I’m currently spending a few days in Tel Aviv, researching and working with firms here, and in between, I’m also visiting the local eateries.
Israel has amazing restaurants, being the melting pot it is, but it also has very good street food. An example is Hakosem (translated in Hebrew as “the magician”), one of the most famous places for falafel in Tel Aviv.
Now, I’m not going to take sides and claim that this is the “best,” “good,” “overrated,” or “not even the best” in Tel Aviv, otherwise I risk losing all my Israeli readers. Israelis, myself included, are not willing to be challenged when it comes to what THEY think (I mean…know) is the best Shawarma, Humus, or Falafel. Most people have strong opinions, even more strongly held.
But I’m no food critic, merely an operational one. And to me, the most interesting aspects of Hakosem are the long lines, which they know how to handle (Israelis are not known for their patience … ok, I’ll stop). Their method allows the people standing in line to sample the freshest falafel as once every few minutes, someone brings out a box with the freshest falafel balls and walks the line, handing out samples to those waiting.
A TripAdvisor reviewer writes:
“Incredible schawarma and falafel
Best falafel and sch[a]warma you can get in Tel Aviv. The lines are always long but go by quickly. While you wait, they give you a free falafel sample freshly out of the frier. It's crispy, flavorful and everything you want in a savory falafel.”
These are the freshest balls; as you can imagine, the freshest balls are the crispiest, most mouth-watering ones.
I know what you’re thinking, “They are just over-promising and trying to get people hooked.” And by the time you get the pita, the falafel balls used to assemble the dish (because it’s a job akin to assembly) won’t be the freshest batch, they will always be older falafel balls.
There are multiple arguments to be made:
First, this is a legendary place among Israelis. This is not a tourist trap. Tourists may not know better and can be baited and switched, but a rational Israeli (I’m not falling into the trap I set for myself) will know and will calibrate their expectations.
This also relates to one of my previous newsletters about line balks:
If you want to keep customers joining a long line and not leaving, you need to keep them engaged, but you may also need to keep driving some value. In Wendy’s’ case, it meant having someone walk out and take orders. With Falafel, the bottleneck is the assembly process. There’s also a need to have the person in charge of assembly hear the instructions directly from the customer (there are several options to choose from) in a way that cannot and should not be intermediated by another person. So, if you want to drive some value and offset the high cost of waiting, you have to do it with the BEST falafel balls. In fact, I would argue that this signals that the cost of waiting is quite high (or that people are highly impatient).
As usual, I would like to take it a step further and offer a third explanation. The practice Hakosem uses when handing out samples is LIFO (Last In, First Out): The last falafel in is the first to go out and handed to those waiting, and I would argue that in some cases, it is the best policy.
FIFO vs. LIFO
Most queueing systems (whether for people or products) use FIFO (First In, First Out); immigration lines, most restaurant lines, even the customer line at Hakosem, are all First In, First Out. People are served based on the order they arrive (or at least join the line). For many of us, this is the only policy.
It seems the logical thing to do when it comes to both products and people, and most people can identify queues with FIFO, otherwise, it’s not a queue. As a side comment, there is a whole argument on who invented queues. The British and the French both contend that they did, and that waiting for public transport seems to be a related explanation.
But we’re so used to this practice and its sibling queue (the priority queue where certain priority is assigned based on importance or severity, on top of the arrival time), that it’s difficult to imagine a LIFO queue —a queue where the last item in line is served first. And this applies to both people and products. It’s important to clarify that I’m not referring to the accounting method of how to compute the value of inventory, but rather to the practice of pulling items from inventory when a customer requests a product, or placing them on the shelf, so customers can hand pick them.
This is what “The Queueing and Falafel Magician” does. And now, there’s evidence that he is on to something… A recent paper by Dan Iancu and his co-authors, On the Management of Premade Foods, shows that this policy may, in fact, be the best in many settings.
Their paper, in collaboration with a multi-national supermarket chain, focused on premade food (the term refers to rotisserie chicken, sushi, fresh-baked goods, to-go salads, etc.).
According to the paper, premade food has been the “fastest growing category for a decade and the COVID pandemic motivated reconfiguration of many stores to allocate more space to production and display of premade food.”
Just like with Falafel (albeit there are differences), when it comes to premade food, consumers care about the freshness. Furthermore, the quality of premade food degrades quickly, so its shelf life – the time that elapses from when the food is prepared until it’s disposed of if unsold – must be short, and a lot of it goes to waste. In fact, the collaborator, would throw away 9% of premade food items —a higher rate than in any other food category.
In situations like these, the grocer has to make three decisions:
Shelf life: How long should items stay on the shelf before removed (once they are no longer fresh enough)?
Issuance order: Should items be displayed based on the order they are made (FIFO) or in the reverse order (freshest items first)?
Time stamp: Should the grocer offer information on how fresh the items are or leave it up to the customer to infer it?
Let’s take rotisserie chickens as an example.
For its inflation-proof chickens, Costco uses a shelf life of 2 hours (after which items are pulled from the shelf), but doesn’t time stamp.
Some Whole Foods stores stamp a large label on each chicken when it comes out of the oven, essentially encouraging (or facilitating) LIFO.
But both of these assume that the issuance order is FIFO. But is it?
So, let’s focus on the issuance order, before we go back to the question of shelf-life and time stamping.
First, it’s important to note that if the shelf life is fixed (i.e., it’s pre-determined no matter what), FIFO is the best policy when the goal is to minimize lost sales and disposal (the amount the store discards).
Because it’s important to ensure that the items more likely to expire are released first, so the store can avoid disposing of them. This means that if we adhere to Costco’s 2-hour policy, items don’t need to be time stamped and they will indeed follow FIFO.
But note, this might not be optimal because, as always, there’s a tradeoff!
What’s the Tradeoff?
Long time readers of this newsletter know that I always look for tradeoffs.
The paper on premade food shows the following tradeoffs from using FIFO versus LIFO for a fixed shelf life:
(a) Disposal and lost sales are strictly smaller under FIFO than under LIFO.
(b) Purchased quality is strictly lower under FIFO than under LIFO.
The key is the latter. Under a fixed shelf life, FIFO is better. But maybe a combination of modifying the policy and the shelf life can be even better.
And here is the surprise: LIFO offers higher quality, so stores can stretch shelf life .
Because people use the number of items on a shelf to estimate how fresh the items are.
My former colleague at Kellogg School of Management, Martin Lariviere, had the “lifelong dream” of writing a paper on “why you should never buy the last rotisserie chicken.” He had the title, but the paper never materialized.
But the idea is simple: people form beliefs about the freshness of an item, and in the absence of a time stamp, the issuance order is taken into account.
However, here comes the interesting part: According to the research, LIFO is universally optimal if you can tailor a product’s shelf life to it.
Because the quality of the products you buy when stores use LIFO is higher than when they use FIFO. With FIFO, you will always buy older products. With LIFO, sometimes you will and sometimes you won’t, but since the value degrades with quality, your average quality is higher than with FIFO. The authors make it more rigorous in the paper and summarize:
“The rationale for the universal optimality of LIFO is that by improving purchased quality, LIFO enables the retailer to use a longer shelf life T (i.e., T LIFO > T FIFO) than the retailer could do with FIFO. With that longer shelf life, the retailer is able to increase both customer welfare and sales, relative to what could be accomplished with FIFO.”
Note the change that occurs when the retailer jointly optimizes the shelf life and issuance rule. When dealing with a fixed shelf life, this is not the case.
It is instructive to compare the extreme case where the retailer maximizes only the profit from the focal product, with the case where the retailer maximizes the sum of their profit and the consumer’s welfare, which is generated by the focal product.
In the former case, the retailer just makes sure that the length of the shelf life is such that the customer will always have some utility from buying the product (knowing that sometimes they buy the last item, and sometimes they get a “fresh” one). If the store also cares about the customers’ welfare, the shelf life will be shorter (such that the value is higher than what they would by disposing of), but still better than FIFO. Note that for research purposes, the authors didn’t allow pulling the product if it was the last one as it would be harder to track customer behavior in that case.
Back to Hakosem:
Knowing that you’ll get fresh falafel off the fryer, the average experience is already above other falafel places (which are offering falafel balls using FIFO). Demand for the product increases (as illustrated by the constant line in front of the restaurant), and as the line gets longer, the possibility to get fresh, crispy falafel balls increases along with it.
To be clear: I’m not arguing that we should abandon FIFO and always use LIFO, but we should absolutely ask ourselves whether blindly continuing to follow operational orthodoxy invented by the British and the French is always the right thing to do.
Thanks for reading Gad’s Newsletter! Subscribe for free to receive new posts and support my work.
Excellent topic and well written to point out that LIFO offers a better average experience compared to FIFO because some fresh falafels are better than no fresh falafels. I've seen some stores use this strategy for rotating new milk on shelves.
Curious on how queue depth and wait time could influence outcomes with FIFO/LIFO.
Great piece! Good job not taking a personal stand on the falafel itself. The only part I would question is where you’re stating that “as the line gets longer, the possibility to get fresh, crispy falafel balls increases along with it”. On a whole, sure. But if I’m an individual standing in line and I still haven’t got my free falafel then actually, as the line gets longer, new people are coming for my fresh falafel and I might end up flowing through the entire line without any falafel (for me personally that’s not an issue though as I’m more of a shawarma guy)