Merchandise planning and control systems can play a key role in increasing profitability, but how do you cut through the hype to find out if they are right for you? John Hobson, Managing Director of The Planning Factory looks at some of the key issues. This article was originally published in 1996 and was used as a source for the Financial Times Retail & Consumer publication “Merchandising & Buying Strategies” in 1999. It was last updated in November 2018. Amazingly, very little has changed!
Merchandise planning systems have enjoyed a very high profile in the retail industry for some time now. If you are contemplating implementing a planning system the first thing that you will have to do is to make a business case for the project. Systems vendors have invested a great deal of time and money in persuading retailers, quite rightly, that effective planning can have a pivotal effect on bottom line profitability. How can we illustrate this?
With a typical cost structure a retailer can add up to 50% to bottom line profit by reducing stock-outs and mark-downs by a couple of percentage points. In the example shown below that means that profits increase by over 3 million pounds for a 100 million pound turnover retailer.
|Lost Due to Stock Outs||12.5%||14,285,714||10.0%||11,428,571|
|Gross Sales inc VAT||100,000,000||102,857,143|
|VAT @ 17.5%||14,893,617||15,319,149|
|Gross Sales ex VAT||85,106,383||87,537,994|
|Variable Costs||Intake Margin %||55%||55%|
|Cost of Goods||38,297,872||39,392,097|
|Actual Sales Value||70,212,766||74,407,295|
|Actual Gross Profit||31,914,894||35,015,198|
|Fixed Costs||Store Costs||23.75%||20,212,766||23.09%||20,212,766|
This reduction in stock-outs and mark-downs can only be achieved consistently by improving the way in which stock is planned and managed.
Are these results really possible or are they just a plausible piece of bait dangled by over enthusiastic salesmen? If we look at Hoogenbosch, the Dutch shoe retailer who run a chain of over 200 stores in Benelux and who implemented IBM’s Makoro software, their own estimate of improvements was as follows:
- In the last two seasons, stock turns have risen from 2.1 to 2.5
- Overall stock reduction of 12 percent; reduced stock of up to 40 percent in some product categories
- Increased gross margin by 1 – 2 percen
- Decreased markdowns by 8 percent
- Up to 25 percent more available open-to-buy
(source: IBM Case Study)
Another area in which savings can be measured is in human resources. According to “Total Jobs” the average salary for a merchandiser in the UK stood at 28,000 pounds per annum in October 2018. Let us suppose that our company has 10 merchandisers and that investing in a planning system meant that they no longer spent most of Monday morning entering and preparing data. The half day saving in labour costs alone would exceed 25,000 pounds each year. Assuming that they used this time in making productive commercial decisions, then the additional return on investment could be many times this amount, and on a continuous basis. Add this to the dividend from improved planning and you have a very compelling case for implementing a planning system
These improvements obviously involve investment in three areas, computer software, computer hardware and human resources, all coming together to optimise the return on investment in stock.
So we can make a case or investing in merchandise planning, but how do you decide if you are ready to make this investment, and if you are, how do you go about choosing a system? To non-specialists, merchandise planning can seem like a black art, but like most business processes it is really just common sense. In this paper I want to try to demystify the process of merchandise planning to help you to discover what questions you should be asking.
How do you decide if you are ready?
The first question that you need to ask yourself is whether or not you have effective core operational systems handling the myriad day to day transactions. If your transactional systems are not well established and understood by the users then you risk wasting your time and money in trying to introduce merchandise planning. Assuming that you feel that you have a mature core system, then you need to take a long hard look at your management culture.
Can you describe your organisation as information driven? Do your key executives understand how to use the information provided by the system, and are they pro-active in trying to get more from it?
These are positive signs that suggest that the time is right to start looking at merchandise planning.
If you do feel that your organisation is mature in its use of information then you need to decide whether or not you wish to make the change process evolutionary or revolutionary. It is easy to be seduced by the “sizzle” offered by currently available products, without taking into account the large investments that will be required to change the way in which your merchandisers operate. Many projects have implementation costs that far exceed the cost of system licences.
You may need to decide whether you need a quick win first. This is most easily achieved by automating existing processes. Once the quick win has shown the value of the approach, then further process changes may be assimilated more readily.
The first step that you need to take in addressing this issue is to ask yourself what you expect the system to achieve.
What should a system achieve?
To put it simply the goal of a merchandise planning system should be to maximise sales and achieved margins by reducing stock-outs and mark-downs.
In order to achieve this there are a set of clearly defined stages that will be followed by most retailers using a seasonally based planning system. Non-seasonally based systems will have similar requirements, but will need greater flexibility in terms of time periods, and will use different methods of extrapolation.
This list is not intended to be prescriptive. It is quite possible that you will wish to add to or to change the list, but it does represent a set of the core elements involved in the process.
You should normally expect a Merchandise Planning System to be capable of providing a large subset of the following:
- pre-season analysis
- normalisation of base data
- plan seeding
- strategic planning (3-5 year time horizon)
- channel planning
- category level plan
- range planning
- store grading
- assortment planning
- line level planning
- store layout design (numeric and visual)
- in-season control and re-forecasting.
Some systems on the market are focussed on the numeric side of planning and some concentrate on the visual, qualitative side. It would be unusual to find a single system that encompassed the entire list shown above.
Each module will consist of a set of inputs, processes and outputs. The overall process is linear, but it is important that it should also be able to be re-iterated and the new results rolled downstream through the system.
Before we look at these in more detail, though, we should first look at the issues that relate to the scope of the system. The question of scope is really one of detail. What levels of your product and branch hierarchies do you need to use in the planning process?
How much detail do you need?
It is important to realise here that, in practice, it is unlikely that you can cover all the bases. There is an inevitable trade-off to make between accuracy and efficiency.
This is an area where the 80/20 rule applies. Each extra level of detail may marginally increase the accuracy, but will require a comparatively huge increase in resources required.
For example, let us assume that we are generating a seasonal value budget and that we are a 100 branch retailer with 100 categories. If we plan at an all branch total level we will have 100 category plans to review. If we take the plan to individual branch level we will have 10,000 category/ branch plans to review. A halfway house here would be to provide a mechanism for “clustering” branches with similar performance profiles and then dealing with these together.
To some extent, however, the scope will be defined by the planning process that you select. If you wish to plan using space by branch, then you will obviously have to maintain and store base data and plan at an individual branch level. If your strategy is served well by your creating a seasonal budget at an all branch level by category, then there is little to be gained by including gratuitous branch detail.
You will also have to accept that there are certain factors that will always be outside your control. These would include the economy, distortions in weather patterns and competitive activity. The fact that these variables can have a strong influence on actual performance is a powerful argument for coming to a realistic compromise relating to the level of detail at which you plan. After all, the one thing that we can be certain of is that your plans will have variances to actual performance!
Regardless of the levels that you select you are going to be planning at a summary level. Even plans at style/colour/size level are summaries of individual transactions. This means that you will have to bring data into the system and store it at the same summary level to avoid having to recalculate the values every time you bring them up onto a screen.
The main message here is that every time you increase the detail in your plan, you vastly increase the amount of effort required to create the plan in the first place and then to keep it up to date. This is a point we should bear in mind when we are looking in detail at the elements of a planning system.
The main elements of a planning system
Analysis is often seen as a separate activity, but in reality it is the foundation on which effective planning is based. Most systems will offer some form of category level analysis in the form of views of actual data as it gets imported into the system. At a simple level, pre-season analysis can consist of reviewing these actuals before the planning process begins.
However, in order to create effective strategies, you really need to get involved in micro analysis. This is typically down at the SKU, Store, Week level and ideally should make use of attributes for both products and stores to allow meaningful summaries of the data. For example we might want to look at summaries within a category by supplier in concession stores (shop in shops).
The output of the pre-season analysis phase should be a clear statement of objectives for the category, based on a full understanding of the historic weaknesses and the future potential of the category.
Normalisation of base data
In order to be able to make a sensible estimate of future performance, it is most likely that we shall use historic data as a reference point. We need to make sure, though, that we are on a level playing field. Historic data will contain all sorts of abnormalities due to such things as promotions, branch refits, bad delivery, bomb scares, and moveable feasts like Easter. If we are to use this data as a basis for extrapolation we must first remove these abnormalities.
This process is called normalisation. In addition to the removal of abnormalities, we might also decide to use more than one year’s data, taking, for example, a smoothed average of the last three years. The result is a set of normalised historic data that we shall refer to here as the base plan.
It is also true to say that the most difficult part of implementing a planning system is getting good, reliable, clean data out of central transactional systems. In some cases the “normalisation” process may need to include estimating data, or using existing plan data, where actuals are not unavailable.
Another area where normalisation is necessary is planning the impact of moveable feasts like Easter. Do not underestimate the time that this apparently trivial exercise can take if it is to be done well.
The process of normalisation is closely linked to plan seeding, Indeed they are sometimes done at one and the same time. Seeding is the process of creating the initial version of the plan based on historical data. Whilst the use of historical data in planning has obvious limitations, the alternative is to start from a clean sheet. This is neither desirable nor practical. The accepted best practice here is to create an initial plan based on historical data, normalise it and then to edit it by exception.
In a properly implemented planning system the process will have been designed so that it is allows tactical actions to be taken to support the company’s strategic plans. A discussion of how to define corporate strategy is outside the scope of this paper. However, in simplistic terms, every business has key performance measures that can be defined in terms like “We are doing all-right as long as sales are increasing by X%, our achieved margin is Y% and our stock cover is no greater than N weeks”. Once these high level targets have been decided, then obviously all downstream plans need to reflect these. If gaps exist then they need to be closed or explained.
Equally, in order to plan infrastructural areas like warehouse space, transport fleets, personnel etc, it is necessary to have broad brush plans for corporate performance. If you are gong to increase sales by 50% then you will probably have to increase warehouse throughput by a similar volume. You may need more staff or a bigger warehouse. In either case you need to be aware of the problem before it is upon you.
This process can be rendered more accurate by providing the user with key performance indicators. A key performance indicator is a measure that shows how well the company is doing in relation to its strategic objectives. Put another way, it is a piece of information that provides relevant decision support. In the context of creating a budget for a company we might show gross margin and weeks cover to help the buyer to make an estimate of the comparative growth potential of its different departments.
The strategic plan can be created in different ways. Firstly we can plan top- down. For example we might know that the financial director is looking for 5% growth next year. We can input this at the top level, and then “cascade” the increase down our hierarchies using the seeding to pro-rata the values. We can then edit the lower levels and re-consolidate the plan. Alternatively, we can plan bottom-up on the basis that our lower level forecasts are most accurate and that the top level must reflect this reality.
The best method is a combination of the two, allowing you to edit the plan at any level and then cascade the values down to lower levels and re-consolidate the plan automatically.
It is important that your system allows you to re-iterate the process easily, as it is unlikely that your first-cut plan will meet with universal approval. You will then need the ability to flex the plan by making amendments at any of its several levels.
The outputs from the strategic plan are typically budgets for sales, margin and stock value and units for a series of seasons, perhaps extending to up to 5 years, broken down to the level of department.
One of the key areas in merchandise planning is trying to assess the impact of store openings, closures and refits, and the impact of any new channels such as E-commerce (the Net New Channel effect). There is a considerable amount of debate as to when we should bring channels into the planning equation. In most cases the principal impact of the Net New Channel effect is on sales. There is not often a great change in key variables like gross margin % as a result of new or closed stores, although it might be argued that the differing fixed costs of bricks against clicks mean that this is worthy of more attention now.
A simple approach therefore is to take account of the Net New Channel effect in sales planning and then to drive the rest of the numbers out using key variables like margin % and weeks cover as planning inputs. It is for this reason that it is of key importance that whatever system you select is able to keep these key variables fixed whilst allowing the sales budget to be flexed and to recalculate outputs like profit value.
As with all planning modules you will be looking for a trade off here between detail and usability. A typical plan will be at store level by week by department. This gives us the ability to model product mix changes as well as openings or closures, but without creating huge plans that are unwieldy to operate and check.
Another variable that you should consider including in your store plan is space. As we shall see later on this is a key variable in the store grading process. Be aware though that maintaining and planning space data requires considerable effort.
One factor that is often overlooked when designing planning systems is that decisions relating to store openings and closures often get taken much nearer to the time of action than is the case with purchasing decisions. If you are going to create your merchandise plan 9 months out, you need to ask yourselves if you will have any meaningful information about new or closed channels at the time of planning. If not, then don’t design store planning to integrate with this stage of the merchandise plan (although do bear in mind that some sort of store plan is essential in order to grade stores for assortment planning).
A further complexity that has been introduced in the last ten years is multi-channel and omni-channel retailing which both make modelling the product flows (both sales and returns) more challenging. This is especially the case if inventory is held separately by channel, but the consumer is allowed to interact seamlessly with the various channels (e.g. buy on line and return to store)
Weekly category level plan
Once we have incorporated the Net New Channel effect into our sales plan (if appropriate) we can start work on our Weekly Sales, Margin, Stock and Intake Plan (often known as a WSSI). The mechanisms involved here are broadly similar to those used in the Strategic Plan. We are simply moving one section of the plan (a season’s worth) down to a lower level of time (week) and product (category).
This first stage of this plan will give us category level forecasts by week for sales, taxes (e.g. EU VAT) markdown and margins.
I have been asked by clients “Why do we plan markdown?” Well, it is almost inevitable that we will have to take mark-down losses. It is important that we plan these so that we can create real forecasts of profitability based on planned achieved margins rather than on intake margins. Mark-down is not a constant, and so it is necessary that we create a phased plan here. However, as we shall see in a moment there is more than one way of defining and planning markdown.
We already have a weekly sale forecast so this module requires us to flex the weekly indicators for VAT %, Intake Margin %, Markdown%, Promotions %, Shrinkage % to give us the resulting Achieved Margin %. This is a margin before the introduction of fixed costs and reflects the sphere of influence of the merchandisers and buyers. For reasons touched upon earlier it is imperative that we plan these relative percentages rather than absolute numbers so that if we flex our budgets the changes can ripple through in away that reflects the reality of the business.
While deciding how to plan your margin you need to think about what methodology you are going to use. Those of you who thought that margin was just sales minus cost may get a surprise at this point.
In the USA and much of the Southern Hemisphere, retailers often adhere strictly to the Retail Method of Accounting. In planning terms this means two things. Firstly your sales margin is calculated automatically based on your available stock margin and secondly your permanent markdown is applied to sales margin, in total, in the week it is actioned. European retailers on the other hand, tend to use a hybrid method where sales margin is planned independently of stock margin, and where the markdown applied to sales margin is either free planned or is an automatically phased version of the permanent stock markdown.
(There is not space in this article to do justice to the relative merits (or otherwise) of each approach, but readers interested in finding out more should feel free to contact me. Suffice it to say that this issue can give rise to Merchandise Planning’s equivalent of a religious war.)
We now come on to what is perhaps the key mechanism in achieving our aim of maximising sales and minimising mark-down – open to buy planning.
Using our phased sales forecast we can generate a stock intake plan that closely matches our intake with the stock requirement. This will probably be done at a category* level , and could involve a variety of methods, the most common of which is to generate a periodic stock requirement based on weeks cover. The difference between the stock requirement and the commitments that we have already made is our “open to buy”. A commonly asked question here is “Should we use flat cover or forward cover? “
*Note that some retailers now take this process down to a category / product seasonality level (e.g. Trousers, Spring/ Summer)
Flat cover says that if this week’s forecast is £1000 and we want 12 weeks cover we need £12000 of stock. Forward cover says that we need to cover the projected sales for the next 12 weeks. With volatile, seasonal merchandise the difference can be highly significant.
Forward cover generally creates more accurate results, but there is a price to pay. If we are calculating the intake requirement for week 20 of a 26 week season and we are using 12 weeks cover, then we need to have a forecast for weeks 21 to 33. This means that we either use week 26 over and over again, enter the stock requirement manually, or we need a simple mechanism to project the next season at the same level of detail.
The depth of cover required in each period will be a reflection of several factors including lead times, safety stocks, service levels required, price/volume trade -offs and closing stock requirements.
It is also important to realise that we are really generating an open to receive rather than an open to buy. If we want to receive 1000 units in month ten and there is a six month lead time then we need to buy it in month four. An open to receive can be made up of goods bought in several different periods. Once the open to buy plan is complete we are in a position to optimise our cash flow, and our merchandise performance by bringing the right quantity of goods into the business at the right time.
Most planning systems also give the ability to convert appropriate KPIs to Cost and Units. To do this we need to use average opening stock and intake prices and margins By inputting these prices and margins we can generate the conversions required without the need for any further effort on the part of the user.
Some planning systems find this a logical point at which to halt the formal planning process. It is certainly the case that from this point on the process becomes more detailed, more labour intensive and the returns begin to diminish. That is not to say, however, that it is not worth persevering.
Planning is about securing a competitive advantage, and the higher quality the plan, the better the buying that will result from it. Having said that, there will always be a trade off between plan quality and plan quantity. What an effective system will do is to allow you to move towards greater plan detail without sacrificing quality or greatly increasing effort.
Having generated a category level plan we need to translate it into something that can be bought. We may know that we plan to take £100,000 in Ladies plain v-neck sweaters and generate a 40% margin on the sales, but we have to decide how many styles and options we should have to achieve this result, and within these how many should we buy to achieve the margin mix.
Decisions here are often “soft” ones. In other words the validity of historical data is questionable and extrapolation can be dangerous. This is especially true of volatile merchandise. We all know that it would be dangerous to say that because beige full length skirts sold well last autumn that they will do so again next time. Equally, the breadth of a range has as much to do with customer perception and the constraints of existing space as it has to do with measurable trends. The lower limit of range width is often described as an “aesthetic minimum”.
What we can do is provide systems that support buyers to help them out of the trap of buying flat. They need to be able to support the winners, and to recognise that some of the styles are probably for window dressing only – they might need to be there, but you do not need to buy a container full.
In this context historical patterns can often be valid. In any typical category the ranked sales participation of styles expressed as a cumulative percentage of total generates a surprisingly consistent curve. For example, it is normal to find that the best seller in a group of ten styles takes about 30% of the sales of the group. The best five styles will take about 85%, and the last five styles account for the remainder. When represented graphically we call this phenomenon a Rank Curve.
If we can predict the likely comparative performance of our range we can use this rule to improve our individual style forecasting.Numeric outputs from the assortment planning process also form a fundamental input to the more graphically based planning systems.
The qualitative aspect of range planning also means that attribute analysis becomes very important. It is essential that the range plan can be analysed in terms of fabrics, colours, supplier, fashionableness etc. The key here is flexibility. A system that only allows you to analyse one attribute at a time will not be adequate when you need to ask the question “How much business am I giving in short sleeved styles in cotton to supplier A?”. Again this may seem obvious, but do not assume that all systems allow you to do what you think they should.
Once the assortment plan is complete we need to ensure that it matches the category sales plan, and the profitability forecasts that we created in the category level plan. In order to do this it is essential that your system is able to create accurate and timely summaries of the key performance indicators (KPIs) for which targets were set in the category level plans. These will include sale (units and value), margin, stock and intake. You should be able to flex the individual option values and create instantaneous summaries by category and by attribute both within and across categories. Finally you should be able to consolidate assortments up you product hierarchy. Be aware that even some “state of the art” systems cannot do all this.
The usual gap analysis will need to be performed and any variances should be explained. We do not need necessarily to close the gap. Plans are created and revisited at different times. Given that we set ourselves some benchmarks with the strategic plan it is always important to monitor against these. However, it is also important to recognise that circumstances change. A flexible plan is a good plan – we just need to be able to explain why we have changed the numbers.
If we accept that it is not practical to create ranges bottom-up for each individual store then we need a way of grouping stores into grades to turn this process into something more workable. There is a tendency to overcomplicate store grading resulting from an understandable desire to cover as many bases as possible in the assortment process. Most of us start off by looking at between 6 and 12 grades, and then some feel that they should go on to add attributes to cover location type, demographics, climate etc. The danger here is that we may end up with almost as many store grades as we have stores, which rather misses the point of the exercise.
Efficient store grading for assortment planning should ideally be based on planned space as this is the key determinant of range width in a retail store. (Web stores are a special case and should be treated differently). If you don’t have space data then you will probably use sales performance to grade with.
There are several different methods you can use to determine the points at which each grade starts and stops. At the most basic you can use equal splits and the more complicated amongst you might like to see the effects of standard deviation in the planned data.
Note: TPF have a ready to use model that allows users to grade by sales or space or a combination using either equal splits, standard deviation or geometric progression to determine breakpoints
Assortment Planning by Line
The role of the assortment plan is to define how we will distribute the styles selected. Planning systems start to show wide divergences here with some restricting the plan to purely numeric outputs and some providing sophisticated 3D animations of how the product will look on fixtures. The key here though is that the limiting factor in retail is space. Any merchandise plan that fails to take into account the constraints imposed by the physical environment risks falling at the last hurdle. Allocating 12 styles to a shop that only has space to display 6 is simple begging to increase the seasonal mark down, at the same time as frustrating your long suffering retail staff.
At a simple level retailers will assign a style to a distribution grade which will dictate which stores receive it. Typically this will result in a wedge shape being created with larger stores receiving greater depth of more styles.
At a more sophisticated level we might look at varying the initial, modular, range by store, substituting items where demographics or climate variances suggest an advantage from so doing.
Sales, Stock, Margin and Price Planning by Line
The final part of most planning systems is to break the category level plans down to a line level. So what is a line level? In most cases we would advocate that this should be style /colour but not style/colour/ size. This is purely on the grounds of manageability. This type of module replicates the stock cards kept by most buyers / merchandisers, but being fed from a centralised system is almost always an area of great efficiency gains. It is also where most in-season monitoring takes place.
Most line level forecast are driven from unit sales forecasts, although I would not suggest that driving them from values is intrinsically wrong.
In season control is a twofold process. Firstly we must monitor the variance between the forecast and what actually happens when merchandise goes on sale. Secondly we must use the information constructively to ensure that we maximise profit. The one thing to bear in mind here is that all the effort that has been put in up to now is worthless unless we monitor variances against the plan.
Variance reporting is essentially a very simple process. It consists of providing reports that compare actual with plan at all relevant levels. In a more sophisticated way we might provide exception based reporting that alerts the users to situations where variances fall outside pre-defined parameters of tolerance. The definition of exception reporting can be a little slippery and I know of one major planning systems vendor who insist that it means traffic lighting data in existing reports, rather than actively seeking out exception situations (they charge extra for that and call it something else)
Once a significant variance has been noticed it should prompt some remedial action.
If a product is outperforming its budget then we might try to expedite existing orders and place more.
If stock covers fall below a certain level we may want to concentrate the stock so that we do not get fragmented ranges.
If a product is under-performing we might try to cancel orders or delay deliveries.
If there are consistent variances from plan then we need to create a re-forecast and flush this through the system again to recast our open to buy plans. This process might be automatic or manual depending on the sophistication of your system. The essential thing to bear in mind here is that our initial forecasts may have been mad e well in advance of the start of selling. The closer we get to the actual selling season itself, the more accurate our forecasts should become.
So there we have the typical components of a merchandise planning system. When assessing the offerings of system vendors you will find that each system approaches the issues with varying strength and weaknesses, by far the most common weakness being in reporting.
Behind all of the systems lurk a variety of technological issues that need to be addressed. Let’s now take a look at some of these.
Getting data into the system
Whilst your transactional systems are no doubt quite capable of generating endless reports, you will not want your response times to grind to a halt every time somebody starts planning. In all likelihood you are therefore going to house your planning system on a separate box.
Depending on the system you select, this box may be anything from a £1,000 PC running Windows NT to £100,000 worth of tin sitting in a cool room, running on Unix with Oracle, and bringing along it’s own requirement for a database administrator who will cost you more, annually, than the total purchase price of an entry level planning system.
Either way you are therefore going to need a discrete summary database to hold the information. The buzz word for this is a datamart.
Getting data into the planning system is easiest where you have an automated process for generating the information summaries and importing them into the planning system.
You need to be sure that you have the flexibility to change or add to the data brought into the planning system without incurring a lot of extra cost.
There are two common methods of getting data into your datamart.
The first is a “push” method where you export data from your main system as a text file and then import it into the datamart.
The second is a “pull” system where you read the information into the datamart using Open Database Connectivity (ODBC). ODBC is a standard for allowing applications to scan external databases and suck data out of them using Structured Query Language (SQL). It has the advantage that it gives you a live link into your transactional databases to refresh the datamart, but it can be difficult to maintain and implement consistently.
The Database format
The datamart itself can take different forms. It may use the same (normally relational) database format as your host system, it may be an OLAP (On Line Analytical Processing) database. If it is an OLAP database it may be multi-dimensional (MOLAP), relational OLAP (ROLAP) or hybrid OLAP (Like Microsoft Analysis Services). Multi-dimensional OLAP datamarts present significant benefits in accessing and manipulating the data very quickly in many different ways (dicing and slicing) and in simple format creation. Relational OLAP (ROLAP) databases provide greater scalability, but can be less flexible, and less easy to implement. The most basic planning systems use standard relational databases to provide low cost but less sophistication.
Operating the System
Another important technological issue relates to the hardware configuration. Here again we see a continuum stretching from the simple standalone PC to powerful mini-computers.
Systems currently on offer include integrated planning systems provided by vendors of central stock management systems, and highly specialised third party stand-alone systems that need to be linked in to your host system.
Third party systems often use client-server configurations that require a dedicated server linked to Windows based PCs. Do not forget to include the cost of providing PCs to those users who do not already have them, but balance this with possible unrelated productivity gains that can result. If a third party system is not fully PC based then you need to ensure that it will run on your existing hardware, or accept the overhead of running another “big box”.
Integrated systems are often less sophisticated, often being designed to offer the minimum functionality that allows a positive response to an Invitation to Tender. At the higher end of the market, however there are some very interesting moves towards tight integration of transactional and planning systems. The two downsides here are that even where it works as advertised, this promised integration is very complex and is generally limited to those systems that cost more than the annual stock purchase budget of your department.
However, it can be beneficial to have a single source for your software when it comes to maintenance and updates, and only one head to bang when you experience problems.
Purchasing decisions should balance the positives and negatives of the above whilst being based on your real needs and capabilities.
When you are evaluating the different systems on offer, you need to take into account the costs and problems involved in implementation. These problems can be split into three areas – cultural, organisational and data-related.
The first problem area is that of changes in corporate culture that will be required if you are going to bring in a planning system.
Change is inevitable when you bring in a new system, and, if it is to be effective, the planning system will need an organisational framework as well as a computer system.
The key question, then, is how much pain you feel that your organisation can soak up whilst remaining effective. The management of change needs resourcing and can be a very expensive process whether it is controlled internally or by consultants.
You can take a revolutionary approach and implement a highly sophisticated system in one hit, but you must make sure that you are in a position to resource, control and monitor the change process.
Alternatively you can adopt an evolutionary approach and introduce the system in stages, making sure that each module is functioning properly and understood before you bring in the next one.
The method of implementation that you adopt needs to balance your need to gain quick benefits with the reality of the change process.
Whilst the change process is normally good for an organisation that manages it well, the chances of a successful implementation increase dramatically when the system is seen to be a servant of the process rather than the master.
You do not want to be told to plan a certain way because this is how the system does it. The systems tail must not be allowed to wag the corporate dog.
The second problem area is organisational, and relates to the personnel who will operate the system.
First of all you need to take into account the level of computer literacy of the staff who will use the system. Once you have done this you need to make an estimate of the cost of training them to the necessary level. These costs may vary according to the sort of system that you buy.
In addition to basic computer skills, your staff will need training on the application itself. This can be a significant element of total costs but can be reduced if you have the right people internally who can be trained themselves and then train others in turn.
Next you need to ensure that you have the necessary human infrastructure to manage the system. Do you have staff at an appropriate level who will sponsor the system and manage it effectively? If not you will need to bear the cost of recruiting the right people. Whilst the expense of this can be high, it can often be cost effective to take on somebody experienced in the planning environment as they will bring a wealth of practical experience with them.
With a basic system your existing staff will no doubt be able to manage the system. With more complex systems considerable demands will be made on the time and intellect of your merchandising staff.
Some companies have taken the step of creating entire planning departments to cope with the additional pressure, thus letting the buyers get on with their jobs after they have provided inputs to the system.
Another organisational issue is that of precedence. Merchandise staff will maintain that a product based plan is most accurate. Sales management will advocate a branch sales based approach. This can give rise to conflict that must be resolved. The precedence given to either is a matter for individual companies to decide as a valid case can be made for both points of view.
Data Related Issues
The final implementation issue I will discuss here is that of providing resources to ensure that the first run of the system does not simply repeat the mistakes of the historical data that you will use as the basis of the plan.
Repeating these mistakes will result in a vicious circle where little improvement is possible. We need to create a virtuous spiral where better merchandising creates a better basis for forward planning.
In order to achieve this there is a considerable amount of work required to interpret the initial base data in the light of inadequate stock provision, overstocking, bad ranging and the like. This process is required each time planning is done, but the load is greatest the first time.
All of the issues discussed above will need to be built into the cost/benefit equation.
Getting the Most Out of the Implementation
Make sure you use the data!
In order to get the most out of the implementation you need to create a strategy for acting on the information, and also to look for spin-off benefits.
To react positively to the new information you need to set up a control framework of in-season review meetings with clearly defined actions that result from them. An end of season review before the planning process begins again will also be of great benefit.
Without these controls you will miss out on a large part of the possible pay-back. The control structure itself is one very valuable spin-off of the system implementation.
As well as reacting to the new information, there are also possibilities for being pro-active.
Look for external benefits
Efficient Consumer Response (ECR) programmes could be the subject of a lengthy article in itself, but in essence it involves a partnership between retailer and supplier. The currency used in the exchange by the retailer is information and that used by the supplier is service.
Your forecasts and in-season re-forecasts have tremendous value if you share them with suppliers, as they are then in a much better position to meet changes in your requirements. ECR often uses Electronic Data Interchange (EDI) to transmit data, but this is relatively expensive and complex and not strictly necessary. You can achieve a lot of the benefits with a free e-mail account!
Look for value added
If we are concerned with maximising the return on investment in planning systems, and indeed systems in general, one area that is frequently overlooked is the value that is added to data in the process of preparing the summary information used in planning.
Many planning systems make use of a datamart of summary information in order to reduce the processing load on central systems caused by repeated query scans on transactional databases.
What is often ignored is that the information in the datamart can also be used in periodic management reporting.
A retailer who sees data on stock, gross sales, mark-down, net sales, profitability and cover as being relevant to the planning process can also gain significant advantage from monitoring the values stored during as well as after the season.
Indeed the technology that allows the views required for the planning process is the same as that used in many cases within a typical EIS (Executive Information System). The multi-dimensional database that allows structured access to planning data can also be leveraged in providing this and different types of information to users throughout the organisation. For example, similar summary information could be generated from financial databases to allow financial budgeting and reporting.
Join up your planning
It is amazing that many retailers buy one expensive system to do merchandise planning and then buy another one that does basically the same job to do financial budgets. If you think laterally and substitute a chart of accounts for a list of products you can save a lot of money as well as integrating your budgeting process and making a reality of “joined up planning”. Wouldn’t it make sense, after all if the sales, margin and stock forecasts in the company budget were the same as the forecasts in the merchandise plan? Wouldn’t it be easier to achieve this if the same system were used for both? Whilst it may be unusual to see this, clients with whom we have implemented such joined up plans see this as one of the most significant benefits from the implementations.
Should you be doing it?
Having read this far, you may still be asking yourself if you should be really be considering implementing a merchandise planning system.
Only you can answer this question but it is beyond doubt that improved stock management is, and will remain one of the critical success factors for retailers in the 21st Century. It makes no difference whether you are a bricks and mortar or internet retailer, you key responsibility is to deliver a return on your company’s investment in stock. You may apply different pressures to the various levers, but the desired goal is the same.
Sales growth forecasts continue to be pessimistic and channels to market become more fragmented and complicated to manage. Increases in profitability will only be available from reductions in fixed costs or better management of variable ones. Most retailers have already shaved fixed costs to the bone leaving better margin management as the only viable remaining lever on profitability.
When you do decide to implement a planning system you will get benefits from four things.
First, by merely getting yourselves prepared to do it you will be forced to take stock of some of the uncomfortable realities of your current position.
Second, implementing the system will require a learning process that will be invaluable, and the change process, if managed properly, will inject new enthusiasm into those involved.
Third, there are the concrete improvements in achieved margin that the system will bring when it is used.
Finally there are the benefits that are available from the new possibilities that the system creates, such as Efficient Consumer Response programmes, creating joined up planning by integrating Merchandise Plans and Financial Budgets and better provision of information.
How should you do it?
To make the most of these possibilities you need to observe some general guidelines.
First of all, do not be seduced by technology. It is easy to become excited about the ability of a system to change data by dragging points on a graph. Features like this are “nice to haves”, but you must look past them to establish that the functionality that you need lies underneath. Equally, don’t assume that because one part of a system has certain functionality, that all other parts behave in the same way.
Second, do not bite off more than you can chew. Your implementation is more likely to be successful if you break it down into digestible sections. You must try to resist the temptation to go for all the benefits at once.
Third make sure that you allow resources to define your requirement before you commit yourself to a supplier. Remember that it is your requirements that must drive the selection process.
Fourth, plan the implementation process properly, and make sure that you have proper internal sponsorship at a high level.
Finally, measure the achievement. You are going to invest a lot of money in the system, and, as far as is possible you, should try to evaluate the return.
Implications for Systems Vendors
So, what does all of the above mean to systems vendors? The fact that users should be looking for the easiest implementation path, and the ability to “grow into” a system means that vendors must provide flexible systems that allow progressive sophistication.
The fact that users need to define their own needs and see these reflected in the system that they buy means that vendors should not be over-prescriptive. Instead they should be enablers providing
- Core functionality
- Consistent Interface
- Access to base data
- Expert advice
whilst allowing the retailer to retain control of the mechanics of the system’s functionality.
In writing this paper I started with the goal of demystifying the process of merchandise planning. Merchandise Planning brings with it a whole raft of new technology and jargon, but you should not allow yourselves to be put off by this.
Merchandise planning is a complex process. George Davies, when he was at Next, maintained that “Retail is detail!” and effective merchandise planning is one way that we can pay attention to this detail.
Let us not forget though that retailing is about common sense. My late friend and colleague, John Beddows, when he was Managing Director of management consultants Kurt Salmon Associates, used to maintain that “manufacturing is a complicated business run by simple people and that retail is a simple business run by complicated people” (I think that this was his revenge for the amount of times he had to listen to the joke about consultants stealing clients’ watches to tell them the time). Without commenting on the accuracy of that statement, the important message here is “Don’t lose sight of the need to keep it simple”.
I have outlined why I feel you should seriously consider such systems, and some of the pitfalls that you should be aware of. I hope, however, that I have also highlighted some of the tremendous opportunities and benefits that they have to offer.
To conclude, the key message that needs to be taken on board is a simple one.
When you start to look deeper into the problem you will realise that the major input to the design of the system needs to come from you, the user.
Developments in functionality may be driven by technological advances. Implementations must still be led by taking a long hard look at your own business requirements.