Advertisement or… negotiation?
Last Sunday I had the pleasure of meeting Anna Bernasek and Dan Mongan, who came to the Alt Banking group meeting to tell us about their book All You Can Pay: How Companies Use Our Data To Empty Our Wallets.
While they were discussing their book, the topic of online advertisement naturally came up. Dan and Anna made an interesting point in that discussion which I’ve been chewing on ever since. Namely, they provocatively suggested that we should never use the word “advertising” to describe the complicated and sophisticated process of tailored and targeted offers to an individual internet browser. Instead, we should call it a “negotiation.” Let me explain their reasoning.
They started by introducing the concept of a “consumer surplus.” This is the difference between what a given consumer would be willing to pay for a product versus what the price actually is for that product. If the difference is positive, the consumer buys the product and has a theoretical bit of “extra” money in their wallet, which corresponds to the happy fact that the price was lower than their maximum price. If the difference is negative, the consumer doesn’t buy the product because it’s too expensive for them.
Does that make sense? For me it only kind of makes sense. I mean, it makes quite a bit of sense for certain situations, like when I’m buying something I don’t actually need, out of funds I consider limited but discretionary. So, for example, after my soggy experience a couple of weeks ago I threw away my faulty tent, so I’m looking at buying a new tent, but then again I’m not going camping any time soon, so the price has to be right.
Here’s an important example where it doesn’t make as much sense. If I’m poor and I need a car to go to work, and I know I have to borrow money to buy the car, then the amount of the loan is less important than the fact that I can get a loan in the first place. The “consumer surplus” is less obvious when it’s a matter of debt rather than cash, and when it’s a desperately needed purchase rather than discretionary.
Anyhoo, let’s imagine consumer surplus to be a thing that people consider useful. The point Is, big data companies are getting better and better and determining or at least approximating what a given consumer’s surplus is, and making offers accordingly.
So, if you’re on a platform where there are advertisements, the algorithms behind those offers might infer that you have a certain amount of money to spend on a car or hotel, and might offer you certain models of cars with or without leather seats, or hotels in certain neighborhoods priced in a certain range, to squeeze out your consumer surplus. Yes, there might be cheaper versions of these things that you’d be happy with, and that would save you money, but the algorithm has determined your spending power, which is all they care about.
Thus, instead of an advertisement, which sounds like something we imagine many many people see, this is a personalized negotiation, just to you personally, and crucially, you don’t see what other people are being offered. So it’s a one-sided negotiation with enormous information asymmetries.
Note this is the opposite of what we’d expect in the “free market,” where all the offers are on the table and you get to choose the one with the lowest price. Partly we ignore that vision because many of the platforms we now spend time on are more or less monopolistic in nature, and partly it’s because of the nature of the auction system of advertising: the “ad” for a more expensive hotel room, that’s still in your price range, will win an auction because it is worth more to the seller.
I think it’s a pretty good distinction, although I’m not sure “online sale negotiations” is a catchy enough phrase to replace “online ads” any time soon.
mathbabe 
It’s an interesting point, but “negotiation” doesn’t seem at all an apt choice of word, since the consumer has no way of negotiating with the company. “One-sided negotiation” sounds like an oxymoron to me, though it may be standard jargon in their field.
LikeLike
Considering how money is distributed in the US, the consumer surplus approach makes a lot of sense. And yes the internet has plenty of opportunities to exploit lower income people as well. There’s plenty of predatory loans being advertised on the internet aimed exactly at those who need a car and can’t afford one, although a lot if that advertising is on the radio and local television stations.
LikeLike
I agree with Ernie, but I think this sounds more like online coercion, the coercive force being the huge information asymmetry. This also sounds catchier than online sale negotiation.
Also consider the ‘weaponization’ of such a technology to make certain goods or service seem unattainable (have negative consumer surplus) by certain people or groups of people which would otherwise benefit from lower cost options if they were presented with them. E.g. exploitative car title loans vs secured property loans from a bank with reasonable rates.
LikeLike
If you prefer, one other way to think of consumer surplus is the collection of other goods you could have purchased with the leftover money.
The loan example points out several important complications:
(1) what is the product, really? In this case, the loan is a product in the form of an on-going relationship with obligations on both sides (though the two-sided nature isn’t true once the initial loan amount is disbursed). There are other products that are clearly structured like this (gym membership) and yet more that can be considered in this way if you see it as a repeated game (price of milk at the local grocery store). When there are different perspectives, the sense of how much surplus exists may depend strongly on how you frame the product.
(2) what is the pricing and does the consumer understand the price? Pretty clear that the price can be made very complicated to understand.
(3) what further information is revealed by the price that is paid? A borrower willing to pay a little too much for a loan is great for a bank, a borrower willing to pay way too much is probably a future default.
In addition, here are some other items that seem important to the consumer surplus model:
(a) our indifference point (max price at which we don’t care whether to buy or keep the cash) is not fixed, it is time-varying. One of the goals of online ads is to get in front of us when our indifference point is relatively high.
(b) our indifference point is uncertain. Because the seller’s payout has a cliff (they make more profit the higher price, up to the point I no longer buy, where they make 0), the more uncertainty, the lower their optimal price point.
(c) advertisement can influence our indifference point. That could be from providing real information (car X has better fuel economy than car Y) or extraneous information (car X has a sexier spokesmodel than car Y).
(d) we can gather information to change our indifference point (but gathering information may be costly).
A key point is the information asymmetry. Who knows the indifference point better: the buyer or the seller?
LikeLike
I think they are wrong: their premise is that purchasing is essentially a left brain activity, where reason triumphs, their logic essentially that buying is triggered when value to the buyer exceeds price for example.
But – as a trained copywriter – I can say that the process of persuasion with copy is very much right brain , playing on emotion, such as fear of loss, done with sophisticated subliminal messages which is why it is so dangerous, and why people who really cannot afford stuff can be persuaded to buy it. Split testing shows small details which are largely inconsequential in the left brain decision massively increases the chance someone will buy, illustrating that the decision is governed by right brain.
LikeLike
I want to play devil’s advocate for a moment. I completely see where the algorithm presenting data to you can seem coercive for brand A, and their goal is to get you to pay — as as Josh so nicely puts it — as much as possible before you fall off of the cliff and give them zero. If only brand A advertises on site Z, then there is potential for brand A to use their information to get you to buy their brand.
What I can image that others would argue is that computational power also opens up the possibility that you leave site Z and go to site Y to comparison-shop. Site Y might have an interest in getting you to pay more, but has a greater incentive in creating a repeat-customer who will return to their site anytime you want to comparison shop. They show you a lower price for the same product made by brand B. They get some cut of the commission, which although potentially lower for this transaction also raises the possibility of future earnings from your return to their site. In other words, the market has a solution to this problem.
This would be the role that sites like Kayak play in the airline and hotel industry. Their incentive is to get you to return to their site to continue receiving commissions. They may try to target you to pay a certain price, but if you find out lower prices exist, you will be unhappy and stop providing them with commissions.
LikeLike
Interesting research taking clean profiles and varying them only by gender appears to produce 1800-300 high value ‘ad negotiations’ for men vs women. I could only image how much it could vary taking other economic or browser history factors into account.
http://spectrum.ieee.org/tech-talk/telecom/internet/women-less-likely-to-be-shown-ads-for-highpaying-jobs
LikeLike