Ineffective entrepreneurship: post-mortem of Hippo, the happiness app that never quite was

60 word sum­mary: I spent two and half years try­ing to start a startup I thought might do lots of good. It failed. I ex­plain what hap­pened, how it went wrong and try to set out some rele­vant les­sons for oth­ers. Main les­son: be pre­pared for the fact you might find the ex­pe­rience mostly stress­ful and have noth­ing par­tic­u­lar­lly use­ful come of it.

Con­tents:

In­tro­duc­tion
What was Hippo
How did we get started? How did things fall apart?
What would I do differ­ently?
Po­ten­tial les­sons for as­pirant effec­tive en­trepreneurs
Conclusion

Introduction

It’s of­ten said EAs should be try­ing new things that prob­a­bly won’t work but will pay off dra­mat­i­cally if they do. If we’re not failing, then we’re not try­ing enough new stuff. One thing that’s been men­tioned quite a few times, al­though less so re­cently, is al­tru­is­tic en­trepreneur­ship: start­ing com­pa­nies as a high-risk, high-re­ward strat­egy: you might build some­thing that solves an im­por­tant prob­lem and ac­quire oo­dles of cash you can give away af­ter­wards. I tried this and it didn’t re­ally work out for me, so I thought I’d share the story. I’ve been mean­ing to write this up for a while – I called time on Hippo back in Oc­to­ber 2017 – and see­ing Joey Savoie’s rather op­ti­mistic post on the value of start­ing char­i­ties has prompted me to fi­nally add my own less op­ti­mistic data point about start­ing com­pa­nies.

What was Hippo?

Hippo is/​was a hap­piness tracker-train­ing app, which I used to try and de­scribe as a ‘FitBit for the mind’. The idea was you could track what you were do­ing, think­ing and feel­ing, get a bet­ter un­der­stand­ing of how your life was go­ing and use that to change how you live. The way it worked – or rather, was sup­posed to work – was that it would ran­domly ping you in the day, you’d en­ter your data, the app would dis­play your data and you could learn what was driv­ing your emo­tions. We never got as far as sel­l­ing it, but we were con­sid­er­ing our po­ten­tial mar­kets as (1) con­sumers, par­tic­u­larly the sort of peo­ple into mind­ful­ness and the quan­tified-self (2) an emo­tional sup­port/​em­ployee satis­fac­tion tool for com­pa­nies (3) a men­tal health mon­i­tor­ing/​self-care app paid for by health providers.

If you have an an­droid phone, you can still down­load it.

What was Hippo sup­posed to be?

I had grander plans than track­ing hap­piness. The goal was to build an app that would ‘solve’ hap­piness. The idea was to cre­ate some­thing that would mon­i­tor how you were feel­ing over time, let you un­der­stand that bet­ter, then analyse your data to give you the sup­port, guidance and tools you needed when you needed them. In prac­tice, this would mean com­bin­ing sub­jec­tive data (i.e. self re­ports) with ob­jec­tive data (lo­ca­tion, move­ment, etc.) to learn what was as­so­ci­ated with your mood scores, and then give you flex­ible, tai­lored guidance.

As an ex­am­ple of how the guidance might work, sup­pose you re­ported an un­usu­ally low hap­piness score. Hippo might then sug­gest you go for a walk, watch a TED talk, prac­tise mind­ful­ness, write down your great­est achieve­ment and how you achieved it, phone a friend, etc and would try to in­tel­li­gently offer the best op­tion. I’d com­piled a list of 40-odd ev­i­dence-based ac­tivi­ties and re-think­ing ex­er­cises, i.e. be­havi­oural and cog­ni­tive in­ter­ven­tions, that I’d come across dur­ing my re­search, and hoped we could use ma­chine learn­ing to give peo­ple ever more use­ful recom­men­da­tions over time. From speak­ing to hun­dreds of peo­ple, I’d found there was a huge gap be­tween the things that were shown to in­crease mood/​sup­port men­tal health and what peo­ple ei­ther knew about or did.

As I saw it, there was a neat and ob­vi­ous way tech­nol­ogy could help peo­ple live hap­pier lives that no one else was try­ing. It was per­haps, in tech jar­gon, a ‘moon­shot’: an am­bi­tious, un­likely-to-suc­ceed at­tempt to use new tech­nol­ogy to solve a huge prob­lem.

How did we get started? How did things fall apart?

TL;DR ver­sion:

I be­came sad. This gave me the idea for the app. I found some­one to help me build it. We worked on it part-time for 2 years. We never quite built any­thing that worked. Even­tu­ally ev­ery­one left.

Longer ver­sion:

I got the idea for Hippo back in Spring 2015. I was work­ing on an­other busi­ness idea, which also never got off the ground (a hap­piness-be­havi­oural sci­ence con­sul­tancy). I felt I was gen­er­ally failing at life when I sud­denly, un­ex­pect­edly dis­cov­ered I was very de­pressed. At the time I looked round for an app to help me—this was back in 2015, when apps were new and the sexy solu­tion to ev­ery­thing—and couldn’t find any­thing. I threw a bunch of things at my de­pres­sion—an­tide­pres­sants, CBT, mind­ful­ness, grat­i­tude jour­nal­ing, etc. - which were, in my case, re­ally suc­cess­ful. After I re­cov­ered I was left with the abid­ing frus­trat­ing that the tech­nol­ogy that would have saved me could ex­ist, it just didn’t ex­ist yet. Armed with my own ex­pe­riences and my util­i­tar­ian hat firmly on my head, I started think­ing se­ri­ously about how to cre­ate a hap­piness app which could, in the­ory, end up spread­ing around the world and chang­ing lives. It seemed like the most good I could be try­ing to do.

The early days (March to Septem­ber 2015) are now a bit too long ago to re­mem­ber. I started talk­ing to lots of peo­ple about it to gather feed­back, Lean Start-Up style. Through a friend’s in­tro­duc­tion, I spoke to a ma­jor high street bank. They said they were in­ter­ested in run­ning a trial; the idea was we’d us­ing anonymised mood track­ing on em­ploy­ees then (1) provide a ‘tem­per­a­ture check’ of em­ployee satis­fac­tion and (2) analyse the data to tell them what they should do to have hap­pier staff (and less ab­sen­teeism, churn, etc.). We were in!

Prob­lem: I didn’t have a product; it was just an idea. I told them I’d get back to them in a few months. I couldn’t code—I still can’t—so I con­sid­ered pay­ing some­one to have an early ver­sion cre­ated. For­tu­nately, I didn’t. That would have been a huge waste of money given I had no idea what I was do­ing. I stum­bled across an an­droid de­vel­oper, David, who was in­tro­duced to me through a friend. David agreed to start work­ing for free in ex­change for no­tional equity (‘sweat equity’), which was a huge coup. We started to try and de­sign the app over sum­mer 2015, which I found hor­ren­dously frus­trat­ing be­cause I have no eye for de­sign. But we were un­der way. Star­dom and riches beck­oned! I could prac­ti­cally see my face on Wired mag­a­z­ine.

In Septem­ber 2015 I moved to Oxford to start a moral philos­o­phy PhD. I was, and still am, re­search­ing ways to in­crease hap­piness, so there was some syn­ergy. I’d been given a place in March 2015, just as I had about pul­led my­self out of de­pres­sion. I didn’t find it a prob­lem to work on a start up and do PhD re­search si­mul­ta­neously. The work I was do­ing for Hippo – de­sign­ing the product and talk­ing to users – I was able to do much quicker than David was able to build the product (he had a nor­mal de­vel­oper job). I’d work out where I think we needed to go next, pass that on, and then get back to read­ing philos­o­phy, psy­chol­ogy and eco­nomics pa­pers. Switch­ing be­tween the two types of ac­tivity was hard be­cause the skill sets were so rad­i­cally differ­ent. The em­piri­cal work I did into hap­piness as part of the PhD did feed into the app, but noth­ing about start ups helped me write bet­ter philos­o­phy. At the time I thought I could eas­ily do both, but now I’m spend­ing all my time do­ing philos­o­phy I can re­ally see how much more effec­tive I am com­pared to when I was split­ting my con­cen­tra­tion be­tween two things.

At the turn of 2016 we launched the first, ex­tremely alpha ver­sion of the product to a few peo­ple. It wasn’t an in­stant suc­cess, but some peo­ple did use it and find it use­ful, which as­ton­ished me. I did a whole new set of user in­ter­views and my co-founder and I slowly worked on im­prov­ing the product. We iter­ated through this sev­eral times. I found the un­cer­tainty of try­ing to work out what was the best thing to do next ter­rify­ing. By the time we launched the alpha it was 9 months since we’d had the ex­pres­sion of in­ter­est from the bank. We went back to them apolo­get­i­cally tel­ling them “hey, we’ve built some­thing” but they never replied. Too slow, clearly.

Around March that year started talked to a com­puter sci­en­tist from my col­lege, who I’ll call ‘A’, about build­ing a ma­chine learn­ing al­gorithm. Ba­sic premise: you tell us how you’re feel­ing, we give you an ev­i­dence-based sug­ges­tions. You tell if it was good and it learns over time. Over the next cou­ple of months, I de­signed a spread­sheet with all the in­puts and out­puts and an ini­tial al­gorithm, which I handed it over to A who’d agreed to im­ple­ment it. He was pretty en­thu­si­as­tic but ba­si­cally noth­ing hap­pened for months and months. I’d get dis­tracted with my work, then some­times chase him and he’d always promise it would be just around the cor­ner. I didn’t re­ally have any bar­gain­ing power or other op­tions, and I be­lieved he would even­tu­ally pro­duce some­thing, so I chose to wait it out. In ret­ro­spect, there was no need to even try to use mach­ing learn­ing: we could have just ran­domly given peo­ple sug­ges­tions if they pressed a but­ton; the sug­ges­tions were only text any­way. Ah, Cap­tain Hind­sight, the smartest man I know!

Through­out 2016, and doubtless to the frus­tra­tion of my then-su­per­vi­sor, I’d go through bursts of work­ing on Hippo, then switch­ing back to philos­o­phy. The main prob­lem with Hippo as a hap­piness tracker was that, fun­da­men­tally, it wasn’t that use­ful to peo­ple. Sounds ob­vi­ous now, but took us ages to work out that was the prob­lem. Peo­ple seemed to like the idea, but didn’t find it use­ful and, where they said they found it use­ful, still didn’t stick with it. We ran into some key tech­ni­cal is­sues – it wouldn’t ran­domly send the no­tifi­ca­tions ask­ing you for your data - and never solved them, which meant Hippo never ac­tu­ally did what we wanted it to.

De­spite all this, and clearly in grip of self-delu­sion, I pitched Hippo to Y-com­bi­na­tor in Au­tumn 2016. It seemed like a big, crazy enough idea they might go for it. At the time, I thought Hippo had a higher ex­pected value than my aca­demic re­search (which wasn’t go­ing ter­ribly well, in part be­cause I was ne­glect­ing it to work on the app) and was con­tem­plat­ing the pro­cess of drop­ping out of uni­ver­sity if we found fund­ing to al­low David to work full-time. I thought the gap in the mar­ket(s) and the solu­tion was so ob­vi­ous I was con­vinced that, if I didn’t get there first, some­one else would.

A con­sis­tent prob­lem, with a sin­gle de­vel­oper work­ing in his spare time, was that we con­stantly strug­gled with the amount of de­vel­op­ment we could do. Late 2016 on­wards I in­creas­ingly fo­cused on try­ing find ad­di­tional de­vel­op­ers who would come and work for Hippo. This was a re­ally hard sell: we weren’t grow­ing, we didn’t have any money. Prob­a­bly the most psy­cholog­i­cally painful part for me was hav­ing to try and con­vince other peo­ple of my idea, an idea I was still strug­gling to ar­tic­u­late, to get them to join the pro­ject I was con­stantly frus­trated with my­self. Around the same point I started en­ter­ing a few pitch com­pe­ti­tions and go­ing to en­trepreneur­ship courses. I thought it would be use­ful to get feed­back on the product and maybe find some guidance on what I should be do­ing. As far as I could tell this was al­most en­tirely a waste of time. I did meet a lot of peo­ple who were slightly en­thu­si­as­tic about helping, but al­most no one who was gen­uinely use­ful, i.e. shared my dream and could tell me some­thing I didn’t already know. I was ap­proached by an an­gel in­vestor in early 2017; he’d heard about the pro­ject through some­one I vaguely knew. This made the whole en­ter­prise sud­denly po­ten­tially se­ri­ous: he was look­ing for a sort of HR-hap­piness-en­gage­ment product already but hadn’t de­vel­oped any­thing yet him­self. In the end, we couldn’t reach a deal. I wanted some in­vest­ment so that David could work full time, whereas what the in­vestor wanted was equity in Hippo in ex­change for (what seemed to me) an ex­tremely vague part­ner­ship ar­range­ment and, pos­si­bly, some in­vest­ment af­ter­wards. To quote a cur­rent line from Bri­tish poli­tics, it seemed that ‘no deal is bet­ter than a bad deal’ and I turned it down.

We’re now in Spring-Sum­mer 2017 and the fi­nal few months of Hippo’s life. A, the com­puter sci­en­tist, got con­scripted into the Egyp­tian army, never end­ing up build­ing the AI al­gorithm we’d been dis­cussing, and dis­ap­peared. This turned out to be good thing in one sense. With him gone, I de­cided I should man­u­ally test the smart-sug­ges­tion sys­tem: I got peo­ple to send me mes­sages on face­book, I asked them some stan­dard ques­tions and then sent them a sug­ges­tion for what they should do now. It ac­tu­ally worked pretty well and caused a brief burst of en­thu­si­asm. After a philos­o­phy talk I’d given, an­other de­vel­oper, call him ‘V’, offered to work on the app and our for­tunes sud­denly looked up. But only briefly. We went our sep­a­rate ways for sum­mer, and when we came back V had found a job in San Fran­cisco and was no longer in­ter­ested. Around the same pe­riod, I saw CEA ad­ver­tis­ing a part-time role as Peter Singer’s re­search as­sis­tant. He was plan­ning to write a book on over­pop­u­la­tion, a topic I’d ac­ci­den­tally ended up work­ing on in my the­sis. I ap­plied, was offered and glee­fully ac­cepted the role (I know, odd twist for a start-up story...). I’d also done about 2 years of my PhD and it looked like finish­ing was both in sight and worth­while. This se­ries of events were sig­nifi­cant be­cause, for the first time (in my eyes) it seemed I would have a higher ex­pected value by stick­ing in philos­o­phy than con­tin­u­ing with Hippo. V’s de­par­ture or, rather, non-ar­rival, kil­led my en­thu­si­asm for Hippo be­cause I knew I wouldn’t be able to sum­mon the en­ergy to try and con­vince some­one else to join David and my­self. By con­trast, hav­ing had a pretty shaky first two years of the PhD, get­ting the RA job made me much more con­fi­dent I might find a po­si­tion in aca­demic philos­o­phy af­ter­wards.

I would, ac­tu­ally, have kept work­ing on Hippo, even de­spite this. We were rea­son­ably close to im­ple­ment­ing the hap­piness-sug­ges­tions fea­ture (‘in­stacheer’), at which point the app might have started be­ing use­ful. David, my co-founder, end­ing up get­ting a differ­ent job and los­ing in­ter­est for his own rea­sons. Dis­ap­point­ingly, I never got to test my un­der­ly­ing hy­poth­e­sis, namely: if we cre­ate a mood­track­ing app and use that info to give peo­ple peo­ple nudges on what to do, it will be use­ful enough that peo­ple and com­pa­nies will pay for it.

And there I was, Oc­to­ber 2017, two and half years later, the last one stand­ing, and de­cided it was time to give up. So it ended. Not with a bang, but with a whim­per.

What would I do differ­ently?

As I see it, our failure was that we never built the thing we were try­ing to build. It’s not that we built a rock­et­ship, launched the rock­et­ship and saw the rock­et­ship ac­ci­den­tally ex­plode into a mil­lion pieces. Rather, we had a blueprint for a rock­et­ship and there are still a mil­lion pieces for it sit­ting on the launch­pad await­ing as­sem­bly when we walked away.

Why did we never get there? I think my par­tic­u­lar mis­take was prob­a­bly be­ing too op­ti­mistic about how much de­vel­op­ment work David and A would and could do. I should have been more crit­i­cal of our de­vel­op­ment pro­duc­tivity and not given them so much of the benefit of the doubt. That might sound in­dul­gent (“my main flaw is that I be­lieve in peo­ple”), but I’ve come to re­al­ise this is a gen­uine flaw of mine: you can be too dis­trust­ful and you can be too trust­ing. I am the lat­ter. I don’t think I was a very good man­ager.

To ex­plain, if I’d re­al­ised H, the guy who agreed to build the ma­chine learn­ing al­gorithm be­hind the smart-sug­ges­tions fea­ture, wasn’t re­ally do­ing any­thing, I would have tried much harder to find some­one to re­place him. Or, I would have done what I even­tu­ally did, which was test the pro­to­type fea­ture man­u­ally. Really, I should have run that test months ear­lier than I did. Equally, if I’d re­al­ised how slowly David would be able to make progress—one guy work­ing part-time some­times—I’d have done one or more of the fol­low­ing: (a) tried harder find an ad­di­tional coder; (b) learnt to code my­self—which I might have done if I’d not found David; (c) looked harder for in­vest­ment; (d) given up ear­lier. As it was, we sort of plod­ded along, mak­ing enough progress that it seemed bet­ter to con­tinue the cur­rent course. If we’d gone much slower we/​I would have changed course. If we’d had the de­vel­op­ment re­sources to pro­duce our product, we’d at least have tested the product we had in mind. We prob­a­bly have failed later too, but that would be for a differ­ent rea­son re­quiring differ­ent ex­pla­na­tion. A ma­jor prob­lem was that, as a non-tech­ni­cal founder, I couldn’t eval­u­ate how much was hap­pen­ing. Of course, if we’d been build­ing a build­ing, it would have been ob­vi­ous...

Say­ing the prob­lem was mi­s­un­der­stand­ing how slowly de­vel­op­ment was oc­cur­ing might also sound in­dolent (“why didn’t you just build it your­self?“). Hard­core techies will prob­a­bly be think­ing I should have just dropped the PhD. I thought about that a lot, but as no point did that seem like the right choice. My ob­jec­tive was and is to in­crease wor­lwide hap­piness. Start a start up seemed like a means to achiev­ing that end. But so did my PhD re­search, which is on the topic ‘how do we max­imise world hap­piness?’ Given I had a tech­ni­cal co-founder but no pro­gram­ming skill my­self, it never seemed like my com­par­a­tive ad­van­tage would be to drop philos­o­phy re­search in or­der to be­come an novice coder.

Po­ten­tial les­sons for as­pirant al­tru­is­tic entrepreneurs

This is a bit tricky, be­cause I don’t know how much I learnt that is also use­ful. My main re­flec­tion on Hippo was that I felt that I spent a huge amount of time to achieve not very much. I would con­sider that the cau­tion­ary tale. Here’s my list:

Star­tups do not nec­es­sar­ily teach you use­ful skills, par­tic­u­larly if your al­ter­na­tive op­tion is to work in a differ­ent field. My ex­pec­ta­tion was that, what­ever hap­pened, I would learn loads to take to what I did next. As an as­pirant philoso­pher, al­most noth­ing I learnt from about en­trepreneur­ship is di­rectly ap­pli­ca­ble.

Failing un­re­mark­ably can hap­pen and is some­thing you should con­sider as a pos­si­bil­ity. If I were to claim I had built ca­reer cap­i­tal as a re­sult of my 2.5 (mis)ad­ven­ture, I would be de­ceiv­ing my­self. As it is I walk away with no cool story, no CV points. At least if I’d blown some money (mine or some­one else’s) and failed spec­tac­u­larly, then I would have a story and I’d have felt that I’d re­ally tried and failed.

You will likely learn things about your­self you didn’t ex­pect to learn. In my case, I think this was (a) I put too much trust that peo­ple will do the work they say they will(/​I have no idea how to man­age peo­ple); (b) I don’t like work­ing un­der great un­cer­tainty; (c) I can­not stand try­ing to sell or ‘pitch’ com­mer­cial ideas (pitch­ing philo­soph­i­cal ideas are fine); (d) I’m not very good at prac­ti­cal de­tails (con­cep­tual de­tails are fine); (e) prob­a­bly, most use­fully of all, I im­proved my abil­ity to be self-crit­i­cal. Start up method­ol­ogy (e.g. Lean Start-up, the Mom test) teaches you how to dis­cover what peo­ple re­ally think when they’re be­ing too nice to tell you your product sucks. (e) does have psy­cholog­i­cal down­sides (see be­low).

It’s not enough to try to build some­thing you think the world needs. I started Hippo be­cause I thought it might do an enor­mous amount of good. Yet, to make a suc­cess­ful product, you need to build some­thing which is use­ful to peo­ple, not just some­thing you think would to be use­ful to them. I po­ten­tial dan­ger with an EA start-up is that you might, as I did, find a hy­po­thet­i­cal solu­tion and try to retro-fit a busi­ness case.

If you try to start some­thing, loads of peo­ple try to help you; un­for­tu­nately, many of them will be use­less. I was hon­estly im­pressed at the fact peo­ple (well, mostly 1 other per­son) were will­ing to spend their time work­ing on my crazy pro­ject and do so for free. Loads of peo­ple offered to help out: they gave ad­vice, were will­ing to let me in­ter­view them to talk about the app, etc. Ad­mit­tedly, lots of the help isn’t very use­ful, and a rele­vant challenge is to sift through all the differ­ent per­spec­tives peo­ple give you and make your own de­ci­sion.

The un­cer­tainty can be ter­rify­ing. The con­stant ex­pe­rience of not know­ing what to do in start-up life was, for me, anx­iety-in­duc­ing and some­thing that didn’t seem to get much bet­ter over time (if only there was an app to help peo­ple with their moods, oh wait…).

While failing is very com­mon, it can still be psy­cholog­i­cally very hard. In the­ory, I should feel glad that I ded­i­cated a con­sid­er­able amount of time to some­thing I thought would have the high­est ex­pected value. As it hap­pens, my abid­ing psy­cholog­i­cal re­sponse, when I think about Hippo, and which is prob­a­bly ir­ra­tional, is one of frus­tra­tion and shame with only a glim­mer of pride. Frus­tra­tion that we never built what I wanted to test. Shame that I feel, in some sense, I wasted my time and should have known bet­ter. A glim­mer of pride that I, at least, stuck with it for so long. The tragedy, as I feel it, is that I never found out if my hy­poth­e­sis about an ideal piece of hap­piness-in­duc­ing tech was cor­rect. I haven’t seen any­one else build it (yet) and I feel suffi­ciently burnt out by my ex­pe­rience I lack the mo­ti­va­tion to try again.

Conclusion

Although I billed this a some­thing of cau­tion­ary tale of failure, to be clear, I’m not sug­gest­ing EAs shouldn’t try to start start-ups. If you have a good idea, you should still go for it. In fact, I’m glad to see there are three other men­tal health apps be­ing cre­ated by EAs—Mind Ease (Peter Bri­et­bart, stress and anx­iety), Inuka (Robin van Dalen, sup­port chat app) and UpLift.us (Ed­die Liu, de­pres­sion) - and I wish them the best of luck. I’m also not sug­gest­ing I didn’t learn any­thing at all. Rather, be­cause sto­ries of start-up failures are so of­ten ac­com­panied be tales of later start-up glory and vi­tal life-les­sons learnt, I wanted to add anec­data: it’s pos­si­ble to fail in a deeply un­in­ter­est­ing way and not learn much. Fur­ther, it’s pos­si­ble there’s some­thing par­tic­u­larly gal­ling about failing to start a start-up where your aim is to help other peo­ple and do good, rather than just make money, but I don’t know any other failed asipirant effec­tive en­trepreneurs to ask them.

I hope my con­tri­bu­tion here has been (1) to provide an ex­am­ple of some­one who did try to start a com­pany with the in­ten­tion of do­ing the most good and failed and (2) to share what hap­pened and what I did (or didn’t) learn along the way, and how I felt about it.

If any­one would like to chat to me about this pri­vately, you can find me on michael.plant@stx.ox.ac.uk