Quick Wins Using Google Forms For Order Processing

Customers can easily use google forms to order, and data to be stored in a format so that its CSV version can be easily ported to another system for invoice generation.

Email alerts can be easily generated when an order is placed. One can also manipulate the item number order to facilitate picking, similar to how you can generate invoices whose item numbers can be shorted for picking straightaway.

Practical Approach To Improve Inventory Inaccuracy

It is common for companies to have inaccurate physical inventory compared to what is reflected in their systems, especially in a high variety, small UOM retail distribution operations in a transhipment flow hub like Singapore.

This leads to:

- time delay on confirm inventory availability as there is a need to check physical stock, rather than generating more sales

- lost sales, lost customers and worst of all, false promises that inventory is available

- higher inventory to ensure that product is available, leading to higher obsolescence

- higher cost of checking with operations and always needing to expedite

A logical approach is to measure cost of these consequence and invest in efforts to dig deep into the reasons for inaccurate inventory and tackle them systematically. This approach to improve inventory accuracy however, is challenging in real-world situations. Firstly, the cost of stock loss is hard to measure as customer demand for each SKU with time/season/promotions. Secondly, most perishable consumer goods distributors allow sales to talk to warehouse operations directly to 'reserve/pick' stocks before pick list is generated. Thirdly, it is accepted that there is a delay in inventory updates. Fourthly, purchasing cost is reduced as certain SKUs are bought in container or pallet loads (reinforcing the acceptability for delay in inventory updates, as there will be enough stock). Fifthly, the multitude of SKUs with different stock loss characteristics makes it challenging for companies to dig deeper into various reasons and collate the cost of different mitigation measures. For example, products that are high value and small by be subjected to pilferage. Products with different packaging sizes may cause inaccuracies from mistakes in picking. Lastly, different inventory accuracy improvement measures escalate cost too.

Rather than collating stock loss and lost sales information, a practical approach to improve stock accuracy is to:

1. Work towards reducing lead time to update inventory. This will encourage operations and sales staff to rely more on system inventory to place orders, and trust that actual inventory is true, or at least closer to the truth. Implement this by trying to get receiving staff to update inventory directly upon receiving. Reserve stock when sales place an order and post out when it is picked. Cutting time to update inventory usually reduces manpower cost to pass inventory forms around too. This will be a quick win that should be enforced as an SOP.

2. With point 1 above done, measure the time reduction in updating inventory and the reduction in calls and paper scribbles operations and sales use to keep track actual inventory, especially for sales in smaller quantities, especially for stable demand items. As this process improvement is instituted for all SKUs, some products may not see improvement due to highly variable customer demands. If the SKU has a long shelf life, bulk purchasing can be implemented to buffer for highly variable customer demands. If it is already done, see of you can send products at high frequencies, tagging along your daily deliveries to pool variability.

3. Try to consolidate orders, rather than allowing daily orders, especially for products with stable demand, pass some savings to customers to encourage bulk ordering rather than piece meal ordering. Reducing transactions this way will reduce mistakes that contribute to inventory inaccuracy. Skip this step if your customers want to order everyday and/or if your business is about customer service through daily deliveries.

4. Then, track system inventory and actual inventory for products with relatively stable demand next. Flat horizontal lines for actual inventory at 0 means there are potential loss sales. Get your sales to note down the quantity in these situations to quantify stock loss/lost sales cost. It may be time to review your safety stock base on variability of demand and delivery lead time.

The above mentioned approach tries to prevent inventory inaccuracies from happening by incorporating focused process improvement and measuring improvements on specific SKUs to see benefits or more accurate inventory. Process improvement is a continuous process, and can also include reducing transactions too. Base on the financial impact of inaccurate inventory, it may be more worthwhile for companies to implement corrective measures like cycle counting (to count faster moving products more often too) and/or have more physical inventory.

If inventory in accuracy is only for a few SKUs, we would want to drill into the source of inaccuracy. Often, there is a combination of reasons for inaccurate inventory. Commonly observed reasons for inventory inaccuracy are pilferage (or shrinkage), transaction error, substitute inventory due to inaccessibility of actual inventory

Pilferage can be known or unknown. Companies can quickly clarify unknown pilferages with spot checks and CCTVs nowadays, and update inventory accordingly.

Transaction error includes:

a. Incoming transaction error - Inventory on electronic packing list is different from what is actual sent, and not corrected in system during receiving

b. Outbound transaction error – Picking/Scanning/Checking wrong SKU

Substitute inventory – The lack of storage space also encourage products to be stored along aisle or spaces meant for movement. This allows pickers to over-ride system directed pick locations. The lack of first-in-first-out flows can easily cause inventory inaccuracy.

The above mentioned reasons for inventory inaccuracy are mainly process related. We also need to layer this with volume related errors for mitigating measures.

Analyzing Application Identifiers (AI) For More Efficient Transshipment Flows

Since bar code is scanned so often, it makes sense to also analyze the many other information that is found on bar codes. I am talking more than just the country code, GTIN and so on.

I am talking about expiry dates, pricing information, batch and other information that can be found on say, a GS1 Databar. It is scanned to track product and location anyway. Analytics on remaining stock on warehouse can highlight expiring items, a cheaper priced product remaining, or what weight of product is not sold. The supply chain partner can then react quickly to the information to reduce mark-downs and expired products.

A list of AIs can be found at http://www.databar-barcode.info/application-identifiers/

Active analysis of bar code information and action will help transshipment hubs like Singapore add value to our partners.

Using Data Mining Measure of Lift To Prioritize Layout Elements

We all know layout of products in a retail store is important. It draws customers into the store, guides them to paths and hope they buy more.

How do we plan a layout to maximize sales? It is usually a mixture of art and science. Some use gut feel base on historical sales of complementary products, some also rely on planograms to develop higher sales from pick-faces. While these are good for smaller layouts and specific shelves, options and opinions may be divided when it comes to positions of different category of products and their relative location to each other. This is where a combination of lift (a measure used in data mining) and relationship chart can come in.

Lift takes into account support and confidence measures in data mining. Support is the probability of a final outcome from all transactions. Confidence is the chance of an outcome given a supporting outcome. Categories with higher lift will take the highest Total Closeness Rating (TCR) used in relationship chart for layout. Presto! You can know locate the category with the highest TCR value first! :)

By locating products close together, trans-shipment flows through Singapore will certainly be faster.

Simple Track & Trace Web Application using App Engine

I have created a simple track and trace web application to anyone to :

a. Associate bar code to a location (http://trackandtraceanalytics.appspot.com/)

b. Check location (http://trackandtraceanalytics.appspot.com/find)

c. Retrieve carton from location to free up location (http://trackandtraceanalytics.appspot.com/scanout)

The text fields “jumps” and “submits” data to webserver automatically, so you do not need to press submit at all. There are also checks on whether there is a location associated with the bar code or not.

I am thinking of adding a page that lets you configure the bar code text fields, and visualize location capacity using bar charts from Google Chart API. Think that it should be more of a track and trace than another warehouse managament system

Let me know what you think?! J