by Brad Welsman, SSI Schaefer

 

The Importance of Online Business

According to a recent NAB survey, Australia’s online retail spending increased by 6.4% to $15bn p.a. in 2014 and represents 6.6 % of traditional retail spending. The Australian Bureau of Statistics estimates that more than half of this online spending goes on imports from overseas.

In order to remain viable in competing against overseas suppliers, it is important that Australian retailers are able to provide high service levels and a competitive and rewarding brand experience for their online customers. Once an effective web site to enable online business has been established, the next challenge is the fast and efficient fulfilment of orders from a distribution centre.

This article seeks to highlight the challenges that online orders present to distribution centre managers and sets out a range of solutions and options to productively deal with them.

 

Why is fulfilling online orders so challenging?

Online businesses are often characterised by many if not all of the following:

• Many small orders
• High number of SKUs
• High percentage of returns
• Special packing requirements
• Fast response time requirements

Why do these issues make online order fulfilment more challenging than for example store order fulfilment?

 

Many small orders

In terms of picking productivity, picking small orders is generally less efficient than picking large orders. This is because the travel distance between picks tends to be greater for small orders than for large orders. For example, unless a special technique such as batching or Goods-To-Person is applied, picking 10,000 2-line orders will take on average significantly more effort than picking 100 orders with 200 lines. In both cases the total task is 20,000 order lines, but the smaller orders will be less efficient to pick.

 

High number of SKUs

In many online businesses (but not all) a key driver for success is the range of products that can be purchased through the online channel. If a high number of SKUs are stocked in the distribution centre, this will exacerbate the inherent inefficiency of manually picking a large number of small orders by further increasing the travel distance between picks. If particular SKUs are bought in from third parties, the incoming products must be sorted to order. Because the number of daily orders is typically very high, this is not a trivial task.

 

High percentage of returns

A key incentive for customers to buy over the internet when they cannot physically see, touch, taste or try a product is the ability to easily return the product if it turns out to be something other than what they expected. While this approach provides a key benefit to the customer and is often a prerequisite for online buyers, it can create major headaches for the distribution centre. The receipt, identification, evaluation, sorting, repacking, restocking and crediting of returns can be both labour and space intensive.

 

Special packing requirements

Because online orders are typically small, they are often despatched in post packs, satchels or small cartons. While picking directly into the despatch unit provides “one-touch” efficiency, this is often not practical with small online orders and therefore a separate packing facility is required. Moreover, many online suppliers offer special services such as gift wrapping or personalised messages to be included with the product. These additional touches, while important for the “customer experience”, come at a cost to efficiency.

 

Fast response time requirement

In this case, response time refers to the time between receiving an order from a customer and despatching it from the distribution centre. A key differentiator for online suppliers is often speed of delivery, with the benchmarks in this respect becoming ever more demanding. While techniques such as batching small orders together can certainly increase picking efficiency, which is discussed below, the creation of large batches can also reduce order response times and may be incompatible with operations where same day despatch times are required or where a multitude of cut-off times need to be applied to suit different courier services.

 

How can these challenges be dealt with effectively?

 

Batch orders together

One of the simplest ways to compensate for the inefficiencies of picking small orders is to batch them together for the purposes of picking. Of course, the items will then need to be sorted back into separate orders again but there are several techniques to do this which are discussed below.

While most modern Warehouse Management Systems have functionality for batching orders together, there are still many factors to be considered in determining the optimal batching strategy. For example, should all items of the same SKU be batched together or should multiple orders containing different SKUs be batched together? What should the size of the batch be?

 

Batch by SKU

Batching by SKU means that orders are batched together such that, during the picking of that batch, each SKU location is visited only once. This provides an enormous boost to picking productivity by maximising the quantity per pick and, assuming the SKUs to be picked are sorted by location, minimising the travel time between picks. The larger the batch, the greater the gain in picking efficiency. Typically a picker can take a trolley or dolly with several empty tote bins and move from one pick location to the next, scanning for accuracy and filling each tote to maximum capacity as they go.

The major benefit of batching by SKU is the very high pick efficiency. The potential disadvantage of batching by SKU is the cost of sorting the items back into individual orders. For example, if 5,000 orders are batched together, several thousand sort destinations will be needed to be able to sort all items to order. The number of sort destinations required can be reduced by reducing the size of the batch but then the gain in picking efficiency will be eroded accordingly.

 

Batch by order

Batching by order means that multiple orders are batched together such that a number of complete orders are batched for picking into the same container, usually a plastic tote bin. Whilst batching a number of online orders together into one tote bin increases picking efficiency, it does not increase it as much as batching by SKU as it is likely that most SKU locations will still need to be visited more than once.

The main benefit of batching by order is that it is easier to sort items back into individual orders later using for example a put wall. While the picking efficiency is not as high as with batch by SKU, the deficit may be offset by the lower cost of sorting to order, depending upon the situation.

 

Single Piece Orders

A typical characteristic of online businesses is a high percentage of single item orders. Regardless of the type of batching employed, it makes sense to treat single piece orders differently to multi-piece orders. This is because single piece orders do not need to be sorted to order; they only need to be identified and packed for despatch. Therefore, very high efficiency can be gained by batch picking single piece orders by SKU. Single piece orders can typically be batch picked into plastic tote bins and then sent to packing stations to be packed and labelled for despatch.

If the items to be picked are very small, they may be able to be placed directly into satchels ready for despatch during picking, using for example a trolley as a mobile workstation, reducing the number of touches and further increasing efficiency.

 

Batch size

An important factor in determining the effectiveness of batching in increasing picking efficiency is the size of the batch. In general, the more orders in the batch, the greater the increase in picking efficiency. In the ideal case, a full day’s orders are batched together so that each pick location is only visited once during the day. This means batching together all of the open orders that are available for release at the beginning of the day and batch picking them ready for despatch at the end of the day. Unfortunately, there are a number of factors working against the implementation of this approach.

Firstly, online orders typically come in irregularly throughout the day, as people order them online, and despatch times may represent an important part of an online business’ value proposition. If orders that are received by say 1pm need to be despatched the same day, they cannot be included in a batch that is created at the beginning of the day. An alternative is to run multiple smaller batches however this erodes the efficiency gain of each batch.

Secondly, the size of a batch is typically also limited by the number of sort destinations available. For example, if there are 2,000 orders per day but only 200 sort destinations available, 10 batches are required. Breaking the work into multiple small batches reduces the cost of the sort-to-order task but again erodes the efficiency gain of each batch.

Another issue with running multiple batches is the potential additional loss of productivity, both in picking and packing, during the batch changeover which often results in reduced activity during the ramp down of one batch and the ramp-up of the next.

It can be seen that determining the optimal batch size is not a trivial matter and a deep understanding of all of the issues in conjunction with some mathematical modelling is a prerequisite for an effective system design.

 

Cluster picking

Cluster picking refers to a situation where an operator picks a batch of discrete orders in a single pass of the pick face but sorts the items to order while picking. Typically the operator will set up a trolley with a number of containers (plastic tote bins or shipping cartons), each representing a separate order. The operator is directed to each pick location using a voice or an RF terminal and directed not only to pick the required quantity but also to allocate the picked items to the correct order on the trolley. In general, the larger the number of the orders on the trolley, the better the picking efficiency, however there is a practical limit to what can be carried on a trolley and once the number of orders exceeds 8 or 10, the incremental benefit of additional orders becomes very marginal.

 

 

Since online orders tend to have a small number of items in them, one option to increase picking efficiency is to batch multiple orders into each tote and then use standard cluster picking to pick multiple totes in one pass of the pick face. For example, if 5 orders can be batched into a tote and a cluster pick trolley can take 6 totes at a time, then an operator can pick 30 orders in one pass of the pick face. Of course, the multiple orders in each tote still need to be sorted out into separate orders, but by applying the cluster pick methodology, the batch size can be effectively increased, improving picking efficiency accordingly.

Cluster picking makes a lot of sense in particular for slower moving items, for example where items are slotted in shelving or binning, as it increases the number of picks per metre walked.

 

Conveyor based picking

For fast moving or promotional items, for example where items are typically slotted in Carton Live Storage or picked directly from pallets, it often makes sense to use conveyors to automatically supply empty order totes to the picking operators and to automatically take picked totes away and convey them to the next picking zone or, if complete, directly to the packing area.

In this way, operators can work in zones to reduce walking and conveyors can route order totes to the zones as required, including to cluster-pick trolley zones for picking of slower movers.

Again, due to the small number of items in online orders, picking productivity can be increased by batching multiple orders into each tote.

 

Goods-To-Person picking

Does Goods-To-Person (GTP) picking make sense for online orders? Although Goods-To-Person picking offers high productivity, often around 1,000 picks per hour, the answer to this question is not simple and the ROI or suitability of GTP picking depends on a number of factors.

Firstly, GTP picking generally provides a better ROI for slower moving SKUs than for faster moving SKUs. This is because the relative productivity gain per pick is higher for slower moving SKUs and at the same time, the investment per SKU is lower because the slower movers generate a lot less throughput.

Secondly, the number of SKUs to be picked in the GTP system is important. It should be borne in mind that every SKU picked in the GTP system must be decanted into plastic totes for storage in an automatic storage/retrieval system (e.g. shuttles, carousels or mini-loads). Decanting faster moving SKUs for automatic storage can be relatively inefficient in comparison for example to picking them from pallet or CLS with conveyor based picking. At the other end of the spectrum, putting very slow moving SKUs in a GTP system is a very expensive form of storage for something that generates little sales. Therefore, particularly where there is a large number of SKUs, the best ROI can be obtained by allocating only a selected range of SKUs to GTP, based on the movement of each SKU.

Another important factor in the ROI of GTP is the overall throughput. GTP systems provide high productivity but if this means the system is only used for one shift or less, the investment is not being well utilised.

 

 

Vertical Lift Systems

Mostly we associate GTP picking with high efficiency pickstations where both order totes and SKU totes are automatically conveyed to and from the pickstation, eliminating walking for the operator. However, while they have a much lower throughput capacity, Vertical Lift Systems also work on the GTP principle. Vertical Lift systems offer an excellent option for space efficient storage of very slow moving or small items, particularly where a secure dust-free environment is required. Vertical Lift systems use the head room to save building footprint and are an efficient alternative for storing and picking items that would normally be stored in binning or VNA shelving.

 

 

Automatic Sortation

As we have seen, batching often makes sense for online orders. However, while batching increases picking productivity, items must be sorted to order afterwards. One of the most efficient ways to do this is using a unit sorter which can automatically sort individual items to different chutes. There is a range of unit sorters on the market including cross-belt, tilt-tray and split-tray sorters, with varying throughput and product handling characteristics.

 

 

Major considerations for automated sorters are the throughput in items per hour and the number of chutes. The required throughput is a function of the number of items that need to be sorted in the peak hour. The number of chutes is a function of the number of orders per batch. Ideally, to maximise picking efficiency, there would only be one batch per day. However, if there are for example 1,000 orders per day, then the sorter would need to have 1,000 chutes which is simply not feasible or practical. An alternative is to break the daily orders into for example 10 batches of 100 orders each. In this case only 100 sort destinations would be required but the smaller batches would reduce picking productivity accordingly.

The problem becomes even more difficult if there are 5,000 or more orders per day. In this case, a potential solution is to sort items into smaller batches of orders and then implement a “secondary sortation” process, using for example a put wall, to sort that smaller batch of orders into individual orders. For example, if there are 10,000 orders per day and the sorter has 200 chutes, then items could be picked in 5 batches of 2,000 orders and each batch could be sorted to a group of 10 orders. The items at the end of the chute then need to be sorted into 10 individual orders, using for example a put wall which is essentially a manual process as described below.

While an automatic sorter enables increased picking productivity, it can be seen that as the number of orders per day increases, the required investment in equipment (i.e. sorter and put walls) will increase and/or the picking productivity will decrease.

 

Put walls

A put wall is essentially a set of shelving locations or pigeon holes into which items are individually placed by operators, enabling them to be manually sorted to order. Typically, the item to be sorted will be scanned by the operator and the operator will be directed into which pigeon hole to place the item, either by a flashing light mounted at the location or via a voice or RF terminal. This put process is only required for multi-piece orders.

Although items have to be sorted manually, put walls are low cost and, unlike automatic sorters, enable thousands of items to be sorted in a relatively small area. This means that larger batches can be picked, keeping productivity high, however it should be noted that the larger the batch, the more walking that is required at the put wall and the lower the “put productivity”.

 

 

As mentioned above, put walls can also be added at the end of automatic sorter chutes to enable secondary sortation and thereby increase the possible batch size without increasing the total number of sorter chutes required.

 

Packing

Because online orders are generally small and numerous, batching of orders and sorting often makes sense. However, this approach precludes the picking of the orders directly into the shipping package, whether that be a carton, tote or satchel. Moreover, online orders often have a wide variety of packaging and labelling requirements and may also require some additional “Value Added Services” (VAS) such as gift wrapping or order specific documentation with for example a personal message for the recipient.

In this case, it makes sense to have a separate packing function with appropriate workbenches at which a range of packaging materials and tools can be made readily available, together with a PC terminal, scanner and quality printers for labels and any additional despatch documentation as required.

Where automatic sorters or put walls are used to sort picked items back into orders, getting items to the packing benches is not always simple, particularly where there are thousands of orders in a batch. Sorted orders can be placed into separate totes and automatically conveyed to packing benches or alternatively put walls can be double side such that putting occurs on the front side of the wall and packing occurs on the rear side of the wall.

 

Reducing the number of touches

While batch picking and sorting of orders increases picking productivity, it also introduces more “touches” into the process of assembling orders for despatch. If there are thousands of orders per day, then the processing of those orders in this way could involve multiple manual steps:

• Touch 1 – Batch picking items by SKU into totes (for transport to a sorter induct)
• Touch 2 – Taking picked items from totes and placing them on the sorter induct (for automatic sortation to groups of orders)
• Touch 3 – Taking sorted groups of orders and putting them to individual orders at a put wall
• Touch 4 – Taking individual orders from the put wall and placing them into totes for transport to placing
• Touch 5 – Taking individual orders from totes and packing them for despatch.

One effective way of reducing the number of touches and at the same time reduce the investment in equipment is to batch items together by order such that a number of whole orders are picked into the same tote. The totes are then transported directly to packing benches whereby each bench is fitted with a small put wall. If for example a maximum of 10 orders can be batched into a tote, only 10 pigeon holes are required in the put wall in order to sort the items back out into individual orders. In this scenario, the packing operator can sort the items to order and repack them without moving from the packing bench.

While batching by order in this way is not as productive as batching by SKU, a sorter is not required and the number of touches will be reduced, which could mean a better overall ROI. Moreover, this solution is more easily able to deal with a significant increase in the number of orders per day and is therefore more flexible with respect to the inherently uncertain growth rates of online business.

 

Non-Conveyables

“Non-conveyables” refers to items that cannot be transported on a conveyor or picked into a carton or tote due to their size, shape or weight. Typically, these items need to be picked separately, usually manually using a voice or RF terminal, and consolidated with other items in the same order.

Non-conveyable items can be consolidated with conveyable items either at a put wall, if pigeon holes are large enough, or at the packing area using for example specially provided shelving. The trick is good timing and clever software which allows the items to accurately come together without one component of the order waiting too long on the other.

For example, non-conveyable items can be picked and temporarily stored in a shelving location next to a packing bench. The Warehouse Control System needs to ensure that the conveyable component of the order is routed to the correct packing bench. When the packing operator scans an item in an order tote which has a non-conveyable component, they are directed to the correct shelving location to retrieve the non-conveyable item accordingly, allowing all items in the order to be packed together for despatch.

 

Returns

The ability of consumers to easily return items is often an integral part of an online supplier’s value proposition. Many online suppliers have a very high percentage of returns which unfortunately create significant work at the distribution centre. Returned items typically have to be checked for validity, whether they can be restocked and resold or whether they need to be refurbished or disposed of. These processes can be labour intensive and time consuming and often represent a significant cost to the business.

It is important that sufficient space and ergonomic workstations are provided to deal with the initial processing of returned items. Placing returned items back into stock is often very inefficient, since it typically involves taking one item at a time to a storage or pick location. An advantage of automated systems for picking is that they can often help with this task. If for example conveyor picking is employed, returned items can be consolidated into plastic tote bins at a central station and routed by the conveyor into the pick zone to which they should be returned. Pickers in the zone, recognising a tote of returned or replenishment items, can be directed via a voice or RF terminal to return each item to its designated pick location in a “reverse picking” process. Where Goods-To-Person picking is employed, returns can be decanted into compartmentalised totes and conveyed into the automatic storage system where they are subsequently given priority over standard stock for picking.

 

Conclusion

A range of automation options is available to help with improving productivity and accuracy in fulfilling online orders. However, due to the unpredictable nature of online order volumes, flexibility remains a key requirement for any automated system that is put in place.

In order to find the right combination between automation and flexibility, a detailed analysis of the profile of the online orders and SKUs is required, including a clear understanding of all key drivers and constraints. In dealing with automation vendors, it is important to find a company that can not only provide this depth of analysis but that also has a wide range of options available in order to offer the best combination of ROI and flexibility.

 

This article is courtesy of: