Category Archives: Miscellaneous

Is the California High Speed Rail Authority Saving a Dime and Losing a Dollar?

A recent article in the Economist on California High-Speed Rail (HSR), touted Cheaper, Slower as if it was a good thing. Quoting the “Fresno Bee”, The CEO of the HSR Authority has decided to extend the first phase of the project, which was due to complete in 2017, until September 18.

As a result of this extension, which is expected to result in less weekend and overtime work, the California HSR Authority is expecting to save $150 Million of taxpayers’ money. This is being promoted as a good thing. I’m not sure I agree.

You see, for Taxpayers to benefit, the State as a whole has to be financially sound. This means that Revenues Minus Expenditures has to be at least zero, if not positive, and anything that increases revenue or decreases expenditures is generally good, unless the State is running a deficit, in which case the State needs to look at each action and see what the costs of the action are.

In this case, the cost of delaying the project is delaying revenues another year. If you look at the revenue projections for the project, available at this link on the California HSR site, you will see that they are massive. Over 2 Billion annually. Now while it’s true that this is just one piece, from Bakersfield to Fresno / Madera, due to the lack of travel options in the area and the fact that it is a vital part of the corridor between LA and San Jose, the revenue projections for this piece alone appear to be over 25% of the total projections. In other words, to save this 150 Million, the State is delaying at least 500 Million of Revenue by at least a year. Now, HSR does have a high operating cost, we don’t know what the profit margins are, and the HSR might actually be projected to lose money early on, but I’d like to see a full cost benefit analysis of what it is costing in the long run to achieve a projected savings of $150 Million. Until someone does this, we have no idea what the projected savings really are, and even less of an idea as to what the savings will even be as they might not even materialize when you consider expected labour increases, expected material cost increases (given the fact that inflationary times are back), and the fact that a whole slew of things could go wrong to cause delays that need to be made up with overtime.

I’m a little disappointed the Economist took the Fresno Bee at their word. An analysis really is needed here.

A Managed Relationship is a Measured Relationship

I have to agree with the author of this recent piece in Inside Supply Management on “Building Relationships” who said that

if your organization isn’t seeking internal customer feedback – and using it as a learning tool – you may not be as strategic as you think.

As Thomas Nash, of First Line Consultants, says, “Supply management doesn’t own internal stakeholders’ budgets, we don’t own their spend and we don’t run their business unit/function.” The reality is that “Supply management provides stakeholders with fact-based proposals on how to better manage their spend based on reality, our expertise and best practice.”

This means that unless Supply Management is providing stakeholders with the proposals they need, in the manner they need, and the support to execute those proposals, it will not be doing a good job of managing indirect spend in the organization.

So how do you measure internal stakeholder satisfaction? The article gives some good tips to get you started:

  • Involve Staff
    Involve your staff in the process of creating the survey, choosing a process to deliver it, and a set of metrics to measure it.
  • Consider Timing
    The evaluation process should be done annually, but not during budget planning or vacation season. You need as many responses as possible, and they need to be good, thoughtful, responses.
  • Watch Your Language
    Use language the stakeholders can understand, not technical Supply Management terminology. In particular, when surveying legal, use their language; when surveying finance, use their language; and when surveying marketing, don’t be afraid to use a few buzzwords.
  • Be Patient
    The relationship-building and subsequent evaluation/measurement process won’t happen overnight. It will be a multi-year process, but with effort, the organization will get there and the results will improve year-over-year.
  • Share Results
    Share what you learned and the changes you intend to make as a result of the assessment. Do so quickly, and make sure the identified changes get implemented in a timely manner so the stakeholders can see that Supply Management is endeavoring to improve their service levels to the rest of the organization. This is how you become the trusted go-to department in the organization and get indirect spend appropriately managed.

The Real Problem With Most of Today’s Supply Chains?

They’re too fast and too slow.

And no, this is not an oxymoron.

As highlighted in this recent post on Supply Chain Digital, the product life cycle is in decline now that 50% of annual company revenues across a range of industries are derived from new products launched within the past three years at the same time that there are 250 supply chain disruptions to public company supply chains every month (Supply Chain Brain) that result in shareholder value dropping by 10.28% on average (SAS) and that takes the company an average of 50 trading days to recover from.

They’re too fast. There are too many products being introduced too fast. For example, how often do you need a new phone anyway? It’s a damn phone. And a dress shirt is a dress shirt. Now, it’s true that you need to constantly improve computing technology (to keep up with the bloatware), but do you need to change the form factor every year? Sure you need to increase the memory, the processing power, and the storage, but there’s no reason the form factors can’t stay the same — especially since density keeps increasing.

They’re too slow. The average company can’t respond to supply chain disruptions or market shifts fast enough to prevent significant stock-outs, significant drops in revenue, or reputational damages that take it, on average, two months to recover from.

Companies need to balance the competing agendas of innovation, renovation, and reverberation. While constant product innovation is needed, the innovation needs to enhance the product lines and not destroy them. Since research is expensive, the gains from each effort need to be maximized. That means reusing designs, components, and innovations to the extent possible for more than just a year or two.

Furthermore, some things just can’t be reinvented. A toaster is a toaster is a toaster. A new design every year isn’t going to drastically increase revenues and is, to be blunt, a waste of time.

Unnecessary efforts need to be eliminated and redirect to risk management. There’s not enough focus on risk or the mitigation thereof. For example, the benefits of an innovation efforts can be eliminated by the failure of a strategic supplier and the expected profits from a new a product launch can disappear if a supply disruption translates into stock-outs across the board in peak seasons.

In other words, your supply chain needs to slow down and speed up.

Robotistan, I Think Not!

In a recent post over on Horses for Sources, Jim Slaby gives us Greetings from Robotistan, outsourcing’s cheapest new destination, and tells us that software robots are going to replace outsourced labour.

According to Jim, you will soon be able to have your own business process analysts create software robots to do the work instead of outsourced labour because you can get the robots up and running in five months and they will do the work for less than half the cost of Indian FTEs.

His rationale, the existence of a UK startup by the name of Blue Prism that makes a software development toolkit and methodology that lets non-engineers quickly create software robots to automate rules-driven business processes.

Pretty flimsy. For starters, here are the caveats that he finds:

01. The process must be repetitive back-office and not require human judgement or much exception handling.
Which probably limits it to data entry, account review, and creation of initial online access credentials.

02. IT buy in is required.
For starters, the software requires a virtual machine cluster. And the maintenance of such adds to what is probably already an excessive workload.

0.3 There is a learning curve.
It typically takes two to four months to master the tools to model, automate, test, and optimize the robots, according to Jim.

And this is just the beginning. Yes, a large wireless carrier and a major BPO services provider may have found some limited success, but you can’t overlook the facts that:

04. When you scale up, any unhandled exception has the potential to effectively crash the system.
Let’s say you created a robot for account review, a prime example for the technology as indicated by Jim, and you define an exception as any new account under a year that is overdue more than 10 days. Let’s say you are a wireless carrier, which typically has relatively high customer turnover thanks to the fact mobile numbers are portable, and you run the robot on a small test set of 1,000 records and only come up with 10 exceptions. You think it’s great and set it loose on the system with millions of subscribers, but fail to realize your sample set was abnormal and the exception rate is actually 5% and not 1% (and that you failed to insure the less than one year test was properly coded) and all of a sudden you get a queue with 100,000 exceptions that need to be manually processed. Chances are the robot will crash when the manual reviewer tries to load the entire queue!

05. BPM software is currently the be-all, end-all of bloat-ware, especially when you’re trying to create an “AI” application.
As a result, the amount of memory, processing power, and storage required to automate even simple queues is exponentially more than what would be required by an application set up to support a human. And while processing power and storage is still doubling on a regular basis, Moore’s Law is coming to an end as we are close to hitting the point where quantum uncertainty will prevent us from shrinking chips any further. This means that, as you try to build more sophisticated robots, the number of machines you require will double, quadruple, octuple, etc. until the cost to run the hardware will exceed what you could pay a human to do the same task in an emerging market (because machines require energy and energy costs and they are going nowhere but up). And, unlike the machine, the human won’t have to push every tenth transaction to the queue for someone else to process as she’ll know how to deal with the majority of transactions by the virtue of her intelligence, dedication, and desire to keep her job and have a better life.

Software is going to continue to get more powerful, and it is going to continue to automate more data processing, and continue to minimize the amount of data that requires human review, but human review is still going to be required and we’re not going to replace humans in any process that matters any time soon. We might reduce the number of humans we need, but we won’t eliminate the need for them or replace them with robots just yet.

And anyone that disagrees with me can bit my gloomy fleshy ass. 😉