Comment by aelhaji | Hacker News Reader

Comment by aelhaji | Hacker News Reader

Parent Full thread

"Asking people what they’d pay for and how much rarely works."

Yes, asking a person about what he or she is willing to pay _hypothetically_ doesn't work. But there are numerous scientific studies that show that using a Vickrey auction, similar to Google's ad auction mechanism, can elicit a person's maximum willingness-to-pay. In fact, it's one of the reasons why William Vickrey won the Nobel Prize in Economic Sciences. The primary reason why you would reveal your maximum willingness-to-pay in a Vickrey auction is because you know beforehand that if you win the auction, you won't have to pay your own bid——you always pay less than your own bid as a winner. How much less is up to the auctioneer but in all cases it's less than your own bid. The pay-what-you-want mechanism cannot reveal how much a person values a product or service because there's such a strong financial incentive to simply state a low amount.

Disclaimer: I'm the co-founder of Veylinx, a platform to measure maximum willingness-to-pay of users using Vickrey auctions.

>But there are numerous scientific studies that show that using a Vickrey auction, [...], can elicit a person's maximum willingness-to-pay.

Yes, but participating in an auction is a concrete action to "commit to pay" -- and therefore not what the author is talking about.

The context of the author's quote you extracted ("asking people what they'd pay"†) is about abstract notions of what consumers think they would pay; e.g. gathering hypotheticals of prices from marketing efforts like customer surveys or focus groups.

Surveys and focus groups have no real money payment commitments. Auctions do.

† author also wrote a few sentences later: "When it comes to money, people are unable to predict accurately whether they’d pay or not. It’s much easier to spend hypothetical dollars than real ones."

I agree that skin in the game matters. But my point is a bit more subtle. Imagine the following situation: I present you a new product and I'm going to ask you how much you're willing to pay for it. Let's agree that you'll pay the amount you mention dependent on the flip of a coin. Even though you clearly have have committed to potentially paying real money, this method won't reveal what I really want to know: your true perceived value of the product (maximum willingness-to-pay). The Vickrey auction solves this problem.

Vickrey (or any) auction doesn't solve the problem if the amount of supply is greater than the amount of demand, which is the common case in software.

That's simply not true. If your maximum willingness-to-pay for a SaaS solution is, let's say $5 per month, then a Vickrey auction will reveal that and the seller can decide to sell it to you at, let's say, $3.

So you set up a focus group of potential customers and vickrey-auction (or VCG-auction) one subscription (or k subscriptions) among them, for the purpose of recording the bid distribution?

For this to work, wouldn't the participants need to believe that the auction really is the only way they can get a subscription?

Depends on what you would like to focus on. If you want to capture the complete demand curve for a product, then you're right. In that case the participants shouldn't be aware of any other alternative to get the product. But you could also choose to focus on users who aren't willing to pay the current price and better understand their distribution. This would allow you for example to determine which promotion would most likely turn a free user into a paying user.

I'm not familiar with Vickrey auctions; can you explain how they solve the problem the author is referring to?

With eBay proxy bidding, you tell eBay what the maximum is you are willing to pay. Then, if you win the auction, you don't pay your maximum. You pay just above the maximum amount of the bidder you beat (i.e. the person who would have won if you hadn't participated).

So, how should you bid if you are a rational person? You should bid the maximum you are willing to pay. As there is zero chance that you will pay more than you needed: you will always pay just enough to beat the second place bidder.

So, the incentive is such that you reveal your maximum price to eBay.

The eBay application isn't really appropriate because eBay always reveals the highest bid, which removes the incentive to reveal if you have a lower willingness-to-pay. A pure application of the Vickrey auction gives all participants an incentive to reveal their maximum willingness-to-pay.

"eBay always reveals the highest bid, which removes the incentive to reveal if you have a lower willingness-to-pay"

I don't understand how it removes the incentive? Would you mind going into more detail?

(I ask because I believe I am rational when I place eBay bids, and I always choose the actual maximum I'm willing to pay.)

Let's say that the first bidder placed as a proxy bid $100 but eBay only shows $1. This will already exclude everyone who has a maximum-willingness-to-pay lower than or equal to $1. A second bidder might put as a proxy bid $120. This wil result in eBay setting the highest bid at $101. At that price nobody with a maximum willingness-to-pay lower than or equal to $101 will reveal their max. The Vickrey auction doesn't reveal anyone's bid and, thus, you collect a lot more bids (demand curve data). I agree with you, however, that the eBay mechanism results in the same person winning the auction.

When I bid on coins I NEVER bid the actual maximum on ebay; I just use gixen.com and place my actual willingness to pay there.

The reason is that for most goods (luxury goods, stuff with resale value, stuff where you might now all the details, etc.) you are actually dealing with a common value auction, so you DON'T want to reveal your cards before hand.

For instance, suppose I see an auction with 5 existing bids.. I stop and say wait that's probably something interesting, and start doing a bit more research on the value of the coin.

Similarly, I might do some quick research and decide that a coin is valued at $10, but if I see bids at $100 that means that either my prior was wrong or the bidders are wrong, and thus I do a bit more research and try to see if there is anything special about the coin (e.g. the quality is actually higher than what the auction description says).

Then besides all these rational motives, you have the irrational ones. A lot of people bid on things almost like gambling (bid the maximum+1, if that doesn't make them the highest bidders, then keep bidding). If you don't place your bid out there, you prevent this problem altogether. [1] https://en.wikipedia.org/wiki/Common_value_auction

The basic problem is what price to set. You don't want to take the risk to set the price too high (probability of not selling at all) but at the same time you don't want to set the price too low (profitability). What if you could measure the complete demand curve for a product? Then you could simply select your ideal price without taking any risk while maximizing revenue or profitability. Vickrey auctions allow you to do exactly that because it incentivizes participants to reveal their maximum willingness-to-pay and a demand curve is simply the set of maximum willingness-to-pay sorted from high to low. How is that possible? Well, it involves a bit of game theory: https://www.quora.com/Why-in-a-Vickrey-auction-does-one-pay-... (Note that a Vickrey auction can be generalized to more than one winner and, thus, the price is not always the second highest price.)

auctions also put up pyschological barriers for some customer segments, so aren't always the best choice.

another (classic) alternative is conjoint analysis: https://en.wikipedia.org/wiki/Conjoint_analysis_(marketing)

it's difficult to do well, but will reveal willingness-to-pay without requiring the commitment of real dollars.

"But there are numerous scientific studies that show that using a Vickrey auction, similar to Google's ad auction mechanism, can elicit a person's maximum willingness-to-pay."

Sounds like a catch-22 situation, how did these studies determine they actually measured a person's maximum willinges-to-pay? Even when a person has decided what their maximum is, there are many known sales tactics that can increase it beyond what the buyer initially decided. Some, like bait-and-switch are so effective that they're outlawed. So, it'd be really interesting to hear how these studies actually determined they actually measured what they said they did.

It's done with so-called value induced auctions. In this type of auction you simply bid on a voucher that's worth $x dollars, where x is different for each participant. Theory predicts that your bid should be $x (your maximum willingness-to-pay). Considering value is induced, researchers know both the prediction and the observation, and calculating the difference determines the 'truth revealing efficiency' of an auction. See for example: https://www.sciencedirect.com/science/article/pii/S016726810...

Reply on news.ycombinator.com