Live data from Hacker News

Walmart: ChatGPT checkout converted 3x worse than website

searchengineland.com

291–300 of 300 posts

Re: Walmart: ChatGPT checkout converted 3x worse than website

#291
post #76

Earlier quoted context omitted.

We are talking about a hypothetical sales chatbot which would be built alongside the business, so they absolutely have the capacity and information necessary to train the chatbot to advise their own clients.

> they absolutely have the capacity and information necessary to train the chatbot to advise their own clients. That doesn't follow. In fact, having this capacity and information creates a moral dilemma, as giving customers objectively correct advice is, especially in highly competitive markets, bad for business. Ignorance is bliss for businesses, because this lets them bullshit people through marketing with less gui…

My brother in christ, the function is just that of a salesman, you are pondering philosophical about whether a salesman's actions are ethical.

Company makes a chatbot that sells their product and advises customers on what to buy.

To be fair with you, you are not asking the wrong questions and you are on to something, it's just that it's basic if you are into sales. Read up on the subject, a source I would recommend, although obscure, is Claude Whitacre. I believe in "Sales Prospecting" he talks about this specific dilemma of giving good advice vs selling your product. He argues that a good salesman will give good advice over selling their own product, and that this is beneficial because it creates trust in the salesman and might result in more sales down the line even if the very specific interaction didn't result in a sale.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#292

ChatGPT recommended me some good hard drives for price per TB, and one particularly cheap one had direct checkout with Walmart, so I tried it, because why not? It let me get all the way to the payment step before it told me it was out of stock. Walmart's website told me it was out of stock when I decided to click on the link. This is probably part of why it doesn't convert.

meta-search sites in travel (Skyscanner, etc) have the same problem of out of date cached information from the merchant. not solvable in practice because there's so much traffic volume and low rate of conversion that you must cache for cost reasons.

Those typically have a list of vendors from lowest to highest (cached) price. IME the prices are correct more often than not but if the best deal has expired you can still check out the next best one.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#293
post #13
post #10

Earlier quoted context omitted.

It depends on the product, if we are talking commodities or mass produced products like groceries, sure. If we are talking custom products or complex appliances that need a lot of guidance, then maybe chat interface is appropriate.

The problem is that the chat transcript is legally binding. If the chatbot makes incorrect statements which the customer relies on for their complex purchase, then you're going to have to refund them. https://www.bbc.co.uk/travel/article/20240222-air-canada-cha...

When comparing a sales chatbot agent with a sales human agent, there is no difference. A chatbot answer would be no more legally binding than a human's answer.

The actual reduction in liability would come down to who can make less mistakes, the fact that the mistakes are or aren't legally binding does not change the decision to put a chatbot.

In fact I think you are plain wrong, it's standard case law that errors in pricing are not binding, if you find a wheel of 50kg cheese priced at 30$ (the price of a kg of cheese), then the supermarket is not obligated to honour that price, they obviously mispriced the cheese.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#294

Earlier quoted context omitted.

Exactly. Just because you're not good at something doesn't mean you don't want to be.

wanting to be good at something is not exactly the same as being interested in getting good at something, which carries the additional, nuanced, connotation of investment (or willingness to invest) towards this goal. e.g. many people want to live an active healthy lifestyle, but fewer are actually interested in doing so.

> wanting to be good at something is not exactly the same as being interested in getting good at something

Are you sure? I've never heard of that difference. To be sure, I checked the definitions for each, and could find no such distinction. Maybe you could provide me with better sources than the ones I found?

Re: Walmart: ChatGPT checkout converted 3x worse than website

#295

Earlier quoted context omitted.

There’s a difference between being interested in getting good at something and being good at something

Exactly. Just because you're not good at something doesn't mean you don't want to be.

Wanting and being interested are not things in physical reality. We are also talking about an organization and not some aspiring teenager.

If someone wants something , the only measure of their interest and desire is how much resources and time they allocate to it. We’re not measuring Sam Altmans deepest desires and fantasies here

Re: Walmart: ChatGPT checkout converted 3x worse than website

#296

Earlier quoted context omitted.

Check out UCP

do we need another protocol for this? seems like a stream data problem more than anything

Agreed, but “protocol” is changing its meaning. MCP is t really a protocol either.

They’re both better described as conventions, but whatever.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#297

Earlier quoted context omitted.

If I can tell ChatGPT to order me what I need to replace the bottom element on my water heater, and here’s a picture of the ridiculously long model number, that saves me quite a bit of time. I used that as an example because I did that last week, apart from me just going to the store to get the links that it brought up.

Conversely, I did something similar, but I took a picture of the model number, copy/pasted the text from the image into a search engine, and that was that. Either way you're taking a picture and performing a manual step (query an LLM, query a search engine) I would never trust an LLM to accurately identify and purchase something for me based on a picture, a prompt, and a prayer.

I mean, it was easy to confirm they were the right parts. Sure, I could have done it in a few minutes of Googling, but I was already asking it to help me troubleshoot (which it did) and instead it took me seconds.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#299

Earlier quoted context omitted.

meta-search sites in travel (Skyscanner, etc) have the same problem of out of date cached information from the merchant. not solvable in practice because there's so much traffic volume and low rate of conversion that you must cache for cost reasons.

Those typically have a list of vendors from lowest to highest (cached) price. IME the prices are correct more often than not but if the best deal has expired you can still check out the next best one.

Well, my point is that there is always a certain level of incorrectness and that can never be fixed because of the cost of retrieving up to date pricing and availability.

Re: Walmart: ChatGPT checkout converted 3x worse than website

#300

Earlier quoted context omitted.

I use AI to shop and it seems easy for an AI to understand when that’s the case (they can do statistics and physics after all).

Sure, I use it for that too. But we’re also talking about most people. Your average Walmart customer is not going to be your average HN reader.

True, though I don’t think they’re too different on that dimension.
Post reply on HN