There can be very few things more valuable to a small business than the ability to anticipate future demand and optimise their inventory accordingly. The goal is to hit the sweet spot between excessive amounts of stock at high storage costs and dangerously low levels risking an array of stock-out problems. Whilst this sounds simple, getting it right is far from easy.
Inventory forecasting is the data-driven method for future proofing your business. With the right tools in place, you can run your business at much higher levels of efficiency, accelerating your growth and profits.
There are multiple different methods for inventory and demand forecasting. Some rely on expert opinion, or â€˜qualitative assessment,â€™ whilst others rely on data deduction, â€˜quantitativeâ€™. The majority of large retail networks depend heavily on this statistical method of forecasting demand to optimise their stock levels.
A forecast is traditionally considered that which has 50% likelihood to be above or below the future demand as observed through sales. This sounds complex, but it is easily worked out by comparing the absolute difference between an original forecast and the actual demand observed later on.
Brightpearl and Lokad write that for all the forecasting your business may do, it is the stock levels themselves which are the true forecasts made by the business, since each stock level is itself a â€œdirect anticipation of the future.â€
So how can you make sure your physical inventory levels match the upcoming levels of demand? A good way to start is to get measuring.
It stands to reason that you canâ€™t expect to improve your inventory efficiency without closely measuring it first. As we have established, your stock levels are a forecast in of themselves, so start here.
Once you look at what youâ€™ve measured, you can begin to assess the quality of your forecasting. You could do this by calculating the actual amount lost (not the percentage lost) due to the inaccuracy of your anticipations.
In truth, itâ€™s never as simple as this. Not least due to the problem of asymmetry when considering optimum inventory levels. The cost of not having enough stock to fulfil an order often far exceeds the cost of storing the extra stock, so companies which highly prioritise customer service will often over-stock to compensate.
In order to get around this problem, more diligent companies measure the adequacy of the stock levels against the target service levels (where those service levels implicitly represent the financial trade off between the cost of inventory and the cost of stock-outs), rather than against the actual cost.
But what about when it comes to reordering? Firstly, to avoid leaving reorder points to guess work (since they are quantile forecasts) you must keep a record. As we previously mentioned, you can only optimise what you measure.
You must be vigilant when taking into account â€˜lead timesâ€™ – the amount of time between the placing of an order for stock and it becoming available to you – in order to hit the optimal reorder point. Reorder points must cover the entire lead time, because no stock can arrive sooner that this lead time. The total demand over the duration of the lead time is known as the lead demand.
Lokad and Brightpearl write: â€œthe reorder point should only cover the lead demand but only with a certain probability as defined by the service level. If the service level is set at 95%, then reorder point should be as low as possible while maintaining 95% chance of being strictly larger than the lead demand.â€
So by now we have recognised that reorder points can be interpreted as quantile forecasts, but we do not quite have the tools to measure the accuracy of these forecasts. Since certain quantile judgement has been used, classical accuracy formulas cannot be used to measure them. In this case, the pinball loss function could be used.
Whilst beyond the scope of this blog post (you can read about it here), essentially the pinball loss function provides you with an understanding of the relative values of different options with your inventory. It cannot be directly used to quantify company costs, but provides a great way to compare options.
So now youâ€™re set up with the introductory knowledge you need to tackle the optimisation of your inventory. Itâ€™s no mean feat, and should be considered an ongoing process, but getting it right will set you apart from your competition. If you want to know more, check out this exceptional white paper â€œQuantitative Inventory Optimization: How to Anticipate Future Demand.â€
Remember, be flexible and always be measuring.
This guest post was written by Brightpearl. Brightpearl are a partner of Creare who provide multichannel retail management software.