Recommendation and discovery and Chris Dixon, oh my!

Chris Dixon recently wrote (in an incredibly brief post) that recommendation systems are features, not products, and pointed out that Hunch was only compelling after being bought by eBay, where it successfully powers both recommendations and discovery on the site (not 100% sure on this).  I agree with Chris and have a few notes.

1) Recommendation takes initial desire and channels it: you want a TV but you’re not sure what is the right kind, it tells you the right kind and thus makes it easier (lowered inhibiting pressure).  Discovery creates desire (higher promoting pressure): you weren’t thinking about food, but you are now.  People tend to mix the two up.

2) The reason that recommendations don’t stand on their own is because the intended action is to satisfy a desire.  If you are looking to buy a TV, the end state is the purchase and telling you which one to get doesn’t end the needstate.  So you’ll get outcompeted by anyone that does (like Amazon).  The only place where you can have just a recommendation engine is where you can’t complete the needstate in that medium: I can recommend recipes, for example, because there is nothing more than can be done online.    But these are dangerous plays, because you can get outcompeted by anyone who brings it closer to closing the needstate. Recommendations worked until Amazon made it so you could get recommended and complete the transaction in one place (just ask CNET).

3) Discovery, in contrast, can work on its own.  Not just because of SEO (Chris’ explanation), but because true discovery isn’t the same as a recommendation: it actually creates desire, often from adjacent desires.  Take Pandora.  Nobody puts on Pandora to purchase new music – that isn’t the point.  But in the course of satisfying one desire (good background music), it occasionally creates another (the urge to buy a song).  If you are browsing blogs about fashion for entertainment, sometimes you find things that you then want independently of your previous desire: buying something doesn’t make you want to stop reading fashion blogs for entertainment  That is generally a clue that discovery happened, when you do something but it doesn’t diminish the need that you came for in the first place.

Bonus: discovery and recommendation map to the two ways I break advertising down for people often as well.  Some ads are about creating a desire for a thing you didn’t previously want, often by coupling them with other things you want (like selling beer with naked women).  You didn’t want a beer, you did want a naked woman, now you do go get a beer, though it doesn’t quench your thirst for a naked woman (studies suggest it may even enhance it).  Other ads, in contrast, are about taking existing desires and channeling them.  You were already thinking about buying a car, you should buy our car, because it is the best car.  It fits your need.  Buy it and you will no longer need to buy a car.

Making the unbanked into the banked with tax prep

I love AdventFinancial.  It isn’t just because their co-founder is a friend and one of the nicest guys I know, but they actually have a great business in a space that often doesn’t have much greatness: tax prep.

The pitch is pretty simple.  Mostly, two kinds of people get their taxes done in the US: the rich and the poor.  The poor know they are going to get a refund, so around tax time, prep offices spring up in poor urban neighborhoods (generally with dancing guys in Statue of Liberty costumes) and the poor enjoy one of the most bizarre business models that actually works.  They get their taxes done for free.  Or at least for “no money upfront”, which feels like the same thing (although of course it isn’t and there is a case to be made for the IRS to just proactively send those folks a letter and a refund check, but that’s another blog post).

Basically, because they know they are going to get a refund, tax preparers will do a poor person’s taxes for no money upfront and then take their fee out of the refund itself.  No refund, they don’t get paid (they used to do it as a “loan”, but the federal government mercifully stopped that), so incentives are aligned.  But what this creates is the need for a middleman, who takes in the refund check and parcels it out to the right folks.  Traditionally, this was done with checks; middleman gets the check, shuffles one off to the tax preparer, and then the poor person has to go back to the tax preparer’s office to get theirs.

Which sucks.  You get your taxes done and then you wait for the refund (which can arrived whenever the government decides to get to it), and the tax guy has to call you or you have to call him, and then you have to go to his office and find someone to give it to you.  Not a good process.

Which is where Advent comes in.  You come in to get your taxes done and the preparer gives you a prepaid card from Advent.  You never go back to the office; Advent just texts you when the money is on the card and how much.  And because they are reloadable prepay cards, what they actually do is a bunch of poor people a bank account.  Direct deposit, ATM, debit – its a great passive way of increasing financial capability, because now they have access to a whole world of financial systems that allow them to save and spend money without the fees associated with the standard options (check cashing, etc).

And therein lies the genius of the product to me.  An often overlooked population (the underbanked) gets increased access, entirely passively, through a system they are already using and that they want to use.  Advent found the thin chink in the system and inserted the right funnel to both make them profitable and enable them to help people.  They didn’t create a new system or make people come to them: they found them were they are.  That’s worth copying.

Product = Perspective

I was lucky enough to get to act as an advisor at USV on Saturday as part of Gary Chou’s great weekend program that allows startups to get three perspectives from field experts on the products they are planning to release.  I say lucky enough because as Matt Smith of Shutterstock pointed out, being an advisor forces you to articulate your perspective.

Which was actually a big part of my advice to startups: have a perspective.  All three of the products I was assigned to review were taking giant stabs at a market, building platforms on an internet that is increasingly becoming about niches.  If you don’t know what you want your users to do, chances are they won’t either.  They were almost all trying to be everything to everyone, and in each case it hampered them from actually launching.

You see it in the UX.  When you’re trying to be everything, users tend to get just dumped into results or dashboards.  If there isn’t a wizard flow, at least for the first time user, than you probably haven’t articulated a clear enough vision.  Even Twitter, the great unwashed platform, onboards users by trying to get them to follow people initially.  “Oh, this is a thing for following what people say” leads to “Oh, and people will follow me and I can say things too”.  Even in the most indefinite product there is, still you find a coherent perspective.

You see it in engineering.  Every engineer everywhere should force their product team to articulate a coherent product perspective and refuse to build new features until they do.  Otherwise, you will either end up building the entire world or you’ll start building the entire world and then watch them cut code as they drop features to pivot into having a narrative.  Engineering time is the most valuable, most expensive time a startup spends, and running down blind alleys is the product team’s fault.

You see it in bizdev.  Having a business model isn’t enough: you need to know who your customer is and what they want.  If you aren’t meeting their need and either taking their perspective or incorporating their needs into your perspective, you will produce a product that no one will pay for and that does nothing.  The greatest fear of every bizdev person, and therefore the thing to fight hardest against, is that they will build a product that everyone “likes” and no one uses.

While I’ve never been big on Powerpoint decks, I do like elevator pitches.  And it is precisely because of the requirement of perspective: a very quick summary forces people to say “what does my product do?” and provide a coherent answer.  It is tremendously hard to face a blank page and come up with something, and dramatically easier to respond to a question, and yet we always think of the marketing of a product from the blank perspective.  The very first expectation I have as a user after learning the name of something is what it does.  Is it a game?  A utility?  Again, you can pull up examples where it is tough to do this, but those are mostly just foils: a large majority of products are defined by their function.

Think of your product like a person.  My name is BRAND and I…?  Now fill it in.

The Hedonic Treadmill, as explained through Training Day (and some Romeo and Juliet, XKCD, and hot showers)

I’m a fan of the movie Training Day overall, but there is a particular scene which I’ve cited time and time again in talks to explain the hedonic treadmill, happiness in general, and a sound psychological method for designing user experiences that keep people both delighted and satisfied.

It always pays to start with a few definitions.  Delight, in a psych sense, is the momentary experience of happiness; satisfaction is its long-term equivalent.  The hedonic treadmill is a way of visualizing adaptation: the tendency to grow used to whatever our current situation is.  The idea is that you have to keep getting more of something to remain equally happy, like needing to keep walking to stay in the same place on a treadmill, because you will grow accustomed to the amount that you have and it will no longer provide the same hedonic benefit.

Personally, I like to think of a shower: it feels hot when you get in, but without changing the temperature at all, it starts to feel colder over time.  So you have to keep turning up the hot water, a little bit at a time, to feel the same warmth.  It is unclear whether that is a better example, but I like showers better than treadmills.
So now, Training Day.  At one point, rookie cop Jake Hoyt is challenged by older cops about his inexperience, and he says he already has the streets figured out.  “It’s all about smiles and cries,” he says.  “You gotta control your smiles and cries, because that’s all you have and nobody can take that away from you.”

Secret of the streets, maybe not, but certainly the secret to a life of long-term satisfaction. Just like there is no better moment than the first one in the shower, there is no better moment than the peak of most of life’s pleasurable experiences: the first bite of chocolate, a particular part of a sunny day.  Because most pleasures come with a cost, whether it is a hot water bill or the calories in the chocolate bar, the maximum happiness/cost ratio is generally the earliest peak.  And that, right there, is controlling your smiles and cries.

If you like something, eating it all the time will actually reduce how much you like it – your brain will release fewer happiness producing chemicals with every repeated instance, unless it is allowed to rest in between bursts.  But the costs remain the same for each burst: you get the same calories from every bite of candy bar, you just don’t get the same rewards.

Thus, if we control what makes us happy, and try to break it up into small bursts that are suitably far apart, we can avoid the need to constantly turn up the heat in the shower.  As Benvolio says in Romeo and Juliet, “Away, begone. The sport is at the best.”  Get out, while you’re still at the peak.  Five two minutes showers is better than one ten minute shower (unless you’re actually trying to get clean).

So product development.  If you constantly bombard your users with positive benefits, particularly the same positive benefits, they no longer act as any kind of motivational reinforcement.  A song on repeat is not a good song.  Good product folks have to build products that allow people to control their moments of happiness, looking for ways to loop together peaks rather than simply make massive, overwhelming experiences that grow stale.  Because the costs stay the same, for you and the user: they have to keep clicking the button, you have to keep serving up the content, and you want to save that for when it matters.

You have to allow your users control over their own happiness, with occasionally serendipity thrown in.  You don’t put Pandora on a party, you throw it on as background music, and when the algorithm is smart, it knows that and performs its lovely background function, while occasionally throwing in songs out of no where that make you stop and dance and sing.  Or maybe just dance, since your coworkers are now giving you strange looks.  It fulfills its basic function, but looks for the places where you have controlled smiles.

Know how your building fits into people’s happiness and try to honor that.  Look for ways to be the first bite of chocolate, periodically.  It is like XKCD – if there was a new one available all the time, I’d get bored.  Your product, whether you like it or not, always occurs in a context.  Figure out what it is and start tuning for not just the moments of happiness but the long-term satisfaction of correctly timed bursts of brain chemicals.  Science…its in there.

Variable pricing is a bad idea (or How your store might become a giant produce section)

The New York Times recently wrote about variable price shopping: the idea that some grocery stores are starting to think about changing the price of items depending on your individual purchase patterns.  And this is a spectacularly bad idea.

I’ve spoken in the media about the psychological consequences of dynamic pricing before, particularly in relation to Groupon and other deal sites.  The basic argument from stores is this: pricing in capitalism is built on a demand curve – the more you want it, the more you will pay – so by moving from understand the demands of the groups to the demands of the individual will allow for a closer fit to the true underlying demand.  It is a perfectly rational economic argument, and one that was used with Groupon: if your current demand is low, make it cheaper so that people sample your product, which increases their demand, which allows you to charge them the non-discounted price next time.

The problem is that it is a spectacularly bad psychological argument (which is why economists created behavioral economics in the first place).  Our willingness to pay for something depends not just on our absolute need for it, but also our understanding of it in a complex web of other prices and other motivations, including powerful anchors.  Which many merchants using Groupon’s found out: giving people something at a discounted price made them anchor their demand on that price and made it “not worth it” at a higher, non-discounted price.

The grocery plan suffers from this problem: you don’t gain nearly as many new loyal customers for an item by making it temporarily cheaper, especially if every other item in your category is also doing the same thing.  Indeed, let’s imagine that this became highly successful.  I don’t normally drink bottled water but I have bought Brand A juice, so Brand A offers it to me cheap ($1) to see if they can get me to buy it more regularly at a higher price ($1.50).

So I buy a case of discounted water from Brand A at $1.  Now Brand B, seeing that I’m a water buyer, offers me a discounted case at $1 to get me to switch to Brand B.  Brand A has gone back up to regular price ($1.50) and my setpoint for water is now $1, so I buy Brand B.  Which Brand C notices and…you get the idea.

But why isn’t this good for the consumer?  After all, I now get water for a constantly discounted price, if I play the game correctly.  And that’s the problem: I have to play the game.

The most volatile current pricing in a grocery store centers around produce and it is one of the things that people hate shopping for the most.  Indeed, stores sell much larger quantities of more stable-priced produce (like apples) than volatile-priced produce (like peaches) because of simple expectations.  I can go to the store and buy an apple with very few psychological ramifications: apples are mostly always the same price, I don’t feel like I’m getting swindled by my desire for it when out of season, and so I have very little regret.  But beware of falling in love with berries, which fluctuate wildly in price depending on season.  I’m happy, very happy, when they happen to be particularly cheap but otherwise tend to ignore them, because I fixate on the higher price: seasonal goods, in essence, have become premium goods.

So now apply that model to everything in the store.  If everything I love becomes more expensive the more I love it, suddenly I’m playing a vast and constantly shifting game of regret balancing.  I really want the cereal, but because I have bought it before and not this other cereal, the other cereal is now “discounted” to a significantly lower price and I constantly have to be making the tradeoff calculations of my want against this new and shifting price.  And what’s more, I know if I don’t buy the cereal I actually like for awhile, they’ll offer me a lower price at some point, so I should resist my urge to buy it now so that it comes down in price.

It doesn’t take a psychologist to figure out that is not a world anyone wants to live in.  Stability is a human component of human happiness, and while the occasional seasonal fruit might make me enjoy a particular day, the world is mostly powered by apples: price stable, quality stable, and thus the staple of the lunch bag.

"So your company is staffed by failures?": a conversation with a Big Company

You know a conversation is going to be rough when it is with someone in charge of cultural change at a Big Company.  Not because Big Company can’t change; I actually think this particular Big Company has done a good job of getting better.  But getting better and being good are a long way from each other, and Big Company Person (BCP) generally wants to tell you all about how great their culture is, not talk about how it could improve.

Imagine, if you will, BCP and me and another Startup Guy (SG).  BCP, of course, feels that we should be honored by Big Company’s presence at the event and wants to tell us how awesome they are.  SG and I both talk about what Big Company could do for startups, to which BCP replies that startups don’t really interest Big Company, because they will get them anyway “if they are successful”.

Fair enough.  I don’t agree, as I think the cream will rely on relationships they made when they were at their seed stage (which is why you see the top few VCs making most the money: everyone wants to work with them from the earliest stages), but it is certainly debatable.  And I’m trying to be nice, having suggested that Big Company is doing good things about their own culture and that after four years of meetings there, I finally saw some people in jeans earlier this week.  I follow up this anecdote with “so you don’t think Big Company can do stuff for startups, what can startups do for Big Company?”

BCP says, naturally, “culture” – that Big Company wants to have more spirited, entrepreneurial people working internally to accomplish big things.  SG points out how companies usually say that but then have cultures of bureaucracy that make it difficult.  BCP says “we’ve got that figured out.  After all, people at Big Company aren’t dumb.  They know how to do it.”

But wait.  If Big Company is full of smart people who already have that figured out, why do they want to get startups to teach them how to be innovative.  Something doesn’t add up.  Maybe, I suggest, startups could also be a feeder for talent. Which led to the following exchange:

BCP: “Oh, we have plenty of talent.  We get plenty of resumes.”
Me: “Getting a lot of resumes doesn’t mean you get a lot of good resumes, though.”
BCP: “We’re pretty good at wooing the people we want.”
Me: “But that’s Google’s advantage: they don’t have to woo me, I already WANT to work there.”
BCP: “But after your second startup fails, you get pretty hungry.”
Me: “So your company is staffed by failures?”

Luckily, because it was loud, BCP didn’t hear that (although SG laughed).  And not everyone whose startup fails twice is a bad employee.  But the odds of top innovators having some previous success is higher than not and I’m not sure anyone wants to be the Big Company that people go to in order to lick their wounds and recharge for their next try.

So to recap: Big Company wants to be talk to startups, but only so they can help Big Company and not because Big Company can help them.  They want to emulate startups and be entrepreneurial but are already really good at it and don’t want help.  And they have a huge talent pipeline, full of really smart people even though most proven entrepreneurs would never consider working there if they could get a job anywhere else.

So…why were they sponsoring this startup social hour again?  If you aren’t interested in helping us, and you don’t want our help, then I would officially like to thank you, oh Big-Company-that-shall-not-be-named, for the free food and booze.  Next time, also please include non-alcoholic drinks.

I actually take that back.  Next time, have no food and no booze and no Diet Coke and just come with an attitude that affords us some respect.  We may be young and unpolished, but if you watch us and listen occasionally, we might just have something that you can use to make both your Big Company and your world a bit better.  We understand bottom lines, know how to make profit, and if we aren’t the kind of people you’d be dying to woo, why are you talking to us anyway?