At the very beginning we were thinking that good vs bad should also include our support costs - not just LTV. But that is amazingly correlated: high LTV means that user has very low support cost (or at least a very nice people to work with).
It is very simple.
A customer "not buying much" is not "bad". If you accept his money for cheap stuff, he is your customer. There's a difference between an honest customer who usually doesn't buy much and then buys something expensive and is getting sadly scammed and a customer trying to rob you.
If you accept the customer for cheap stuff but refuse to help him when he gets scammed on an expensive item, you are the bad, pathological, person/business.
For every refund system you will unfortunately get people who attempt to game that system. They'll buy many negligible cost items without issue, then a single high cost item that they'll claim never arrived hoping they've outsmarted the store or algorithm. These bad actors are why honest consumers have a hard time. I don't know what a business can do to actually differentiate the two.
Costco kinda figured it out, by charging a membership fee they got rid of most low LTV customers from the get-go. Unfortunately, their generous refund policy still got hacked by bad actors which is why they have a lot of restrictions around returning electronics. (People were returning laptops after a year of use just to upgrade). This is why we can’t have nice things.
source: I worked at Costco before they had restrictions on their electronics return policy, when they bragged about accepting returns on anything for any reason… before the cost of doing so became prohibitively high.
LTV refunding like Amazon is doing is a huge leap forward that lets every business kind of act like Costco.
You seem to be saying 'well they did say any reason' and that good faith on the part of the buyer didn't/shouldn't enter into it. Yes they did, and people behaved the way they did. Now the generous refund policy no longer exists. See how that works?
It’s a shame the policy is gone. There was really no peace of mind quite like it, knowing that if your device failed after years of use, but still due to a defect, you had recourse. And for poor people, it was a godsend of a program especially (and not because it could be abused, but simply Costco providing a service that anything you buy will work as advertised)
[1]source: some internal training I received back in 1998
People suck, especially when they are disconnected from the human side of business.
It makes me sad. Ebay two decades ago was a lot different. I never got burned back then, and did a fair bit of buying and selling snowboard gear, including boards that get up to around $800 each. There's absolutely no way I'd do that today. Just impossible not to get ripped off on it.
Could the retail organization use a physically distributed escrow service to somehow tie payment to physical goods delivery? That essentially outsources the "brick and mortar" aspect to a third party and removes delivery risk. Amazon and Currency Exchange, for example, can have a much healthier economic relationship than Amazon and me.
Alternatively, this sounds like it could be addressed as a sort of "consumer protection" insurance similar to my homeowners or auto insurance. I file a claim so insurance covers legitimate large losses. Risk is pooled and the underwriter is the organization that is evaluating my trustworthiness, not the individual retailers.
Just ideas off the cuff, maybe there are huge holes here.
On eBay fraudsters buy large quantities of eg $1 seed packs, to build up artificial account ratings. Then they try to steal expensive things (didn't arrive, wasn't in the box, arrived broken, etc), and eBay shields the accounts heavily because they favor buyers to an extreme. All the thief has to do is be disciplined about it, buy $40-$50 of cheap items for positive feedback, then go after a $300-$500 item. Sellers can't even leave negative feedback on eBay, which further cuts off a mechanism for at least warning other sellers of a running con (sellers sometimes leave the negative language in the neutral/positive feeback segment instead, which rather points out the absurdity of eBay's broken system).
Is anyone actually arguing for fewer "easy refunds" and more investigation? I though the point was that given the fact that most of their refunds are handled algorithmically, they have plenty of time to handle the rest of the refunds quickly and with care.
If I don't get my order, I want my money back. If the algorith decides I can get that with no investigation required, sure, I'll take it. But if the algorithm thinks they shouldn't take the "shortcut", I still deserve my money back and a human should be dispatched to deal with it ASAP.
Algorithmically adjudicating in favor of the customer is fine - it's their money they're risking. But algorithmically adjudicating against the customer should never be allowed without a human looking things over.
In theory, Amazon's T&C's attempt to pass on the risk of loss to me after delivery of the item to a carrier. In practice, this is far less clear, because of merchant agreements / cardholder agreements and overall FTC regulation of goods sold directly to consumers.
But it may be moot, anyways: if Amazon decides to screw me, they have a bunch of digital assets that they hold hostage and will remove my access to if I initiate a legal dispute or chargeback, even if I'm in the right.
And I'm not just using the term "responsibility" im a moral sense here - consumer protection laws in many places agree with this. The seller is an "informed market actor, taking on the risks of the transaction". The consumer is not an "equal market actor" and is therefore afforded protections from the risk.
The law will oblige either of the two sides to carry the burden of proof. Either Amazon has to proof the package was delivered, or the customer has to proof it wasn't. If the other sides does not accept the proof received, it has to sue.
Update: I should be clear, the shipper can prove it. The customer themselves can't without the shipper's acquiescence.
The shipper might be able to prove that the transport company picked up a package. Beyond that it's all hearsay if we are talking about fraudulent sellers.
The transport company might have evidence of a tracking number and thus most probably an actual package making it through the system. Shipments do get lost or stolen though. No way to prove anything really.
The customer can't prove it either. There are enough cases of packages with a tracking number having arrived at my and other people's doorsteps without the tracking ever showing that it happened. Why wouldn't the opposite be true? Driver just needs to scan package arrival not actually deliver. Of course there can be repercussions and I'd gather it's easier for a package to be stolen off other parts of the delivery chain. Anyway in this case both the shipper and customer have 'proof' / no proof to various degrees and in reality something 'fell off the truck'.
It's all about being convincing enough to whoever will be adjudicating this but 100% proof is impossible. The closest the customer could get I think would be 24h Video surveillance of their front entrance. No motion detection, just 24h continuous tape.
Yep, I think this is exactly how things work. My household buys pretty much anything it can from Amazon unless it's more expensive than we can get elsewhere because our Amazon credit cards automatically give a 5% discount (cash back). Since we pay our balance every month this is basically like having 5% more income proportional to the amount spent on Amazon**
We never have a problem dealing with customer service. A simple chat solves pretty much everything immediately, and when it's Amazon's fault they'll apply a courtesy credit towards our account. (Actually that hasn't happened a lot lately-- maybe something they don't do as much any more, or maybe they've calculated that we're solid enough customers that we're not quiting Amazon just because Amazon didn't give us free money)
*The first time we used Amazon Fresh was a complete epiphany for us. I was sick so couldn't go food shopping. Instead I gave Fresh a shot, and found out that on average their prices were 10% to 15% lower than I paid at the supermarket, and* we still got the same 5% cash back from using our Amazon credit card. Even after a good tip for the driver, we were still saving money. In terms of selection, they probably lack about 10% of the things we normally buy or they're much more expensive (meats especially) so I still do the occasional trip to the local grocery store to stock up.
But Amazon has more information and uses that too. They know loss rates by zip code or by block by carrier.
Usually items I order for them that are higher shrink risk don’t go through their Amazon distribution network and come UPS or USPS. Anecdotally, I’ve noticed those returns are easier — they probably dump cost on the 3rd party shipper.
I asked the question because they spoke with the assurance of an insider... so I wondered if they worked on it.
Taken to the extreme (and I don’t see anything in the argument that prevents it being taken to the extreme) and combined with acceptance of utility functions (_i.e._ that the value of anything is comparable to anything else, including human lives) this seems to also reduce any and all rule enforcement to wergild or something radical in that vein. Which—I’m not sure how it worked historically, but it also seems to be the current state of patent law (it doesn’t matter if the patent has merit, the side with the better and more expensive lawyers will starve out the other), and the results look utterly miserable.
Vague generalities aside, the complaints (and hypothetical downvotes) seems to indicate that, however simple and transparent a rule this is from the inside, people don’t want to pay for what is not their fault (cf experiments where people rejected a $1/$99 split in favour of a $0/$0 one to punish the other guy). I’m reminded of people queueing up in droves to the first McDonald’s in Moscow nor for the food or the prices, but for the customer service, because that was literally the first time in their lives they could get that anywhere.
Unless it’s marketed as a luxury good, people will actually pay for decent customer service that, among other things, satisfies their sense of fairness. That a model which does not reflect this can be simple is not a virtue of that model, it’s shoddy model-building. Not providing that service might still be the better economic decision (with caveats), if few enough people end up bitter over it (and complain on HN, etc.), but that’s a much more involved claim than “but it’s so simple!”. It also has very little to do with its moral merit, and if the better economic decision is too far from the moral one, it may well be time to go tune the incentives again and not to give up on your set of morals.
I’m not convinced customer protection regulations are the answer here, but the results of no customer protection regulations also seem completely idiotic from the outside. The third option of piling upon the company online also seems revolting for entirely different reasons.
They're not always the cheapest, but the overall experience is great - and I just love that I can see all my orders for such a long time span.
Now, I've always had good customer support experiences with the company, and your model helps explain why I suppose; loyalty has a value of its own.
Sure, it is, when you're rich. If you're some poor bastard who's scrimped up a whole lot of money to buy an expensive item, and Amazon figures you won't spend much money in the future, you get ripped off.
There's a couple of externalities you've left out, too, because your LTV doesn't predict them:
* Reputational costs
* Risk of adverse legal outcomes, where you decline to provide deserved refunds after failing to deliver merchandise.
You’re talking like there’s no legal recourse (or even financial, like a chargeback). There simple aren’t enough resources to track down the source of every sob story, the hard truth is this “inhuman” efficiency is a huge boon for the vast majority of us.
There may be very valid reason to shortcut claims and to grant dubious claims of high value customers. But you really shouldn't just ship an empty box and say "sorry, you're not getting your money back unless you sue us". (And worse, break the customer's existing digital content if they do, because you can hold it hostage).
I mean, I get that it's financially appealing to externalize the costs of in-warehouse and in-transit shrinkage to random low-value customers, but brah-- that's just not right.
IMO this should be a flag for regulators to start taking a look, too, to understand how systemic this is.
If Amazon chooses to refund without question to save disputes money, that's their choice. But it's not the customer's problem. If you pay for something and you don't receive it, you must get refunded.
I wonder if that factors in somehow.
I also wonder - since the box containing five CPUs looked like a "single product" box - how often that must have happened.
For instance, in the short term it's better to take the loss instead of investigate. But that then makes your platform an easy pick for fraudsters, so you end up with way more fraud and way more costs overall. If you were a bit more thorough and spotted scammers better, you would have a higher cost _per incident_ but lower cost overall in the long term.
This is particularly true if you are such a big player that you are the market in practice (so general equilibrium effects cannot be ignored anymore). And reminds me a bit of what eBay did in the past (sided with buyers almost every time without investigating, and made selling much more difficult).
Where people haven’t seemed to have learned their lesson with this is with airlines and to a lesser extent hotels. People will just buy the cheapest ticket, which is completely fine if that’s what you want or need to do, but they then complain that the airlines don’t bend over backwards when things go awry.
Example: you have to check in to international flights 60 minutes or more before the flight leaves (from the US). This is a reasonably strict rule. I once arrived 45 minutes prior and they checked me in anyway. I give that airline a lot of business and they know it.
So I order a lot of packages from Amazon and have had zero problems with returns and cancellations but almost all my purchases are low end.
For, say, new electronics, I’ll just go straight to B&H or Adorama. Or even Best Buy.
No. If you're 30, they might (or might not) group you with other 30 year olds, if that's important, and run this model against you all as a group. But you don't punch in "he has 50 years left to live" in the model. The model doesn't know any of that, it just knows your transaction history, and it works better when the population of histories are more similar than dissimilar. Though it's also very effective when run against a large population too.
Basically, these models can pretty much predict whether you are pathological or good one. I know that models also include things like if you are a pet owner, number of pets, siblings, number of disputes in cc, etc.
(I have never bought anything expensive from Amazon, and so I have no idea what its stated policy is.)
Behaving otherwise would just be shooting yourself in the foot for weird ideas of fairness.
I'd say three centuries, but who's counting ...
(or whenever printing got cheap enough to use for advertising)
What is frustrating to people is: it doesn't matter what happened to the package. You feel like you're this good person pursuing this great justice, but the simple facts are it doesn't matter what actually occurred. This is coming from someone who had two $3,500 computer deliveries stolen, probably because they had RTX 3080 written on them, but like, Amazon didn't care, it just refunded me. I didn't have to provide evidence or explain anything. Because my LTV was really high. Same with getting stuff refunded on my credit card. They simply did not care, and it was thousands of dollars of stuff where, obviously I was in the wrong, you can't just refund like airfare and such, but I asked for it and I got it because LTV.
Is this just? Like how else could it work? They investigate everything, spending more money than the cost of the refund? They would just not provide refunds, as was the case historically in e-commerce! And it's not Amazon, it's everyone. Indeed, the real innovation here isn't the strategy, it's that LTV forecasting has gotten really accurate. Otherwise I feel like these threads devolve into, people who don't know anything complain about stuff they keep using, and nobody critically looks into what's really going on.
There's an emotional desire, ironically, for this slow, legalistic, argumentative, plaintiff versus defense world. You know, when it suits you. The algorithmic approach, as soon as you open that Pandora's box, well of course it optimizes all that crap away - the LTV model of "justice" here is, in my opinion, a lot lot better.
It's just so inscrutable to a lay person, such a baffling reorganization of at least two decades of conditioning, I can see how the reaction is, "Oh my god, who is this blowhard, downvote." All I'm trying to do is share the facts of how remedies really work when you interact with giant, growing, successful companies.