Inventory management for an e-commerce store – whether as a stand alone or as part of a multi- or omni-channel operation – is a process that affords retailers the pleasing opportunity to offer customers a vast assortment, as not everything needs to be in your own warehouse. It is quite common that most of the items in an online assortment are only ordered from the supplier when a customer makes a purchase. Some retailers’ online catalogues run to millions of items, and it is quite obvious that most of them are never sold.
So the traditional inventory management limits as to what you can sell, and what you can’t, no longer apply, but unfortunately you still need to decide what to keep in stock. This is not a one-off decision, as in online retail things happen fast. This means you have to monitor your stock constantly to make sure it doesn’t become obsolete. So then, how do you choose a few thousand items to stock from the millions of possibilities, and not just once, or occasionally but constantly? As an engineer, my suggestion would be a fact- and cost-based model – even if it doesn’t cover all the exceptions, it provides a solid basis from which it’s easy to automate. Additional information or opinions can be factored in as needed.
Traditional inventory management limits as to what you can sell, and what you can’t, no longer apply.
As you well know, keeping goods in stock is not free. You have to factor-in capital, obsolescence-risk and warehousing expenses. On the other hand, unit-costs are relatively lower as you can order in larger batches: time spent per product in all phases of the ordering process is quite small, and you can take advantage of freight-free limits and cheaper means of transportation. When ordering only against customer orders, the cost of warehousing tends towards zero, but ordering costs increase significantly when you source in small batches, or even single units, that you need to get relatively quickly. So it gets expensive if you do that very often.
Simply put, the trick is to compare the costs of these two models and select the cheaper one. In general it makes sense to stock cheap items that sell relatively well – and by the same token, expensive items that sell very irregularly are best kept at the supplier’s warehouse.
The principle itself is simple, but the difficulties come firstly in gathering and maintaining all relevant cost data, and secondly, in running the process efficiently. With regard to the first, my advice would be to make a start and improve as you go along. Every incremental improvement you make will produce results. You could start with a subset of products – perhaps the most problematic or the most important to your bottom line. But whatever you do just take the first steps. You’ll soon see the impact and gain in confidence. Rough estimates work pretty well here. When it comes to the execution, it is not a difficult task to programme that approach into a well designed system and to run it automatically as often as needed. We’ve helped many clients facing similar challenges and it really does bring structure to decision making.
Simply put, the trick is to compare the costs of two models and select the cheaper one.
Obviously there are a lot of exceptional situations that must be addressed. Most of them concern the rate of sale, a key factor in the decision making. For instance, how do you know how new items will sell? Well, you don’t. But you can use estimates based on earlier introductions of similar products (this is something that should also happen automatically) and then automatically update your decisions as you go on. After the basic process is in place you can then easily tag exceptional items such as iPhones, for instance, that always need to be in stock.
To take things a step further, you might also take into account the fact that stocking policy itself has an impact on sales. Or more specifically, the lead time, which is a result of the stocking policy, has an impact on sales volumes: If an item comes from a supplier via your warehouse and the scheduled delivery time is 14 days, but a competitor offers it with a significantly shorter delivery time, or the customer simply needs it immediately (think, for examples, of consumables such as nappies) the customer might end up buying it elsewhere. Such products can be identified automatically using data, but in practice it’s often more easy to simply have them tagged by members of the team.
Now that we’ve touched briefly on the topic of assortment planning, I’ll explore what richer analytics allows you to do with the data you gather online in my next post.
Is your supply chain optimised for online?
Our ‘Supply Chain Planning in Online Retail’ ebook covers five steps to improve inventory management in your e-commerce operations. Read it to find out more.