It is easiest to set up data collection to do customer analytics if you run an online business.
Require your customers to create their account and track the when they log on, duration, pages that they look at, products that they click and of course what they buy.
From the data, you can carry out Latency analysis by checking if customers are taking a long time between consecutive purchases. For example, by having a column of 1st to 2nd purchase, 2nd to 3rd, 3rd to 4th purchases etc, and another column showing the days of the next purchase, one can find trends that the days of next purchase may significantly lengthen after the 4th or 5th purchase, then you can target promotions at this group of customers.
Another analysis is of course Recency, Frequency and Monetary (RFM) analysis. Develop an index to compare R, F and M measures for a sample size and focus on your top, and customers groups you can convert to top customers. Remember to validate your expected profit with a control group. Costs of a promotion also includes inventory capital and storage costs. Per product price and shipping cost may decrease because you ate ordering more.
Oh, and please look at changes in time trends for action base on RFM, perhaps using statistical quality control?
It may be worthwhile looking at variation for your latency and RFM besides averages too for for targeted marketing. Remember to overlay demographic data with location associated with IP addresses too, for better targeting like changing prices and language for different customers.
Thursday, September 27, 2012
Monday, September 24, 2012
Responsive & Resilient Food Supply Chain And Logistics
Similar to what Cisco has done to recreate a resilient global supply chain, can we create a responsive and resilient food supply chain through Singapore?
Singapore as a transhipment hub thrives on breaking bulk and creating a wide variety of assortments for the region. It is critical that the inflow to Singapore be resilient to minimize cost over huge economies of scale.
How do we create resilient inflows? For food supply chain, one way is to be able to respond quickly to changes in food harvesting conditions. One practical way of going about this is to overlay food harvest locations over the world and and current supplies on the world map. One will then see where supplies are from currently and where are the alternative supplies. Overlay this data with sea/airfreight lead times. This will allow importers to adjust their reorder points quickly in bad weather affecting harvests. Overlay real time weather data for buyers/purchasers to assess impact to inflows and quickly identify alternate sources.
Algricultural produce regions and volume can be found at US agriculture records. Freight lead times can be gotten from maritime data sources. Weather as natural disasters now come in geoRSS feeds. Experienced buyers will be able to provide ratings weather events as a risk factor that one can multiple volume with to assess potential impact. If potential impact is large, the buyer can consider alternative supplies.
With historical data, one can also do a regression on weather patterns and food harvest yields and prices. For more accurate risk factor value.
Singapore as a transhipment hub thrives on breaking bulk and creating a wide variety of assortments for the region. It is critical that the inflow to Singapore be resilient to minimize cost over huge economies of scale.
How do we create resilient inflows? For food supply chain, one way is to be able to respond quickly to changes in food harvesting conditions. One practical way of going about this is to overlay food harvest locations over the world and and current supplies on the world map. One will then see where supplies are from currently and where are the alternative supplies. Overlay this data with sea/airfreight lead times. This will allow importers to adjust their reorder points quickly in bad weather affecting harvests. Overlay real time weather data for buyers/purchasers to assess impact to inflows and quickly identify alternate sources.
Algricultural produce regions and volume can be found at US agriculture records. Freight lead times can be gotten from maritime data sources. Weather as natural disasters now come in geoRSS feeds. Experienced buyers will be able to provide ratings weather events as a risk factor that one can multiple volume with to assess potential impact. If potential impact is large, the buyer can consider alternative supplies.
With historical data, one can also do a regression on weather patterns and food harvest yields and prices. For more accurate risk factor value.
Responsive Online Retail Using Google Insights and Trends
For online retailers wishing to quickly capture search trends and insights to decide which designs to develop and promote can certainly use Google Insights and Google Trends.
Overlay this with Twitter posts for even better trendspotting? anyone? :)
Below is a tabulation of rising searchings in Google Insights for 7, 30 and 90 days. For example, gangnam style is 60% times searched more in the past 7 days than 7 days before. Breakout means that it is searched more than 5000% times more than the previous period. One can easily download, or even poll Google Insights data to track rising trends. For example, if gangnam style suddenly becomes very popular in the past 7 days, you can have your own gangam style product design to capture impulse buys.
Similarly, I tried to use Google Trends to search for supply chain and logistics. Clearly, logistics as a search term is more popular. Someone is South Africa, Singapore, India, UAE, HongKong will likely see my posting if I use "logistics" in my post. There are also trending issues too, to complement Google Insights. In fact, you do not have to key in anything in the field to see what are the top overall searchs on the web.
Overlay this with Twitter posts for even better trendspotting? anyone? :)
Thursday, September 13, 2012
Serializing Products - Aligning Proprietary Serialization with Global Open Standards
Many companies are not familiar with how to use the Global Trade Item Number (GTIN). Even for smaller companies who have their own serialization practices, it is still good to know how GTIN does it so that proprietary serialization approaches can be subsequently matched onto GTINs when business grows.
Always try to get the company prefix with the lowest number of numbers possible. It can range from 4 to 6 digits. As GTIN (whether EAN 13 or UPC 12) has limited digits, try to use just 4 digits for company prefix so that you can have the rest for your product type, colour, size, etc.. identification. You can use '1' or '2' to differentiate between products or '10' and '20', just in case you have new in-between products.
The GTIN can be associated with various product characteristics in your database.
Consider layering the ITF-14 Data Structure if you can't embedded it in GTIN. Add 1 to 8 in front of GTIN to differentiate packaging and aggregation (item, case, carton, pallet) layers.
Approach to adding in packaging levels in front of GTIN using ITF-14 standard.
Examples of application of ITF-14 on different packaging units below.
Spcficiations for bar code type, size, placement and quality all depends on where the bar code will be scanned. it could be at Retail point-of-sale, distribution, marking and repackaging locations.
Below is an approach to selecting the right serialization and bar code standard.
Remember to update GS1 your production sites too, so that they can relate the Global Location Number (GLN) to your GTINs for traceability.
Now that you have your GTINs (or even GLN, ITF-14s) in place, you can then track these products along your supply chain. Tracking and tracing allows you to measure and improve your supply chain performance.
Always try to get the company prefix with the lowest number of numbers possible. It can range from 4 to 6 digits. As GTIN (whether EAN 13 or UPC 12) has limited digits, try to use just 4 digits for company prefix so that you can have the rest for your product type, colour, size, etc.. identification. You can use '1' or '2' to differentiate between products or '10' and '20', just in case you have new in-between products.
Your GTIN can be for the smallest packing possible, and subsequent packing (or SKU) to use another GTIN.
The GTIN can be associated with various product characteristics in your database.
Consider layering the ITF-14 Data Structure if you can't embedded it in GTIN. Add 1 to 8 in front of GTIN to differentiate packaging and aggregation (item, case, carton, pallet) layers.
Approach to adding in packaging levels in front of GTIN using ITF-14 standard.
Examples of application of ITF-14 on different packaging units below.
Spcficiations for bar code type, size, placement and quality all depends on where the bar code will be scanned. it could be at Retail point-of-sale, distribution, marking and repackaging locations.
Below is an approach to selecting the right serialization and bar code standard.
Now that you have your GTINs (or even GLN, ITF-14s) in place, you can then track these products along your supply chain. Tracking and tracing allows you to measure and improve your supply chain performance.
Responsive Supply Chains For Blogshops
From selling fast fashion, many blogshops are offering customized products tailored to individual taste. Does that mean that each product has to be individually manufactured?
Certainly not. From the principles of Design for Logistics, some blogshops apply the principle of economic packaging and transportation for buying in bulk (sometimes by themselves and carrying it as hand luggage).
For some blogshop that customize products though, there are opportunities to batch produce, and standardize products as much as possible in a manufacturing process, and differentiate only at the end. For example, do you sew a bag first or print customized design on it first? your manufacturing process time is certainly shorter if you sew the bag first. The end result? the customer will be delighted that you are sending the customized product faster than they thought. You can also track and encourage customers to tag the company on Facebook or Twitter, and providing a bar or QR code for them to scan to retrieve the specifications easily. This helps promote the product, source new product ideas, and serialize products to track and trace the manufacturing specifications (e.g. zipper configuration or velcro, size? kind of stitching, etc...) for warranty purposes.
Additionally, one can also use concurrent and parallel processing principles to shorten production lead time. For example, can you print on both sizes of a pouch at the same time then sew it together rather than one side at time?
With Twitter and Facebook API we can even try to capture the popular text and designs to push possible designs to customers.
Certainly not. From the principles of Design for Logistics, some blogshops apply the principle of economic packaging and transportation for buying in bulk (sometimes by themselves and carrying it as hand luggage).
For some blogshop that customize products though, there are opportunities to batch produce, and standardize products as much as possible in a manufacturing process, and differentiate only at the end. For example, do you sew a bag first or print customized design on it first? your manufacturing process time is certainly shorter if you sew the bag first. The end result? the customer will be delighted that you are sending the customized product faster than they thought. You can also track and encourage customers to tag the company on Facebook or Twitter, and providing a bar or QR code for them to scan to retrieve the specifications easily. This helps promote the product, source new product ideas, and serialize products to track and trace the manufacturing specifications (e.g. zipper configuration or velcro, size? kind of stitching, etc...) for warranty purposes.
Additionally, one can also use concurrent and parallel processing principles to shorten production lead time. For example, can you print on both sizes of a pouch at the same time then sew it together rather than one side at time?
With Twitter and Facebook API we can even try to capture the popular text and designs to push possible designs to customers.
Monday, September 10, 2012
Machine Utilization For High-Mix, Low-Volume Operations
We understand that Queue Time, Flow Variability and Process Times affect process flow variability. Always remember to add up the utilization of the same machine base on the products using it. Except for utilization = arrival rate x effective processing times for mutiple products for the first machine, you always have to consider queue time, process times and the flow variability of previous departing flow for subsequent machines. Use the correct CTq equation for single or multi-machine scenarios too. Then compare the results with Best Case, Practical Worst Case or Worst Case.
The below pictures gives a good approach to reviewing operations if the results are close to worst case.
The below pictures gives a good approach to reviewing operations if the results are close to worst case.
Wednesday, September 5, 2012
Digitizing Shipping Documents For Speedier Supply Chain Information Flow
Want to relate incoming invoices and other shipping documents with internal production and outgoing shipping information?
The fastest way is definitely EDI or some form of electronic information exchange. However, many companies, especially smaller ones, still prefer to issue hardcopy documents. So what do you do?
There are many OCR softwares available in the market, but none can help users OCR their documents easily and quickly associate with new data, like say on a 128 bar code or QR code.
I have created www.transhipmentocr.appspot.com to help users OCR their jpg, bmp, png and gif pictures. The application can be easily customized to pick up specific information in multiple documents, and associated with other product/shipping information from my barcode 128 decoder at www.transhipmentflow.appspot.com to do it.
The fastest way is definitely EDI or some form of electronic information exchange. However, many companies, especially smaller ones, still prefer to issue hardcopy documents. So what do you do?
There are many OCR softwares available in the market, but none can help users OCR their documents easily and quickly associate with new data, like say on a 128 bar code or QR code.
I have created www.transhipmentocr.appspot.com to help users OCR their jpg, bmp, png and gif pictures. The application can be easily customized to pick up specific information in multiple documents, and associated with other product/shipping information from my barcode 128 decoder at www.transhipmentflow.appspot.com to do it.
Subscribe to:
Comments (Atom)