FAQs (frequently asked questions) on supply chain Analytics
Following are some of the frequently asked questions (FAQs) on supply chain analytics that we have encountered in our speeches, workshops, seminars, and other forums. Feel free to ask more questions if your particular question is not answered below.
Why we are qualified to write this list of FAQs on supply chain Analytics?
VERY FEW PEOPLE KNOW SUPPLY CHAINS LIKE WE DO - retail, beverages, food, milk, dairy, meat, livestock, explosives, chemicals, cotton, rice, graphite, solar power, natural gas, crude oil, fertilizers, electronics, packaging, glass manufacturing, machine parts, automobiles, industrial goods, mining, etc are just some of the industries where boards and executives have benefited from our proprietary knowledge of the supply chain analytics.
Click on our project methodology above to see how supply chain analytics is an integral step in each and every project that we have undertaken in the last three decades.
Since when no one had heard of supply chain, our co-founder Vivek Sood has been considered one of the most authoritative professionals in the field when it comes to the subject of supply chain analytics in Australia, Asia, North America, South America and Europe.
He has written four seminal books about restructuring supply chains to gain massive advantage in business. He also regularly delivers keynote speeches at business schools and conferences such as University of Technology Sydney, Supply Chain Asia, Asian Bankers Forum, APEC Business Advisory Council.
He has been quoted in the authoritative business press and over 100 academic papers written by supply chain researchers around the world. Vivek and his team have examined thousands of supply chains during their projects over the last three decades and helped hundreds of executives build safe, cost effective and sustainable supply chains and careers.
supply chain analytics - FAQs
Every supply chains generates heaps of data.
With five flows integrating hundreds of nodes in a typical supply chain, and efforts to optimise the entire supply chain underway 24/7, supply chain events and transactions are happening all around you all the time.
All this data is generated, collected, parsed, cleansed, analysed and studied only for one reason - to benefit the humanity. To increase the efficiency, to enhance the effectiveness so that basic needs of humanity can be met in a better manner, supply chain analytics play a vital and essential role.
Data tells its own story - a more objective and 'truer' story in most cases.
Let me illustrate this point with an example from a project. We started a project recently, and prior to starting the project everyone in the senior executive team told us a story that can be summarised as 'we have a good supply chain and we are very proud of it, but now the time has come to take this supply chain from good to great.'
The story was maintained till we spoke to the first few large business customers who showed us their data on the companies delivery rates, which could only be described as appalling. This company was in the bottom quadrant among all the suppliers of their most important customer. In fact this customers' unhappiness was the primary reason for starting the supply chain rejuvenation project and for getting us to talk to this customer at the very beginning of the project.
It was clear that everyone in the company wanted us to hear the truth either from the customer, or the data.
When we independently collected and analysed the data on the company's supply chain performance it became crystal clear that the supply chain of the company far from good.
This is not an isolated incident. I can quote hundreds of such incidents. Almost every company internally agrees on a story. In many cases the supply chain data analytics tells a very different story.
Both stories are important and useful.
One type of story is useful for making the day-to-day life useful and bearable. The other kind of story is useful to know the truth so the company can make progress.
Supply chain data analytics is about 50% of the overall company data analytics task. It is also the hardest 50% for the simple reason that a large chunk of supply chain data is not sitting within the company, but in the supply chain 'out there'.
The other two large chunks of data analytics in the company are differentiated as follows:
Financial data analytics:
This consists of about 25% of the data analytics in the company and is usually the easiest. CFOs signed the bills of ERP systems only after the FS module was well and truly under control, irrespective of what happened to the rest of the modules. If you understood what I just said above, then you will also understand why supply chain analytics is the hardest 50% chunk.
Customer data analytics:
This consists of the remaining 25%, and is a relatively new field opening up after internet connectivity came into place. Now sales and marketing get accurate data on actual customer behaviour and can make decisions based on this rather than on market research data where professed opinions were the norm. Being a relatively new field, businesses are grateful to get whatever insights they can get out of this and are still learning to use it effectively. Moreover, privacy issues and data theft issues make it extremely difficult to push these analytics to their limits.
Data visualisation give life to the data by presenting it in a graphic manner. If you add interactivity to the data visualisation process then the data starts telling its story by itself.
The history of supply chain analytics runs parallel to the history of supply chain management itself. The world's first supply chain project in 1978 was analysed on an Apple II computer by Dr Wolfgang Partsch who is now a senior partner in our company.
The entire story is told in this article titled "How supply chains got more powerful every ten years".
Since then as the computer technology progressed, evert ten years a new generation of supply chain was conceptualised and created by the leading companies and their chosen consultants. Unfortunately, the academia had barely any role in this process.
In parallel supply chain analytics capabilities also progressed because while the mathematics was already available, data crunching capability as well as the systems to crunch these data became available only progressively.
Currently, we are in the fourth generation of the supply chain management, and in the fourth generation of supply chain analytics. The article linked above talks about these generations of supply chains in details.
Supply chain analytics covers the entire gamut of activities in supply chain - from top to bottom and from purchasing to selling.
At the top it covers strategic elements of supply chain. At the bottom it covers the execution of supply chain activities - ranging from buying, to storing, making, moving, selling and delivering.
In the middle of these two levels it covers scheduling of shifts to planning of an entire season. Take a look below to see the entire gamut of supply chain activities. Each of these activities is measured, analysed and evaluated for its effectiveness and efficiency using supply chain analytics.
The four key methods of supply chain analytics are related to the four key levels of supply chain activities as described in the pyramid you see in the answer above.
The bottom layer - supply chain execution - uses descriptive analytics to showcase the data on what was done.
The second layer - supply chain scheduling - uses diagnostics analytics to showcase the data of how well the work was done.
The third layer - supply chain planning - uses predictive analytics to work out what is likely to happen.
The top layer - supply chain strategy - uses prescriptive analytics to to work out how to respond to what happens.
We will talk about each of these layers in detail in the next few answers.
All across the entire supply chain, from purchasing to selling, five flows of supply chain are happening in unison moving goods in lock step manner without too many bottlenecks and dry pools. Hundreds of events or transactions are happening every second. These are recorded in the ERP systems or some other system as they happen.
Descriptive analytics takes the data of these events and transactions and describes the events. The key intent is to answer the question - what happened?
For example - we delivered 569 truckloads of wheat to the wheat mill from the wharf yesterday. Or, in the morning shift, we milled 72 tonnes of wheat into flour on monday.
All you are doing is taking the data from the transaction processing system and making sense of all those millions of transactions to create a picture of the flows through the supply chain.
Supply chain dashboards are extremely useful for descriptive supply chain analytics. Behind the dashboard the tools can range from simple excel or google sheets to sophisticated data crunching or data visualisation tools such as R, python or tableau etc.
Supply chain diagnostics analytics take the descriptive analytics one step further and add a value judgement based on artificial intelligence. We already saw above how descriptive analytics try and describe the supply chain events. The next step is to assess what happened - was it good? OR, was it bad?
To continue the example from above - we delivered 569 truckloads of wheat to the wheat mill from the wharf yesterday. This was 21 trucks over our weekly average for the past 10 weeks. That was good as our productivity was higher than usual.
Or, in the morning shift, we milled 72 tonnes of wheat into flour on monday. This was 8 tonnes below our shift average for the past 3 weeks. We performed 10% below average on milling - let's find out why.
Three things to note about diagnostics supply chain analytics:
- All dashboards are useless without a diagnostic element that assesses the statistics. You will frequently see red, green or yellow colour coding on the dashboards. Sometimes they might have another way to indicate that the level of performance was better, worse, or as expected.
- Frequently the diagnostic element is based on some sort of artificial intelligence (AI). Many of them are rather rudimentary, though some of them are starting to get quite sophisticated. However, this is a good way to start understanding the use of AI in supply chains.
- When combined with human intelligence, diagnostics supply chain analytics are frequently used for root cause analysis. For example, if you find that on every Sunday you deliver 20-30 trucks over weekly average, and on every Monday and Friday you deliver 10-12 trucks below weekly average, you might investigate the data on average run times on these days and find that traffic around the wharf is the root cause of these variations. Similarly if you find that every Monday, the first shift struggles to meet the average shift target you might conduct and investigation and find that there are significant number of last minute sick calls on Monday morning as the supervisor struggles to fill up the shift roster on Monday. What corrective actions you can take is generally part of the next two steps.
The tools used for diagnostic analytics have to go past the basic excel, tableau type of analytical tools to include computer programing languages that can make value judgments based on heuristics or algorithms.
The next step up from diagnostic analytics is predictive analytics where we try and figure out what is likely to happen next.
In the above two examples, you can safely conclude that :
- Next Sunday too, the traffic will be lighter around the wharf and the productivity on the wharf to mill run will be higher as a result. You may want to conduct further analysis on whether it will be cost effective to put some extra trucks on Sunday to take advantage of the lighter traffic. The answer might surprise you because the common assumption is that higher overtime rates of Sunday will override any traffic advantage. What about extra utilisation you get for your equipment? What about the lighter traffic and shorter queues at the wharf and at the mill?
- Next Monday too there will be some stragglers on the first shift who will come in late, while others might not turn up. What is the average no show rate on Mondays? How can you structure the shift rosters to take into account the no show rate? How can you reduce the no-show rates?
In general, predictive supply chain analytics makes predictions about future events based on the historical data about the same, or other events.
More and more sophisticated predictive algorithms are created every day, and many of them incorporate quite sophisticated artificial intelligence - way beyond our simple examples above. As the data sets become larger and larger, and the predictor elements grow in numbers, we enter the realm of use of big data in supply chains. It is unclear where we cross over into the boundaries of what is known as THE BIG DATA.
There are a number of sophisticated statistical packages available for predictive analysis, and almost all the supply chain decision support tools available commercially deploy predictive analytics. We will publish a guide to supply chain decision support tool to cover these separately.
Let us continue the two examples from the answer above to look at prescriptive analytics:
- You may do a cost benefit analysis of putting extra trucks on Sunday and take into account all the extra costs of overtime and extra trucks and offset these against the higher productivity due to lighter traffic and less waiting time as the wharf and gates as well as higher utilisation of trucks which would stand idle on Sundays otherwise, and conclude that per tonne cost of movement is indeed lower on Sundays than on the weekdays. This conclusion will then lead you to prescribe that "let's put extra 20 trucks on Sundays" to take advantage fewer bottlenecks on Sundays.
- You might conduct more analysis and prescribe that extra 17% shift rosters on Monday mornings will cover the no-shows adequately and HR needs to create and roll out a program to reduce no-shows.
Prescriptive supply chain analysis frequently involves scenarios, sensitivities and simulations. The aim is to create options and choose the best option for forward action. Decision science models can get very complicated especially when the stakes are high. More and more artificial intelligence is being incorporated in these models as they go from one generation to next. A number of these models are created and utilised inside the large companies and outsiders will never get an access to these models.
There is indeed a very simple 4 step supply chain analytics process that will apply to all type of analysis. We incorporate this process in our strategy building practice and it is given below:
ERP (Enterprise Resource Planning) systems are massive deployments that cost billions of dollars today. It is difficult to summarise them in a short answer such as this one.
But, clearly they are good at some things otherwise why would companies spend hundreds of millions of dollars on buying these systems and then a few hundreds of millions of dollars on making them more useful.
So what are they good at?
Firstly, true to their name they are good for planning the resources needed for task at hand. If you tell a ERP system that you are thinking of making a car on next Monday, it will spit back that you will need four wheels, four doors etc down to the last nut and bolt you need to make that car. It will then tell you that given the lead times you better order the supplies by this day. It can plan for the workforce, machine times and materials supplies in adequate amount of details - presuming all those assumptions fed into the system are correct. We can go into a lot of details here, but suffice it to say here that theoretically all your resources for making the car will be available on Monday when you start making it.
Secondly, ERP systems are good transaction processing systems to record every supply chain event as it happens. They are repositories of records and factual information about what happened in supply chain and when.
So, why is there such a lot of negativity about the ERP systems?
Firstly, depending on how they are set up, and due to the human propensity to take short cuts many installations do not do either of the above two tasks beyond a level considered satisfactory by the finance department. If the finance department can compile their monthly financial report without too much trouble, they consider the ERP deployment successful. However, operational staff has a much higher requirement from such a system for both of the above two tasks and feel that the system falls short on both measures.
On top of it ERP system sometimes have a tendency to purport analytical capabilities that can best be described as vapourware. Whether it is purported ability to incorporate AI, or predictive analytics or BIG data, or any of the latest buzz words, the real capability may be far short of what it purports to be.
The real deficiencies of ERP system lie in two areas - both of which we have covered elsewhere in these pages:
- What happens outside the four walls of the company is rarely recorded well enough in the transaction processing system. That is why ERP systems are so poor in helping you manage the freight spend.
- Decision support analytics for supply chain decision making related to inventory planning, demand planning, production planning, transportation planning and warehouse planning. You will need specialised decision support tools for these applications that may sit on top of the ERP system - drawing data out of them and churning the numbers to provide you with good enough prescriptive answers that you can act upon.
As indicated in the answer above decision support systems are frequently utilized to supplant ERP systems in more sophisticated companies to hep carry out planning and control of supply chains at much longer time horizon. These systems help with:
- Inventory management - inventory targeting, planning, movement and control
- Demand management - demand forecasting, planning, and shaping
- Production planning and scheduling
- Fulfillment planning and delivery execution
- Transportation management - route planning, optimisation and control
- Warehouse management and optimisation
- Sales and operations planning
This is not the appropriate place to note the key functions of all these various decision support tools or their pros and cons.
Almost all the decision support tools utilise some sort of constraints based optimisation function. They all there two common deficiencies
- The first is their inability to prioritise constraints - whereas we humans know that some constraints are more bendable than others.
- The algorithms and heuristics built into the system may become outdated at a later date without anyone realising that has happened.
We have, in practice seen countless examples of both these deficiencies in the companies. Putting blind trust in a DSS (Decision Support System) is never a good thing.
Whether you buy your supply chain analytics tool off the shelf or build it yourself - the only reliable test is to check it against the ultimate machine - the human brain.
Conduct the same analytics going back to the basics and compare the results. If there are unexplained descrepencies something is going on that must be further investigated.
Notes on FAQs
Clearly, any such list of frequently asked questions (FAQs) about supply chain can never be fully exhaustive. Neither is anyone, including us, the final authority and arbitrator on this or any other topic.
You will have your own opinions on many of these topics, and will have many other questions.
We throw open the comments section to you for your opinions and questions. We will try to address all of these, and the best ones will attract a reward in the form of one of our books, or publications.