Stuck in the middle of a canal

One of the biggest supply chain stories of the past week has been the saga of the Ever Given, the ginormous container ship blocking the Suez Canal. Bloomberg has a nice podcast interviewing the head of a global shipping line about the crisis (Baystate Business: Laurence Odfjell, Mar 30).

There are two interesting part to this. First, he gets into some of the physics that likely contributed to how the ship got stuck. Second, he talks about the challenges his firm faced — particularly with ships on the way to the canal and whether those should be sent to a different route.

The Wall Street Journal also has had some interesting reporting — emphasizing the knock on effects of the delays (Suez Canal Traffic Resumes Slowly as Some Ships Weigh Anchor, Others Wait, Mar 30):

Logistics experts were forecasting port congestion in Asia and Europe as some of these diverted vessels arrive at ports around the same time as the delayed vessels now making their way slowly through the canal. That is on top of regularly scheduled traffic.

“This backup risks leading to a concentration of volume,” said Luigi Bruzzone, an analyst for the port of Genoa, one of Italy’s busiest. “What we were expecting to come throughout April will now be concentrated in the last two weeks of the month.”

In short, while the grounding of the Ever Given has been a very visible event, its impact is going to be last for months and likely much less visible to those not in the industry.

Read Full Post »

What is Takt Time and how use it? What is Cycle Time and why do I not use it?

Takt time is a key concept to plan and run an operation. In this video I explain what takt time is and how to calculate and use it. There are many other “times” used in operations and, in contrast to takt time, some of those other times can be confusing and have different meanings at different organizations. Cycle Time is one such example. In theory, the word takt means cycle or beat, but in practice, cycle time would better be called unit workload. Watch why 🙂

Please let me know in the YouTube comments if you have any other topic or question you would like me to address in a future video. I remain curious whether this type of #operationsmanagement content can ever get the “4,000 viewed hrs in 12 months and 1,000 subscribers” to pass YouTube requirements 🙂

Read Full Post »

The times they are a-changin’: Who’s Next? When waiting in a single line is better

The Corona crisis brings changes, including (or especially?) to academics. Marty wrote a blog again. And I got myself re-engaged in producing Youtube videos on operations. I must give credit to INFORMS: our annual international conference went virtual and all speakers were asked to record and upload their 15min video presentation. This made me rediscover the creative challenges of educational video production and I figured I may as well share my presentation on my YouTube channel.

There is joy in creativity and continuous improvement. I quickly realized that 15min videos are often too long and so I have embarked on the endless path of continuous improvement to make better videos. Let’s see how long I stick to that path; feedback and pace of improvement will matter. When you have time, check them out and leave some comments below the video. And if you like them subscribe to the channel and hit “the bell” so you receive an announcement when I upload a new video. (YouTube tracks number of subscribers and number of hours watched. Hence “YouTubers” ask you to subscribe. As we teach: metrics drive behavior 🙂

In this video I discuss why and when waiting in a single line at airport checkin, at the bank, at the supermarket is better. I explain the intuition but also quantify how much better a single queue is over service systems with two separate queues. The largest improvement stems from sharing queue length information (which leads to Join-Shortest-Queue JSQ); the second smaller improvement comes from postponing server choice (which is equivalent to allowing customers to jockey among queues).

The fine legal print: The video addresses “80% of what is important to 80% of viewers” :). It also focuses on customers waiting in line. For the mathematically inclined: The graphs consider simple M/M/1 and M/M/2 queues. There is a deep theory behind “resource pooling in heavy traffic” that shows that the insights in this video extend to Gi/G/N, but notice the required i : we must have independent inter arrival times which is fairly reasonable for customers arriving for service, but not for data network switches…

Read Full Post »

Uber, COVID and the future of public-private partnerships

Over the last few weeks, we saw several interesting announcements related to the gig economy and the future of transportation.

First, from Dallas: “Uber Receives Three-Year Contract to Supplement DART Microtransit Service,” and similar arrangements emerged in Miami-Dade County:

“Their continued partnership also reflects a trend of transit agencies turning to rideshare services to fill transportation gaps. Although Uber has seen a major loss in demand due to the COVID-19 pandemic, places such as Miami-Dade County have turned to Uber and Lyft’s services to subsidize rides along its bus routes that have been suspended at night.”

Of course, this is part of a broader trend: “Some U.S. city transit agencies turn to Uber as ridership drops during coronavirus crisis.”

“An Uber spokesman said the latter initiative was not a major revenue stream compared to Uber’s pre-coronavirus business, according to Reuters, but that it underscores the company’s hopes to further expand into the public transportation sector.”

So, if Uber does not see this as a significant revenue stream, why does the firm do that?

Revenues are meager for Uber these days (91% fewer rides), so it’s clear that any way to keep drivers occupied, while not losing money is a good idea. But I think this goes deeper. As part of its S-1 filing, Uber identifies a “massive market opportunity” in the estimated 4.4 trillion miles traveled by people on public transit in 175 countries in 2017.

Is Uber using these private-public partnerships to get deeper into this “massive market opportunity”?

This is an excellent example of one of the gig economy’s main benefits: the ability to match supply and demand in almost every possible time scale. But it also potentially exposes the main issue. These are market solutions, which may be different than the ones that benefit society. Everything can be subsidized, but it is easier to reverse these decisions than the long-term investment in infrastructure, making them much more dependent on the political climate of the moment. In that sense, it is essential to note that Uber has a dark past when it comes to these private-public partnerships.

For example, Pittsburgh has been less than happy before with its relationship with Uber.

“They currently operate as if they have been given carte blanche access to our city,” Pittsburgh City Controller Michael Lamb wrote, per a transcript of the letter published at WPXI.com. “At Uber’s request, the city of Pittsburgh has opened its streets to a fleet of data-collecting robotic vehicles. This is much more than ride sharing. These vehicles are capable of collecting endless amounts of data about our city. Who owns that data?”

In some areas (Denver is one example), people view it as a way to reduce investment in public transportation. It is clear that once you give firms the ability to do that, the outcomes will be governed by market considerations. The NY Times had a longer op-ed about that last year, “How Uber Hopes to Profit From Public Transit.”

“But by reducing the cost of individual rides, Uber and Lyft also draw a privileged subset of passengers away from public transit systems. That, in turn, undermines support for public transportation… Researchers have also found that ride-hailing tends to make cities more congested and polluted, not less. Alejandro Henao of the National Renewable Energy Laboratory, who drove for Uber and Lyft as part of his research, showed that in Denver, ride-hailing was responsible for an 83 percent increase in the miles that would otherwise have been traveled by car. Much of that increase came from ‘deadheading,’ or driving in search of the next fare. As Mr. Henao puts it, Uber may be reducing the public-transit base without providing enough services ‘to make up for that negative effect.’”

There are many questions, such as, how do we make the market competitive enough so one firm cannot win and take the city hostage? I am also not sure that public transportation is the solution to every area (particularly areas that are not very dense), but how do we make sure that we optimize for the long term and not only the short term? Finally, there is an issue of equity here. How do you ensure that these firms continue to serve low-income neighborhoods or cater to the elderly, non-English speakers, or people with disabilities? Of course, the regulator can do that, but it’s harder and harder to find active regulators.

This brings up a much broader debate on private and public entities’ role in the provision of collective goods (or their substitutes).

Burton Weisbrod, in one of the most seminal papers on this topic , shows that the higher the heterogeneity among customers (both in terms of preferences for quality of service and income), the lower the support for government provision of these goods, and the higher likelihood of an emergence of the private sector substitutes. Given the increased inequality and the significantly more heterogeneous nature of American society, it is not surprising to see these private transportation solutions supplanting the public sector. As researchers predict (pre-Corona) that car ownership will decline by 80% by 2030, it is not all that surprising that the replacement comes from ride-sharing firms replacing public transportation.

Read Full Post »

Will pick up last past the pandemic?

One of my favorite examples I have learned from doing this blog is Chronodrive, a French chain specializing in pick up groceries. Specializing in the sense that this is all that they do. It makes for a nice example since it allows for a contrast between a firm that has tailored all of its operations for one niche against conventional supermarkets that have tried tacking on pick up or delivery onto standard stores.

Of course, in the current environment, lots of firms have had to tack on pick up or delivery options onto their existing stores. To paraphrase Don Rumsfeld, sometimes you have to serve customers with the processes you have, not the processes you might want or wish to have at a later time. But will pick up — some form of click and collect — have legs?

The Wall Street Journal reports that for both restaurants and grocery stores, pick up has been a good business and has been holding up even as states have reopened (Pickup Gains Ground Over Delivery, June 25).

Pickup grocery sales were up 81% in the week ended June 13 from the start of this year, according to Nielsen, while delivery sales rose 33% in that time. At restaurants, carryout accounted for 42% of orders by dollars in May, according to data from research firm NPD Group Inc., compared with a 13% share of sales for delivery. Carryout has maintained its share of restaurant sales since dining rooms began to reopen in May, NPD said, while drive-through and delivery have lost some ground to dine-in orders.

Read Full Post »

The supply chain of Kettlebells and why nothing is going to change post COVID-19

I hear a lot about how many things are going to change posts COVID-19, including supply chains (and global supply chains in particular). I have already expressed skepticism of, what I see, as primarily wishful thinking. So, view the next post as just one more data point, strengthening my confirmation bias.

The NY Times had an interesting article a few days ago and GQ magazine covered it a bit earlier. The story begins with the following interesting observation, telling the story of two trainers.

After the stay-at-home order was put in place, the two began holding virtual workout sessions. But clients would need to supply their own basic equipment. The problem? Basic workout equipment was all but impossible to find. “I did not see this coming,” said Meron Tamrat, 32, a Harlem resident and JTW Fit client. She purchased a dumbbell and kettlebell immediately after the stay-at-home order was announced in March. But a few weeks later, when she needed more weights, she hit a dead end. “Everything was just gone,” she said.

In trying to understand the root cause, one has to start going upstream through the supply chain, as well as down memory lane to the previous recession.

Take, for example, Rogue Fitness, a U.S. manufacturer and retailer of strength and conditioning gear:

Rogue prides itself on manufacturing and selling American-made goods, but the company’s kettlebells are normally manufactured overseas. Most of the kettlebells that you could have ordered before March 13 were; it’s probably not surprising that, in 2020, there are few American foundries eagerly pumping out large bulbs of iron.

When Rogue ran out of Chinese made Kettlebells, they turned to Rhode Island’s Cumberland Foundry.

The irony is that Cumberland Foundry doesn’t really want to be in the kettlebell business. Cumberland isn’t automated, and its president, Tom Lucchetti, estimates that it takes a full day to produce 40 to 50 kettlebells (with Rogue handling last steps, like painting the bells). Rogue typically buys internationally-produced kettlebells by the shipping container. “I’ve been clear with them from the start that isn’t something we can keep up with,” Lucchetti says.” Besides, Lucchetti has no illusions about the current, likely fleeting situation. Foundries in America have, since the ’80s, been decimated by globalization.

If you ask yourself, why I am not too optimistic that this is a sea change and the beginning of a new renaissance of U.S. manufacturing? The same story happened in 2008. And what was the outcome: 

Around 2010, Lucchetti says, the gym owner started selling his one-piece kettlebell to Rogue. Rogue, in turn, quickly figured out it could instead mass-produce its own kettlebell design. Cumberland helped Rogue prototype that design—then Rogue took that design overseas to be manufactured, a fate Cumberland anticipated.

So, let’s review what we just went through: customers view Kettlebells as a commodity and thus are looking for the cheapest option. Manufacturing Kettlebells is more complicated than it seems and not insignificant capital investment, and given the relatively small scale of U.S. foundries, it is much cheaper to manufacture these overseas. This creates a cycle by which the local manufactures do not invest in scale (but are more responsive), and the retailers are unwilling to pay for responsiveness (which is not valued on most days) until it’s too late, and when responsiveness is needed, we can’t scale it.    

What’s the solution? 

Kettlebells — which are valued for their versatility and used in endurance, cardiovascular and weight training — are so hard to come by some New Yorkers have paid mysterious vendors nearly $400 for one set, more than four times what the average kettlebell cost two months ago.

Some people call this price gauging; I call that just the value of responsiveness. If you are unwilling to pay on the day-to-day to maintain a responsive supplier, you should be prepared to overpay when you are in crisis. If you are not thinking proactively, you should be “willing” to pay the penalty of being reactive.  

Read Full Post »

Are supply chains and networks truly resilient

Bloomberg had an interesting article a few days ago “Coronavirus Will Stretch, Not Break, Global Supply Chains” which is very much related to Marty’s yesterday’s post.       

The article starts with the following assertion:

Modern supply chains are often likened to biological systems in their complexity and interdependence. As with the coronavirus itself, a weakness in one part can metastasize into a sickness that causes the entire organism to collapse. As the pandemic has spread, limits on exports of medical supplies grew from a few locations to near-global prevalence. Problems with food trade, such as the restrictions on rice exports that Vietnam imposed in March and the closure of some of America’s vast meatpacking plants after virus outbreaks among workers, threaten a seizure in the food supply chains on which billions of people depend.

This is maybe one of the most interesting aspects of this crisis from the supply chain point of view. Networks are much more resilient than chains. The reason is obvious: in a chain, any point is a single point of failure. You can attack a network in one location, and supposedly, very little should happen.

However, the underlining assumptions are critical:

1. The network is optimized: For networks to be more resilient, you need to have a centralized management and planning authority that ensures the network is indeed optimized for maximum resilience. If each node actually only optimizes for its own profit, I don’t think it’s clear that this is indeed the case.  
2. Full visibility into the network: everyone within the network knows how everyone else is connected, so if one needs to hedge against risks, they understand how their actions impact others, and they know to what extent they depend on other (and usually deeper) tiers of the supply chain.

It is evident that this crisis has demonstrated that both assumptions are violated. We see a lot of short terms “local” decision making in different nodes, and how it impacts others.  When it comes to supermarkets, many of them reduced inventory levels with the assumption that their vendors are going to carry this inventory and will allow them to be “just-in-time”. We see the same behavior when it comes to personal protective equipment, where hospitals carried very little inventory (even though, their mission is to prepare for a crises), again, in the name of cost cutting and being “lean”, not considering the fact that their suppliers are probably weaker than them financially and will not be able to respond in time to any crises. We observe the same in the automotive and electronic supply chain worlds.

We also understand that many firms scramble only now to start mapping their supply chains. Many of them only now realize that their strategic supplier is not necessarily, their at-risk suppliers. So, we see Ford offering loans to lower tier suppliers.  The advantage of a simpler supply chain (with the emphasis on “chain”) is that you know exactly who does what and there is very little diffused responsibility.

And that brings me back the article, which ends on a positive note:

Still, the lesson of the decade or so encompassing the 2008 financial crisis and the trade drama of the Trump administration is that, in spite of the efforts of governments and managers, supply chains in both physical goods and intangible assets have a remarkable ability to heal themselves and continue along their previous paths. The world is struggling to come to grips with one willful biological entity right now. We shouldn’t expect intricate trading networks to be any more tractable in our efforts to bring them under control.

And this might be really the main redeeming quality of networks. While they might not be as resilient as we think to attacks and outside disruptions, the fact that they are locally optimized and built around many decision-makers reacting and responding to the situation around them may have the ability to heal in the long run, sometimes faster than what a central planner can achieve. If this theory is correct, only time will tell. But in many ways, this is where our supply chain theory, epidemiology, and political theory meet.

Read Full Post »

Facebook, Trust and Scalability

The Verge had an interesting article “Facebook will pay $52 million in settlement with moderators who developed PTSD on the job”

The article documents the details of the settlement and the situation leading to it,

“Each moderator will receive a minimum of $1,000 and will be eligible for additional compensation if they are diagnosed with post-traumatic stress disorder or related conditions. The settlement covers 11,250 moderators, and lawyers in the case believe that as many as half of them may be eligible for extra pay related to mental health issues associated with their time working for Facebook, including depression and addiction.”

The moderators were hire, through consulting firms, after the 2016 elections when Facebook was criticized for failing to remove harmful content from the platform. 

Multiple interesting aspects can be learned from this case (before I get to the PTSD aspect). Still, the most important one is the sheer number of moderators needed to moderate content for a tech-driven platform.  

Three key assumptions are driving this situation:

1. Facebook is not a media company, but a platform, and in that sense, it cannot and should not police content. That was the initial point (and still to this day, the legal point) Facebook was making.
2. If it tries to police anything (and it clearly polices some content by virtue of showing us some and not showing us other content), it should be done algorithmically. Why? Machine learning is better than humans.
3. Humans, if can actually do that work, will deem it as non-scalable, which is antithetical to anything Facebook (and I should say, anything Silicon Valley).

It’s the last part I want to focus on, so why did Facebook resort to humans? The fundamental dichotomy that has usually been discussed in the tech sector is Bits vs. Atoms.  Atoms are the physical assets of a business, such as inventory, property/infrastructure, and people. Bits are digital or otherwise intangible assets, including software and intellectual property. In my scaling course I discuss the “Scalability paradox”: in an attempt to become more scalable firms, tend to rely on tech solutions (and in that sense, Bits), even in a situation where technology is premature in addressing a vital aspect of the service, in many cases slowing their ability to scale and enjoy these key benefits. 

Why do I think technology is premature here (I know… machine learning)?

“In the settlement, Facebook also agrees to roll out changes to its content moderation tools designed to reduce the impact of viewing harmful images and videos. The tools, which include muting audio by default and changing videos to black and white, will be rolled out to 80 percent of moderators by the end of this year and 100 percent of moderators by 2021.”

The fact that technology is going to be rolled (slowly) with the ability to mute and move the content to black and white shows

1. How little technology can do
2. How good humans are in tricking these systems, so human judgment is actually needed to do a good job.

So, it’s clear: the solution is a mixture of humans and technology. The question is, how can you reduce the complexity of the task, reduce the need for human intervention, and increase the predictability. Astonishingly, most of the inroads were in providing straightforward tools to minimize the impact, post-trauma, rather than reducing the likelihood of encountering the trauma.

So why is Facebook even engaging in this, after claiming all these years that it’s not a media company and that algorithms are superior to humans: Trust is an existential threat to FB. And this is the scalability paradox for Facebook.

Read Full Post »

Uber and Grubhub: Love in the Time of Corona?

The New York Times, as well as many other outlets, wrote over the last week that “Uber Said to Be in Talks to Acquire Grubhub.” 

The New York Times continued with the subtitle of “A deal would unite two large players in food delivery as more people order in meals during the pandemic.”

This has nothing to do with the pandemic (but also has everything to do with pandemic).

On the one hand, the trend was evident even before COVID-19: people order more food, and restaurants are moving a larger volume of their sales to online platforms. A new phenomenon emerged over the last few years: Dark Kitchens (the term used in the UK) or cloud kitchens (US-based). These are restaurants without a storefront or a seating area. While still early, the emergence of these cloud kitchens is just the beginning of a much broader trend: we will see more efficient value chain from kitchens/restaurants to homes through a decentralized distribution system, powered by these online platforms such as Grubhub, DoorDash, and ClouldKitchens (a VC run by the founder of Uber).  

What we see in the acquisition talks between Uber and Grubhub is also a continuation of the consolidation of this industry. It is clear that the competition among these online platforms is extreme at this stage, and it is evident that the perception among them is that in order to be financially viable, they need to have a significant scale, potentially dominating the entire market.  

However, while this market has some of the hallmarks of a winner-takes-all one (strong network effects, and homogenous customer preferences: we all want availability and fast delivery), it also has many features that counter that (we don’t mind multi-homing, being on multiple platforms, and restaurants prefer being on various platforms). So, while we will most likely not see a winner-take-all situation, scale is going to be a significant driver. We already see consolidation: Amazon invested in Deliveroo, DoorDash acquires Caviar, among other deals.

So, is this even remotely related to the Coronavirus? Uber just fired 20% of its employees. It hasn’t had one profitable quarter, and things are looking pretty grim with the uncertain Corona return to normalcy (even though, I am pretty optimistic about it, as I share here). Since food delivery has been demonstrated to be counter-cyclical to the transportation part of the gig economy but has some positive externalities (same drivers, for example, same customers), this has the potential to be a very positive way to address this volatile time, with significant upside even beyond that.

Read Full Post »

Online Grocery post COVID-19

The Information recently published an interesting article on Instacart, Instacart Weighs New Financing That Would Boost Valuation by at Least 50%.

It starts with an expected, yet somewhat surprising in its magnitude, statistic:

“Instacart’s business has risen more than fivefold since last year, as millions of customers have chosen to order groceries for delivery or pickup rather than go to the stores themselves.”

The Shelter-in-Place, together with the extreme social distancing in some places and/or complete business shutdown in other places, has contributed to the fact that people not only stay home but stay at home and cook. Instacart has more locations, as well as more variety and optionality than Amazon which has contributed to the fact that the firm has increased its lead over Amazon.

But one question is looming, as it does for every business these days: Is this the new normal?

“One question is how many of Instacart’s new customers will become permanent after more businesses reopen across the U.S. A higher valuation might put pressure on the company to retain new customers to sustain its current revenue.”

I have been analyzing the online grocery competition in the US (and globally) for many years and one thing that has been surprising for me throughout the years was the slow pace of growth of this industry. While around 30% of the people have tried ordering online groceries, the market penetration (pre-COVID 19) had been around 2%: 2% of the dollars spent on groceries are done online.

This is a really small percentage, and even more so when compared to other online retail sectors. Why? As I discussed in earlier posts, one reason is that online shopping for groceries is not as convenient as it seems initially: the biggest cost of grocery shopping for most of us is actually going to the store, which means that it is enough that I need to go for one product, I might as well buy everything at the store. Outsourcing and trusting quality control to the picker is something most people are still reluctant to do. The second reason is that when it comes to groceries, people are much more price-sensitive (at least when it comes to paying someone else to do my shopping). Culturally, we are still hunters and gatherers.

How have these factors changed during the corona-days? The cost of going to do shopping has increased significantly: putting on face masks and baring the risk of COVID have changed the math altogether. We don’t want to pay someone to do our shopping and save us time. But we actually like the fact that someone else is bearing the risk for us.

How will it change post-corona? Clearly some people are going to realize the benefits and time savings of online groceries. More people have learned to cook, and how to be more efficient (or settle) for online grocery shopping so I definitely expect to see an acceleration in the penetration rate of online groceries, potentially justifying the new valuation. But will this be the new normal? I don’t think so.

Read Full Post »