster aiding
staff retention, which is important for long-term stability and
progress."
In terms of the kinds of benefits that company accountants,
rather than personnel managers, like to see - the kind with pound
signs attached - Horner identifies two key winners when it comes to
getting return on investment from the datawarehouse.
"It's saving a hell of a lot of time," he says. "It's cutting
tasks by 50%."
The second major bottom-line benefit is that the company can
avoid being out of stock at key selling times, like the
weekend.
"When we first started we found that around 100 stores might run
out of stock at the weekend. After six weeks it's down to 50. That
means we're making 50% more sales at the weekend - that's an awful
lot of money," says Horner.
Moreover, of course, the on-going analysis of sales trends and
how consumer tastes evolve means that the datawarehouse also plays
a key part in range planning. The information in the datawarehouse
is not just useful to Blacks, but to its suppliers as well, who
obviously have a clear interest in how their products are
selling.
"We do provide some information back (to suppliers such as Nike)
and it's clearly an area we're looking at," says Horner.
The nuts and bolts of the datawarehouse system at
Blacks
Blacks' IT is based on open systems. Transaction systems such as
purchasing, distribution, stock management and sales, plus the
financial systems, all run on Hewlett Packard HP/UX Unix servers,
and the predominant database architecture is Oracle 7. The 150
gigabyte Penine Galaxy datawarehouse, on which Blacks spends around
10% of its IT budget and which went live in March 1998, runs on a
separate Unix server, and was migrated from Oracle 7 to Version 8
last year. Although held in a single database as sets of tables and
scripts, the warehouse consists of a collection of data marts, such
as the till sales datamart and the branch stock datamart - all
holding different information at different levels of
granularity.
"They can be brought together," says MIS manager John Horner,
"but they need to be stored in different marts because they are in
different schemas."
The warehouse is accessed by users running Business Objects on
their desktop PCs. Business Objects fires users' queries to the
datawarehouse, retrieving the requisite data from the data marts,
processing the data on the server to yield the answer and returning
it to the user's PC.
"There's a year's worth of data (in the warehouse) making 700
million rows of data, so picking out, say, 100 rows, is a sizeable
chunk of processing and it's done on the server," says Horner.
When the warehouse was first deployed, Business Objects was well
suited as the means of user access, and still is for the 25 core
users.
"In the main these are the merchandisers - they are the people
who have to get the right stock to the right place at the right
time, who need to know who's buying what, at what stores," says
Horner.
But increasingly Blacks wanted to be able to make the data in
the datawarehouse available to a larger range of users, most of
whom would not need to do heavy-duty analysis and who therefore
probably did not need the very rich range of functionality that
desktop Business Objects provides.
"It would be a costly overkill," says Horner, to deploy Business
Objects on the desktop of every employee who might need some of the
datawarehouse information, some of the time.
"We needed Business Objects at first, but we don't need to
mass-distribute it," he points out, particularly not when it came
to rolling it out to all the stores. Licence costs apart, "rolling
Business Objects out to 400 stores would have been a maintenance
problem," says Horner. With only eight staff in IT it would
inevitably have meant contracting out maintenance.
"There was no need and no benefit" to mass deploying Business
Objects, he says bluntly.
In order to open the warehouse to the masses, therefore, Blacks
decided to adopt Web delivery. At the time, Business Objects did
not support Web delivery, so Business Objects' resellers, Acuma,
which had implemented the query and reporting tool for the initial
core of users, installed its Information Distribution Server (IDS)
software. This sits on its own Windows NT server, together with a
Business Objects client (although IDS can also take in data from
data feeds not in the warehouse).
IDS comprises a management console, allowing users to specify
every aspect of how they want information delivered, including
content filtering options and output format. The report generation
and distribution is driven by the IDS Engine, comprising a core set
of functions supplemented by a series of plug in processes that
allow IDS to accept input from different sources than the
datawarehouse and output to a number of delivery channels.
IDS can be set up and scheduled to run reports in combinations
of formats, including Business Objects, Word, Excel, Adobe Acrobat,
HTML and rich text, and distributed to different people at
different times. Some of the information put upon the Intranet can
be seen by everyone, some only by designated users depending on
their security levels. At the moment the Web server sits on the
same box as IDS, but, as the Intranet, grows it will probably get a
box of its own, says Horner.
"One of the key benefits is that we can deliver (datawarehouse
information) to our regional and area managers who could never have
seen such reports before, such as our list of top 20 best sellers,
which we analyse on a weekly basis", says Horner.
In the next six months this will be extended to the 400 store
managers using thin client technology, without the need for
installing PCs in the branches. Thin clients will allow the company
to run the software on the head office servers, but allow store
staff to access the Intranet.
"They'll be able to get on board and communicate with head
office - up until now they've only had the Epos till providing a
very limited e-mail," says Horner. "PCs would be a nightmare to
maintain (in 400 stores)."
"IDS has taken us to the next level of data warehousing."
Factfile: Blacks
Blacks is the UK's largest retailer of leisure-related goods, it
has about half a dozen retail fascias, including First Sport, Pure
Woman and AV. It turned over £200m last year and has 400 retail
outlets. There are eight people in IT and three main IT departments
- data warehousing, PCs and networking, and the main transaction
processing systems. Blacks recently acquired the Outdoor Group
whose most famous fascia is Millets. At the moment, the two IT
infrastructures are quite separate.