261–270 of 304 posts
Earlier quoted context omitted.
It's mind boggling how most people think it's a simple matching FIFO matching algorithm, instead of how mind bogglingly complex making ride-sharing happen at scale is I would know.
Maybe the world needs a ride share matching game, in the same vein as Elevator Saga[0]? [0] https://play.elevatorsaga.com/
The problem with Uber Eats is exactly the same as the problem with Uber's rideshare: there's no difficult-to-copy element of it. It's been clear for at least seven years that it's easy to spin up a food delivery service, just the same way it's easy to spin up a rideshare service. The basic technology is something a small team can conjure up out of nowhere in a few months, the two sided marketplace is heavily incented…
I disagree. Even Lyft has noticeably worse service than Uber. Compare the prices and time it takes to get a ride on both platforms. They are not comparable.
Uber is an example of a company that wants to be a tech company, but (1) really shouldn't (2) kinda lacks the technical leadership to really be one. It feels like the right call, around 2017, was to either: (A) double down and build a platform for every kind of physical multisided marketplace (matching as a service, maps+routing+locations+geofences as a service, physical product ops as a service, etc) Where Amazon we…
Wow, strong disagree. Having interviewed a couple of candidates from this space, even for just rideshare pricing for both demand and supply sides is an incredibly sophisticated statistics and economics problem, where appropriate modeling can have long term revenue impact. Even within this vertical it seems you’d need a lot more than 15 engineers.
This is the fallacy of completeness and sounds vaguely like Wassily Leontief's input-output planning and has been proven to not work [1]. In the absence of complete information - demand and supply, "appropriate modeling" will always fall short. Economics has again and again showed that sophisticated models always fail and simple heuristics based models work far better. I suspect a simple rules driven system would work just as well.
Uber big innovation IMHO was making supply extremely flexible by massively lowering cost to become a driver and making it possible to be a marginal driver. As a society, we should run with that innovation. The rest of stuff, the fancy app, the car tracking, the dynamic pricing is fluff hiding the real value add.
Uber is an example of a company that wants to be a tech company, but (1) really shouldn't (2) kinda lacks the technical leadership to really be one. It feels like the right call, around 2017, was to either: (A) double down and build a platform for every kind of physical multisided marketplace (matching as a service, maps+routing+locations+geofences as a service, physical product ops as a service, etc) Where Amazon we…
Wow, strong disagree. Having interviewed a couple of candidates from this space, even for just rideshare pricing for both demand and supply sides is an incredibly sophisticated statistics and economics problem, where appropriate modeling can have long term revenue impact. Even within this vertical it seems you’d need a lot more than 15 engineers.
This is the fallacy of completeness and sounds vaguely like Wassily Leontief's input-output planning and has been proven to not work [1]. In the absence of complete information - demand and supply, "appropriate modeling" will always fall short. Economics has again and again showed that sophisticated models always fail and simple heuristics based models work far better. I suspect a simple rules driven system would work just as well.
Uber big innovation IMHO was making supply extremely flexible by massively lowering cost to become a driver and making it possible to be a marginal driver. As a society, we should run with that innovation. The rest of stuff, the fancy app, the car tracking, the dynamic pricing is fluff hiding the real value add.
[1] https://en.wikipedia.org/wiki/Input%E2%80%93output_model
> This writer, previously an avid Uber Eats user, now avoids it in almost all cases and instead, orders directly from restaurants, which have mostly built an online ordering systems. Going direct means your food isn’t ruined I don't know if this is confirmation bias, or just a lie to add weight to the thesis, but it's obviously absurd, as anyone who has ever worked for, or even ordered directly from a local restauran…
This in very small restaurants or when things were slow was handled by the actual drivers/bikers/walkers for their own deliveries, and in large restaurants or during rush was handled by a person who didn't leave the restaurant.
The metrics were repeat business and tips. First orders were given priority, business meeting lunch orders were given priority, drivers would give big tippers priority. The methods were timing when orders were given to the kitchen, keeping on top of orders that were not coming out of the kitchen fast enough, and grouping orders that could go out with the same person. Orders that had problems leaving the restaurant would result in calls to the customer. Things stayed on the line too long without being delivered and got "trashed" (which probably meant somebody took them home after shift.)
Everything needed to be flawless, because mistakes got expensive quickly. The cheapest mistake would be to forget a condiment/sauce, where you might be able to satisfy the customer by offering them a discount, or you might be able to give the sauce to somebody going out immediately. This cheapest mistake probably obliterated all of the profit from an order. If you have to replace an entree due to a mistake? That's better to consider a new delivery order that you're not getting paid for, in addition to not charging for part of the one you already sent out, and if it was a really stupid mistake you had better send some free desert along with it, and if you fuck up again you're going to lose the customer. Worst case, you get everything wrong, you have to remake and redeliver it again, they get it all for free, and you offer them a credit to beg them to come back. Instead of 1 sale, you've had -4 sales.
The process was actually so choreographed and practiced that I'm realizing that it's a lot to remember. I think the assumption that people would do something for 6-12 hours a day and not optimize for least effort and maximum income is weird. You get that by not doing everything randomly, by not getting complaints, and by getting your big tippers to love you. It's not like a tech business where you can just pivot to a different customer - you are locked in a geographic area, and your available customers are your available customers.
The key difference? Now restaurants can just blame the delivery app/driver.
edit: the big takeaway I would put out there is that delivery orders got a lot more attention and processing than inside orders.
Earlier quoted context omitted.
Only a few high volume high margin restaurants are able to do in house delivery economically. Typically pizza joints where delivery is already baked in to the price as that is their primary sales funnel.
>Only a few high volume high margin restaurants are able to do in house delivery economically. Typically pizza joints where delivery is already baked in to the price as that is their primary sales funnel. That's not true where I live (obviously, YMMV). Granted, I live in a very urban area (NYC) and most deliveries are done on bicycle, as the distances are shorter and it's much faster for the delivery folks. Most rest…
Earlier quoted context omitted.
It's mind boggling how most people think it's a simple matching FIFO matching algorithm, instead of how mind bogglingly complex making ride-sharing happen at scale is I would know.
Sometimes I wish it were simpler. I’ve had multiple situations where I couldn’t get a ride into my local city recently. The app says there are rides within 10 minutes, shows me a cheap rate and I hit yes. In 10 minutes no matches have been found. In 10 more minutes it finds someone 20 minutes away. 15 minutes later the driver cancels and I’m back to square one. Now the thing is the cost of the ride is a pittance and…
Earlier quoted context omitted.
> plus he can control his income relative to impacts to benefits In other words, benefits abuse?
If the welfare program says you only get money if you earn under $2000/month, what's wrong with keeping your income just under $2000 month?
Earlier quoted context omitted.
>Only a few high volume high margin restaurants are able to do in house delivery economically. Typically pizza joints where delivery is already baked in to the price as that is their primary sales funnel. That's not true where I live (obviously, YMMV). Granted, I live in a very urban area (NYC) and most deliveries are done on bicycle, as the distances are shorter and it's much faster for the delivery folks. Most rest…
NYC is on a whole other scale than just about any other city in North America. The density of NYC changes the economics of just about anything drastically.
Absolutely. Which is why I qualified my statement with "YMMV."
That said, I think one aspect is broadly applicable: Most low to mid-price restaurants (not chain restaurants) operate on pretty thin margins. And the 12-30% (just for "listing" on their website) can often eat all the profits from the business.
And that doesn't include the "service" fees and delivery charges either. Making the whole enterprise a bad deal for everyone except the UberEats/DoorDash/GrubHub/etc.
Edit: Fixed typo.
Earlier quoted context omitted.
Right, but every additional sales channel is also work , and so the sales need to be large enough to actually justify doing this. If they're not, in a world where some mix of users and restaurants get charged the true costs of delivery service, then you're left with a service that gets people to order things for you, which is basically just Postmates, and they're also not doing hot.
I think some of the actual value of delivery apps has been that they have low start-up costs for restaurants. They aggregate demand for that channel and some of the genius here is that it's pretty simple putting a restaurant's menu online. It seems clear to me that there is plenty of demand to justify the service. Doordash's revenue was about $5B last year (I believe that's revenue and not GTV, which is probably 5x t…
Pizza delivery works because there is only one place you need to go to pick up the food that needs to be delivered, and there is a very high volume of food to be delivered. There's no additional traveling cost to flit between restaurants, most of which don't churn out food at high production rates like a pizzeria.
App delivery drivers spend a lot of their time not traveling directly to customers, or waiting around for food to be ready for them to deliver.
---
> Even if you believe that some of that demand is illusory, brought about by subsidized prices, that still seems like there's a pretty straightforward business there if you can charge a sort of "ordinary business profit."
This remains to be seen. MoviePass was never able to successfully both retain customers and raise prices, streaming services are starting to see the same issues. The premium over pickup is often $10-20 already, if it increases anymore that probably starts having a negative exponential effect on market size. (How many people are willing to pay $30? $50? over pickup?)