The Business of Emotion

The Business of Emotion is a paean to the era of big data by Big Data: monetizing attention, addicting users and somehow creating loneliness in the biggest, most sociable crowd ever seen on earth. It’s dark, dystopian, dehumanizing. It’s about how faceless organizations use data to turn people’s attention into money.

I love it. It might be my career’s theme song. But to understand why, you need to sing it with conviction. 

Bear with me. I need to take you on a brief, nerdy, detour.

This is one of my favorite scenes from Star Trek. (I warned you). In it, Patrick Stewart learns that a god-like being Q is testing all of humanity through a game. Q expects humanity to fail, and quotes Shakespeare at Stewart. So Stewart quotes Hamlet (a pretty negative guy all around) right back at him: 


“What he might say with irony, I say with conviction”. 

That stuck with me. Stewart takes a cynical, sarcastic judgement of humanity, and with full appreciation of current  shortcomings, and chooses to bet on progress and growth as the core of what we are.

It’s easy to be cynical. Pessimism seems smarter than optimism. The ability to say “we’re going to work hard to do the right thing for our customers” can sound naive, and some in finance would outright snark and smirk at it. Well, plenty of people live their lives shorting virtue, I’m sure they’ll get along fine. That’s just not my bag, baby.

I’d rather speak with conviction.

Back to our main act. The Business of Emotion. Listen to it or read the lyrics.

Now, let me say with conviction what he says with irony.

Oh I been watching you/ I’m gonna get you high

The things I do to you/ They’re gonna make you cry

So, yes, behavior is tracked. At some level, any web company has to. We need an auditable trail of if/when a client or employee did something. Timestamps, location data, pageviews: it all automatically generated from the fact that your interacting with the web service. 

So, how do you helpfully get people high, and make them cry? Encouragement, positive feedback, relief. I once received an email from a client who was a long-distance trucker, who used us to save before he spent the money. He was proud of how much he’d saved up, and how it meant he was more secure and less stressed than before. When clients withdraw money to make a house downpayment or buy a new car, that’s a reason for rejoicing. 

That email was better than cocaine.  (I’ve never done cocaine, but I’d hope so. )

Doesn’t matter if you like it or not

Doesn’t matter if you don’t wanna play my game

You signed up, You gotta participate

Ok. so this one is a bit trickier, but it’s worth discussing.

I’m helping design a piece of technology that will run, automatically, for (hopefully) millions of people. Mistakes scale up just as fast as successes. It’s important to pay attention to not just the median client, but the whole distribution, and try to make every change a Pareto improvement.

We test new products and services, we do it as an experiment, and we try and learn and iterate. Most of the time, the people who are experiencing the ‘alternative’ design or service won’t know about it.

That can be important. We act differently when we know we’re being watched. Placebos are surprisingly powerful. I want to know people are acting honestly, and that if if a new design is ossified into the system, it’s not unintentionally hurting anyone for years.

Second, and equally if not more important, is adverse selection. Suppose we let clients opt out from ever seeing the ‘alternate’ design while it was being tested, and that people who opt out tend to fall on the extreme of some behavior relating to investing. A design might deployed into the world which has a disproportionately large effect (good or bad) on the very group who we have no data on. That’s scary. 

Doesn’t matter to me cause you’re all the same

By default, all clients are just a number. Or a whole bunch of numbers, really.

And I’m not sure as a client, I’d really want it any other way. Do I really want service providers treating me differently because of how they perceive me? Their biases about my name, ethnicity, gender, skin tone, or taste in music? Should I help more attractive clients first? Or perhaps the wealthiest?

Doesn’t matter to me, cause you’re all the same.

That said, I likewise think they’re all different. We should make left-handed and right-handed scissors. We should create a service which is systematically personalized, not ad-hoc or haphazardly personalized based on the perceptions/biases of one person.

I do respond to client emails and take phone calls, and at that point they become more than a number: a voice telling me how they understand the world, where they find themselves, and where they’d like to be. And that can hit you in the gut with emotion. But you have to take that fire that this actually matters and keep it burning for all the people who don’t call in, who didn’t raise their voices.

Because they’re all the same. They all matter equally. 

Feel good

Make you feel good

I’m looking for emotion

So I know just what to show you

Yes, duh, of course, come on. I want to make people feel good. People who feel good persevere longer, solve problems better, and I want people to enjoy the time they spend with my product, and tell their friends about it.

What he says with irony, I say with conviction.

It’s worth noting that this shouldn’t just depend on the good will of the designers. Incentives matter. I chose to work for a company with a client aligned compensation strategy so that I don’t get paid to be the investing worlds version of an attention merchant

And remember, I’m not only a Betterment vice president, I’m also a client.


Incentives and behavioral design: brokerage UI edition

One of the most powerful behavioral tools is… incentives. And how commercial incentives influence behavioral design deserves some more attention.

Let’s quickly compare the incentives of brokers versus fiduciary advisors.

Brokers generally make money when customers trade, either through direct trading fees (e.g. $7/trade) or routing trades to a liquidity provider who pays the broker for those trades. Brokers lend out client cash balances, and do standard banking lending. And they get paid by fund companies to prioritize the fund companies funds, either by making them more salient, or allowing them to be traded for free (these are generally called ‘platforming fees’).

Fee only fiduciary advisors are paid directly from the client for their advice and ongoing management. Many use an assets under management based structure like fund companies do, though there is a movement towards flat fees or annual retainers.  There should be no conflicts of interest like commissions or kickbacks from fund providers. The more happy, wealthy clients an advisor has, the more money they make.

Now, I want you to imagine you’re a behavioral designer working at either a broker, or a fiduciary RIA. You are still part of of a commercial profit-seeking business, so you’re expected to contribute to the bottom line, either directly or indirectly. What impact do  incentives have on your work?

Let’s use as an example recent research on giving brokerage customers personalized feedback about their activity and success: over 18 months clients were given monthly feedback on how well they had traded considering costs, risk taken, and diversification. Customers who received the performance reports traded less, diversified more, earned higher returns, and became more successful investors over time. Pretty valuable stuff to the clients! 

Now imagine the conversation between the behavioral designer and the business manager:

Hi boss, we have the results of the personalized performance report!

Great, what was the outcome?

It was incredible! Customers who received the reports held more diversified portfolios, traded less, and had better performance.

Uhm, traded less? So if we gave this to everyone, we’d decrease our revenue?

Well, yeah…

Great work. Please shred the report and let’s not talk about this again. And please focus on getting people to trade more, not less. 

Now imagine if you worked for a fiduciary advisor: it’s a win-win. Clients grow their money fast, and you’re providing a valuable service to help them do so.

The crux is that brokers incentives aren’t well aligned. Clients want to grow their money and be smart. Brokers want them to trade and hold cash. The design you’ll see at a DIY brokerage has been designed, either by commission or omission, to maximize trading and cash holding.

If, theoretically, I worked for a brokerage, how would I design things to maximize trading and cash holding?

  • I’d use color to make gains and losses very emotional: green for gains, and bright blood-red for losses.  → more trading
  • I’d default clients performance view to as short a time period as possible, ideally daily, to maximize myopic loss aversion. → more trading
  • I’d emphasize individual positions, to maximize narrow framing. Diversification means there will likely always be something underperforming, and I want to focus them on that. → more trading
  • I’d put up biggest gainers & losers to focus them on volatility and opportunity → more trading
  • I would would not make a dividend reinvestment program the default. → more cash
  • I’d make cash look safe and reassuring, and not proactively push them to invest it. → more cash
  • I’d definitely not mention inflation. → more cash

Here’s a screenshot from the web of the logged-in homepage of one of the biggest brokerages. See how many items you can find.

The one element I haven’t see is this: the disposition effect tells us that clients will hold onto losers longer than winners. That means less trading. So I’d want to give clients a positive emotional reason to trade out of losers. Tax loss harvesting works in taxable accounts, but not tax advantage accounts, so it’s not universally useful. 

So, I hope you can see why I’m very cynical about all the ‘free trades’ brokerages out there today. Much like a casino offers free drinks and perk for big rollers, a ‘free’ brokerage isn’t arranged that way because it’s a charitable organization. These brokerages have just figured out other, less direct ways of charging you, and still making money. Minimizing your trading costs at the expense of maximizing your money’s growth is just silly.

To be clear, these designs also reflect what brokerage clients like to see, and what it’s easy for brokerages to build. People who want to trade like detailed returns information. The page is full of ‘historical facts’ which are easy to produce, but very little ‘future insight’ which is expensive. It’s the convergence of broker incentives and client motivations which create this design.

So please, alway consider the incentives of your financial counter-party. Know how they get paid, and what that means about how they’ll optimize their service to you. If you aren’t paying, you aren’t the customer. You’re the product. And generally a company spends time optimizing how to serve its customers, not it’s suppliers.

Outsourcing self control

Self control is one of the biggest struggles most people have. Me included.

I’m beset by the near-term desire to behave in many ways that I wish I didn’t. I snack too much, and drink too much. I spend too much time on twitter.

There are a few approaches to to deal with this:

  • rely on my willpower
  • pre-emptively avoid tempting circumstances (hacking the akrasia horizon)
  • create quick feedback loops which deter it
  • Commitment mechanisms: out-source the self control.

The technology version of last option is the newest. It’s a recent development that I can use technology to craft systematic self control, and so the most interesting to me. I often mention it as an example, and frequently receive follow-up questions asking what I really mean. Let’s dive in.

Continue reading Outsourcing self control