When dealing with spare parts, the primary challenge is the sheer number of product references which very frequently count thousands, if not tens of thousands of different items. In addition, not only are there many different references, but except for consumables, the number of units serviced per day for a given reference is usually zero, and sometimes one.
Lokad has developed a forecasting technology that is uniquely suited to handle such situations. First, we forecast the entire probability distribution of the demand – because forecasting the average demand provides near-worthless information. In other words, Lokad forecasts the probability of observing zero units of demand, one unit of demand, two units of demand … for every single reference. This information is much richer than traditional forecasts, and in particular it provides the necessary information about “rare” demand levels, instead of focusing on the “average” demand level, which is zero for most references anyway.
Second, we leverage high-dimensional statistics in order to correlate the demand patterns between the different references. Indeed, statistics are supposed to rely on the laws of large numbers, but with spare parts, the sales history of most references is very far from what could be qualified as “large numbers”. In order to overcome this limitation, when forecasting the demand for a given part, Lokad leverages not only the history for that one specific part, but also the history for all the other similar parts at the same time. Since there are typically hundreds of similar parts, Lokad leverages all this data to considerably refine the forecasts.
Spare parts is a “need” market: clients come to buy parts because they need them, which is very unlike the market dynamics found in say , the fashion industry. In an ideal world, your business would be able to serve every single client need. However, in practice this is not possible: there are too many references and too little demand, and serving spare parts is always a trade-off between the quality of service and the cost of inventory. Then, as the quality of service comes with diminishing returns – e.g. increasing the service level from 95% to 97% might require twice as much inventory for only 2% of extra demand served –, there is a point where the cost of inventory balances with the returns gained by the quality of service.
Any time your company is holding at least 1 unit of inventory, your company is making an implicit statement about the future demand: implicitly holding this unit of inventory indicates that the future demand is estimated as sufficient to cover all the expected inventory costs to justify carrying this one unit. Unlike traditional forecasting solutions which emphasize optimizing percentages of forecasting errors, Lokad minimizes the forecasting errors expressed in, say, dollars. We leverage our predictive technology to directly optimize the financial returns associated with your spare parts inventory.
Targeting 100% service levels is unrealistic because mathematically it is the equivalent of infinite inventory, which is hardly practical. Yet, the optimization of inventory is intended to remain closely aligned with your strategic business goals. In particular, delivering a great service through high service levels is frequently a critical ingredient for boosting the loyalty of your clients who are coming for the service and not for the price.
For spare parts distributors, Lokad delivers purchase and divestment priority lists as well as all the relevant KPIs. The purchase priority list indicates which SKUs need to be reordered sorted by order of decreasing returns. Naturally, each SKU appears multiple times in the list, as each extra unit to be purchased comes with a lower priority than the previous ones. In practice, for any given purchasing budget, Lokad can compose the exact list of SKUs with their associated quantities which will deliver the best returns for your company. Lokad can also take into account minimum order quantities or your supply constraints to in order to deliver purchase orders that are compatible with your existing business processes.
Similarly, no matter how good the forecast, servicing spare parts tends to generate some dead or dormant inventory that accumulates over time. Naturally, with better forecasts, dead inventory accumulates at a much slower pace, but it keeps accumulating nonetheless. Hence, Lokad can deliver the prioritized list of inventory that should be liquidated or divested, possibly through a large discount on price and/or an alternative sales channel. Indeed, dead inventory still costs money through carrying costs and clutters your storage areas. The prioritization list typically emphasizes SKUs that have the lowest probability of being sold, but also puts an emphasis on SKUs that are either too bulky or expensive, therefore incurring high carrying costs.
Finally, Lokad also delivers all the KPIs needed to assess the inventory performance achieved through our ongoing recommendations. Through our experience in spare parts services, with companies is verticals as diverse as automotive, aerospace or machinery, we can provide you with the KPIs that have already proved to be well-suited for such businesses. However, we are also very keen to design tailored KPIs based specifically on inputs from our clients, in order to really align the inventory optimization process with your company’s strategic business goals.