0:00

Welcome to Rework, the podcast by 37signals about the

0:02

better way to work and run your business. I'm

0:05

your host, Kimberly Rhodes, and I'm

0:07

joined by the co-founders of 37signals, Jason

0:09

Freed and David Heinemeyer-Hanson. If you've listened

0:12

to the podcast at all, you've heard

0:14

Jason and David talk about trusting your

0:16

gut and trusting your gut in

0:18

business. This week, we're going to talk about

0:21

that. Jason, you recently wrote about this on

0:23

your Hey World in a post called, It's

0:25

All a Judgment Call. Tell us about

0:27

that. Yeah, I was having

0:29

a discussion with someone about data driven

0:31

decisions and whatever. And it

0:33

just like it kind of dawned on me that like, okay,

0:35

so you've got some data. Is that the

0:38

only thing you're making this decision based on? Really,

0:40

truly? This the data? That's it? Well,

0:42

then why not just have a computer do it? And

0:44

then person's like, well, because it's

0:46

like exactly, because you

0:49

have experience, you have judgment, you've seen other

0:51

things, there's a million points of view that

0:53

you have, you don't even know where they

0:55

came from, you don't know why you were

0:57

influenced to make these decisions. But you're making

0:59

a human decision. A human decision is a

1:01

judgment call. Every single time. It's a judgment

1:03

call. You're weighing all sorts of things. You

1:05

don't even know what you're weighing. But you're

1:07

weighing millions of different data points, and not

1:09

just the data that's in front of you. And

1:12

I started thinking about like, well, why do companies hire

1:14

when they hire executives? Who do they

1:16

hire? Do they hire people who are really good at

1:18

Excel? Do they,

1:20

you know, no, they hire people

1:22

who have good judgment and good

1:24

experience, they're looking for experience. What

1:27

is that? Experience is a very

1:29

fuzzy, blurry thing. It's not cut and dry. It's

1:31

like, I've seen this, and I've seen that, and

1:33

I've been around this, and I was around this

1:35

when that happened, and I saw how this came

1:37

together. These are all things that color your point

1:39

of view. And then you bring that point of

1:41

view to the next thing, and you paint with

1:43

that. That's what people actually want. So

1:45

I just feel like it's time to be honest about that.

1:48

Everyone's trying to find like certainty in scientific

1:50

decision making and whatnot. And there's like some

1:53

data that's obviously more correct than

1:55

others, but that's not the only thing that

1:57

makes A decision. And If it does, might as well

1:59

be a machine. Okay so

2:01

a lot of got decision To me

2:03

I think intuition and harness saying what

2:05

you know to be true. How do

2:08

you guys kind of take that step

2:10

of trusting your intuition and trusting your

2:12

that. All. The time. As.

2:15

Part of it is that it's a privilege to be able

2:17

to do so. And that a

2:19

lot of the. Object

2:21

to pretty soon. The lot

2:23

of the rationalization comes when

2:25

people are trying to impress.

2:28

Other people. With. Their.

2:30

Thought. Process if you. Well, it's not

2:33

even the thought process because it was just

2:35

that I would be more benign. but it's

2:37

like trying to rationalize, trying to break proof.

2:39

Oh, I made this decision because. A

2:42

B C A D Yes, you can

2:44

only percent be no factors of that

2:46

decision to other people. What about all

2:48

the other stuff? the just lives in

2:51

your gut? I meet. The reason

2:53

we call it god is that it's literally not

2:55

your head. That. There is a

2:57

computer inside of your stomach. The job

2:59

even really know how operates. It's a

3:01

little bit of a black box. you

3:03

feed it some problems and will spit

3:05

out and intuition Ill spit out a

3:07

direction that go like you. What? I

3:09

think we should go over here. I

3:11

have us. I have a hunch I

3:14

have a sniff that this is the

3:16

path forward. I can't explain what that

3:18

is. It's funny my oldest kids with

3:20

talk about how they do math and

3:22

a lot of how you do math

3:24

is you have to show you. What

3:26

to your throat show? Need seven? You

3:28

go through it and like that. Makes

3:31

sense when it's like a math problem.

3:33

The variables are know they're fixed, the

3:35

values it. Nothing. Almost

3:37

nothing in the realm of business has

3:40

that privilege that you get to hold

3:42

all the variables constant. Just tinker with

3:44

a few know you have to. Deal.

3:47

With the fluidity of the situation as people's

3:49

expects noises use just me that there is

3:52

no fixed point. The kids stand in one

3:54

position and go like I use my five

3:56

inputs. and here's my to outputs and that's

3:58

what it is. you know. The even know.

4:01

What the magic to me is your have

4:03

to know. If that

4:06

got computer you have. Has

4:08

been trained are not not really pushing

4:10

the metaphor, your steady diet. Of.

4:12

Bomb. Experiences.

4:15

And different tastes and different. Also the

4:17

I've been him a lot of different

4:19

places. You can learn to trust it.

4:21

I think that's what we've. Essentially.

4:25

Allowed ourselves to do is to trust

4:27

that that black box is not going

4:29

to reveal all his magic to was

4:31

in terms of how reaches this but

4:33

it's gonna reach good decisions and this

4:36

is where I see the clearest can't

4:38

trust. Two people were very good as

4:40

a tickle aiding what they know under

4:42

inputs. There's on million Mbs who could

4:44

tell you exactly how to run through

4:46

this form. Last, who could do the

4:48

Boston Consulting grades. Oh

4:51

you gotta to Disney got do this and they

4:53

suck. They. Fucking suck. They can

4:55

not make a good decision that leads

4:58

are great outcome to save their damn

5:00

life. Or. Their lives or companies

5:02

Which is why so many companies who

5:04

end up with professional managers go down.

5:06

damn to they go down the drain.

5:08

They can explain all these you supposed

5:10

to do a none of it fucking

5:12

wants because they don't have it tuned

5:14

black box in their guts to guided

5:16

in the right direction and as the

5:19

same time you can see entrepreneurs and

5:21

founders and and so forth be unable

5:23

to sick like why do I do

5:25

think that I do? I don't know.

5:27

I. Just have all these are employed. By.

5:30

Make good decisions and that's ultimately what

5:32

does this is why businesses such as

5:34

satisfying to made to be is because

5:36

you get. Verdict: Back on

5:38

the market. Or you make good

5:41

decisions. I can't. Articulate it.

5:43

I can't even predict it. But. You

5:45

can trended sort of backwards. You can

5:47

have a moving average. Their entrepreneurs are

5:49

when normal business partners, Just entrepreneurs, but

5:52

business people in general. Who.

5:56

If. You have a conversation with them through do

5:58

not necessarily the hundred and forty. Q

6:00

Super with is right. In fact,

6:02

I've seen some studies show is

6:04

always a detriment to you if

6:06

you are too highly attuned to

6:08

the intellectual rigors of whatever you

6:11

lose sight of all those intangibles

6:13

arms and you're actually a worse

6:15

business person. So I find that

6:17

dumb learning to trust your gut.

6:19

His. First of all the realistic thing and

6:21

business Because we have no alternative. We don't

6:24

get to pick the barrels, We don't get

6:26

to live in the Harvard Business Case study.

6:28

To me what we can just look back

6:30

was not we have to jump into the

6:32

unknown. With. The only. Real

6:35

compass that we have with us

6:37

and we have to trust that.

6:39

and the sooner you get to.

6:42

Reality. Of that. The. More bullshit

6:44

you can cut out. To. Sets really

6:46

what it is when it comes to

6:49

this to the bottom line is so

6:51

much bullshit in rationalises in projections in

6:53

Excel, spreadsheets and all the stuff that

6:55

goes into this decision making process. And

6:57

to get concrete with this, we had

7:00

an example at There's Hims Eagles for

7:02

a while we're trying to say new

7:04

prices for the product and were like

7:06

this it's time to experiment and patsy

7:08

prices for long. Timeless must rise of

7:11

the and we try the heavily rationalization

7:13

and. I'm partial to the

7:15

stuff. I love statistical significance. I love the

7:17

rigor a bit, and I could also look

7:19

back upon that. This is making process go

7:22

like you know what was gone with Jason

7:24

Scott. We would have done it fucking fifteen

7:26

minutes and instead it took three months to

7:28

ride at the same goddamn outcome. For

7:31

the Us I want to add by the

7:33

way is is a x Have gotten a

7:35

debate with people bought this I'm often facts

7:37

of the like. We did all the market

7:40

research we asked or of we've we've we

7:42

surveyed five hundred and a customer's we do.

7:44

These surveys is extensive surveys with eighty questions

7:46

We have all this data are you telling

7:48

me we shouldn't follow it and I would

7:50

say is what were they were The questions

7:52

come from. So. You you ask the

7:55

questions. Why do you decide those questions? And.

7:57

then they go was because don't

8:00

the ones, it's like, well, what

8:02

if I asked 80 different questions? Like,

8:04

why those 80? It's

8:06

all a judgment call. You thought that those

8:08

80 questions were the important questions to ask

8:11

the customer. I might think there are different

8:13

80 questions to ask the customer or only

8:15

30 or only six that mattered. These

8:17

are judgment calls. So, if you

8:19

just strip back a few layers, it all comes

8:21

down to, I don't really know,

8:23

we're making our best guess here. We're bringing

8:26

our judgment to bear and we

8:29

think that these are the

8:31

questions we should be asking. The data you

8:33

get, you might feel is certain but it's

8:35

still based on something you thought to

8:38

ask. Where'd that thought come

8:40

from? Experience? What's that? Like,

8:42

so you just have to go back a

8:45

few layers here and you realize like, it's

8:47

all very, very soft. It's not

8:49

concrete. It's very soft, it's very blurry,

8:51

it's very muddy, it's very full of

8:54

unknowns and these things

8:56

are not based on solid foundations

8:59

even if you have the answers that

9:02

you think are. The questions themselves came

9:04

from a really muddy, blurry, you

9:07

know, obscure place. So, anyway, that's

9:10

how I've approached this and that's kind of what the what the

9:12

erogant was about. And

9:14

I think the fundamental misunderstanding here is that

9:16

there are people who think all this rigor

9:19

is valid or applicable because what

9:21

they're doing is actually like hard science. We're

9:23

being very scientific about this stuff. It's almost

9:26

like we're doing physics or we're doing chemistry

9:28

or we're doing God forbid math.

9:30

No, you're not. You're doing fucking sociology,

9:33

psychology. Do you know what's been going on in

9:35

the social sciences for the past 50 years? A

9:38

lot of fucking bullshit. Half of it,

9:40

just off the top, half of it, all

9:43

the famous studies you've heard about it, they don't

9:45

replicate. No one knows anything.

9:47

No one can repeat any of the fundamental

9:49

insights that we're supposed to have learned from

9:52

the social studies for so long. That's what

9:54

business is, is social studies. And no one

9:56

can agree on what the terms are. Everyone

9:58

dresses the same way. stuff up

10:00

in statistical rigor and it's

10:03

p-hagging out the wazoo, it's

10:05

picking your conclusions after the

10:07

fact, it's all this post-rationalization.

10:10

I want this certain outcome and I'm going

10:12

to backtrace my way into the steps I

10:14

need to get to together. Just

10:17

give up all the steps. Just go

10:19

straight to your preferred outcome. Test

10:21

it against reality. Conclude

10:24

that one experience or experiment within

10:26

this context. At a verdict, do

10:28

you know what? I can extract 10% rough

10:31

guidance from that. I can't extract life

10:33

lessons that they're going to apply universally, even

10:36

in my own damn business. This is one

10:38

of the reasons I enjoy testing our priors

10:40

so much. This pricing

10:42

experiment I referred to was one of those examples.

10:45

We tested pricing many times in the past.

10:47

You know what? A pricing test from 2013

10:50

is not worth a whole damn lot. It'll

10:53

tell you something in some extract ways or

10:55

at best prompt you to put some good

10:57

questions to the market. But you can absolutely

10:59

not use data that's 10 years out of

11:01

date because it's just out of date. All

11:04

the variables change. All the circumstances change.

11:06

All the expectations of customers change. Accept

11:10

the fact that you are in the social

11:12

sciences and that the social sciences are mostly

11:14

bullshit most of the time. Okay,

11:16

so David, you mentioned pricing when it comes

11:19

to Basecamp, which makes me think of this

11:21

new once umbrella and those kind of pricing

11:23

decisions, which there's no history to go back

11:25

to like there is with Basecamp. How do

11:28

you make that kind of judgment call when

11:30

it comes to something like that for

11:32

either of you? Pricing, you mean? Yeah.

11:36

You pick a price that feels reasonable. It

11:38

takes three seconds and you're done. You

11:41

put it out in the market, you see what happens. Hey,

11:44

we're selling enough. Great. We're

11:46

not selling as many as we thought. Could it be pricing?

11:48

Could it be something else? We could try this then. The

11:50

whole point is like if you want an answer, you got

11:52

to do the thing, put it out in the market

11:54

and that's where you get your answers. You don't get

11:56

your answers from research in this case. You don't get

11:58

your answers from surveying. your answers from asking

12:00

people, what would you be willing to pay? Because

12:03

it doesn't cost anyone to give you an answer. The

12:06

only answers that matter are the ones that

12:08

actually cost the answer. So if

12:10

you put it out in the market for $2.99,

12:12

if people buy it for $2.99, the

12:15

answer is yes. If you

12:17

say, would you pay $2.99 and they say yes,

12:19

well, it costs them zero. So it's not an answer.

12:23

It's that, that's how you do it. I

12:25

was just on a call with a couple of

12:27

entrepreneurs running a business where we're talking about exactly

12:29

this pricing stuff. And they were like, oh, we're

12:31

trying to find the right advisors, right

12:33

advisors who can tell us what to do. And I'm like, no,

12:36

no, no, no, no, absolutely not.

12:39

Do not take anyone's advice.

12:42

Where did they pull that advice from? From their

12:44

own circumstances in their own business that has a

12:46

million variables that do not transfer to your context.

12:48

The only way to get it is just test

12:50

and start tomorrow. They've been like, oh, we've been

12:52

talking about pricing for a few months. What do

12:55

you know? No, start testing

12:57

tomorrow. You get new customers,

12:59

especially if you're dealing with web based stuff.

13:02

I mean, we have new customers coming to

13:04

basecamp.com every single day, thousands

13:07

of them who are fresh. They

13:10

have heard of basecamp, they've got a

13:12

recommendation, whatever. They're going to see our

13:14

offer more or less blind for the

13:16

first time. We can change that

13:18

offer from day to day. Now, there's some secondary

13:20

effects and dah, dah, dah, dah. But fundamentally, you

13:22

should look at these pricing tests as they're isolated

13:25

experiments. They are actually

13:27

relatively replicable within your

13:30

own context within a narrow

13:32

timeframe. So you can test one thing to one

13:34

day and another thing the next day. And you

13:36

can be reasonably confident that like, that's going to

13:38

give you some answers to go on. But that's about

13:40

it. That's what transfers from

13:43

one context to the other. And

13:45

this is why advisors and other

13:47

inputs into this is just mostly

13:50

useless. Start with your

13:52

own intuition. That's what's going to give you

13:54

the foundation for the experiment and then run

13:56

the damn experiment. The thing is that people

13:59

really are surprised. when I tell them

14:01

this, like literally pricing

14:03

decisions, like we just make it up. Like

14:07

there was not a meeting about this,

14:09

there was not a two-hour conversation about

14:11

this. We didn't put 17 different options

14:13

on the table and debate each one.

14:16

We're like, I don't know, $2.99 or

14:18

I don't know, maybe it was even like $4.99? I don't know, how about $2.99?

14:21

Yeah, that sounds pretty good. Let's try that. I

14:24

mean, literally, honestly, it's like a minute

14:26

to three minutes. That's

14:29

it. Most decisions should probably be like that, frankly. I

14:31

mean, like the idea that you have to sit there

14:34

and debate something that you'll never know the answer to

14:36

until you do the thing anyway, just do the thing.

14:39

Use your best, again, your best judgment, whatever that and

14:41

people, what is that? I don't know what it is.

14:43

It's your best judgment. That's what it is. Make a

14:45

decision, have a feel, feel it out,

14:47

go for it, test it on the market, see what happens.

14:50

That's your answer. There's no other way to get there. And

14:54

this comes to the point

14:56

of leaning on that gut computer. We've

14:59

been training your gut computer your entire life. But

15:03

even within a given domain, Jason and I

15:05

have been training our product computers for 25 years.

15:09

Whether we talk about this for five

15:11

minutes or 10 minutes or two days or

15:14

two weeks, it's such a vanishingly small

15:16

amount of time compared to 25 years we've

15:18

been training the gut computer. Now, it's

15:20

going to give us an answer. And

15:22

sometimes the answer is 42 and you don't

15:24

know what it means, but you've got

15:26

to go with it anyway and you just

15:28

go like, all right, we'll start there.

15:30

That's the jumping off point for the experimentation,

15:33

which is, I think, as Jason says, it's

15:35

shocking when people hear about it in the

15:38

context of pricing, but it's actually shocking in

15:40

the context of everything. Jason

15:42

and I very

15:44

rarely deliberate for long periods

15:46

of time in real time or

15:49

even offline time. Like

15:51

if you compress most of the major decisions

15:54

we've ever made, you

15:56

could compress them to probably 10 minutes, maybe

15:58

30 minutes if it's a... truly one-way

16:00

door, as Jeff Bezos would say,

16:03

there's no coming back from it.

16:05

Maybe we take a combined like

16:07

two hours. That two hours might have been spread out

16:09

over two weeks because we wanted to like see how

16:11

it feels tomorrow. This is actually

16:14

a good input to this whole thing about

16:16

the black box computer. What we often do

16:18

is we will reach a decision

16:20

or a preliminary decision in about five minutes

16:23

and then we'll go like, let's sleep on it. Do you

16:25

know what that means? Let's sleep

16:27

on it means feed it to the

16:29

gut computer. Let the gut

16:31

computer run its calculations. Digest

16:34

it. Digest it literally. Yes,

16:36

digest it literally. And in

16:38

the morning, it'll deliver an answer. It'll

16:41

deliver an answer that you can feel like you have

16:43

some confidence about because I think this is the other

16:45

part. When you make a decision quickly, a lot

16:48

of people feel like they don't have

16:50

confidence in that decision because it's very

16:52

monumental. Perhaps most aren't really. People just

16:54

pretend. But they believe it's

16:56

monumental. And therefore, they believe

16:58

it requires more input. Again,

17:00

all the input is already in the

17:03

computer. So if you

17:05

need to, and sometimes even we do, it is

17:07

a big decision. We just go like sleep on

17:09

it. Sleep on it is letting the

17:11

computer work and it will deliver the answer. And

17:14

then don't fuck around with the answer anymore. Once

17:16

you know, you know, this is the other thing

17:18

that's so curious. And I found this again many

17:20

times when Jason and I talk. We'll

17:22

show up to a discussion. We already know what the

17:24

fucking answer is. The

17:26

discussion is about getting onto terms

17:29

with the answer that is already

17:31

known. This is relevant

17:33

in pricing. It's relevant in employees.

17:35

It's relevant in thousand evaluations.

17:38

We show up already knowing

17:40

what the answer is because the computer has computed

17:42

what we have to do. We have to do

17:44

these little dances to get comfortable with what we

17:46

know to be true already. One other thing I

17:48

want to add, Kimberly, really quick about pricing. Someone

17:52

might say, who's listening to this, they might say, well,

17:54

like, if you don't test and keep testing, you might

17:56

be leaving some money on the table. Maybe

17:58

you could have charged $349. instead

18:01

of $299 and the answer is you're right,

18:03

maybe we could have and maybe

18:05

you'll eventually find that better price

18:07

or maybe we won't. But we're

18:09

comfortable not trying to be perfectly

18:11

optimized. Like that's not,

18:13

we're not after certainty. We're not trying

18:15

to find that perfect, perfect point. We're

18:18

trying to find a big enough circle that's good

18:20

enough for us. Like it's

18:22

not about the point. It's not about exactness.

18:25

It's about this is pretty good. Yeah, maybe

18:27

we're leaving somebody on the table. Whatever. Like

18:30

it's fine. Especially if it

18:32

would have taken nine months to find that point. Don't

18:34

really care. Not that important. So anyway, that's another

18:36

way to look at this too. This

18:39

is one of the pure

18:41

privileges of running a company

18:43

where you don't answer to anyone. Because

18:46

the problem with having investors and a

18:48

board or executives, whatever, other people that

18:50

you have to convince that your gut

18:52

computer was correct is that that

18:54

requires the level of exactness and you can't say

18:57

things like, eh, we're just gonna leave some money

18:59

on the table. Like, hey, what

19:01

do you mean leave some money on the table? That's my money. I'm

19:04

an investor here. I don't give a shit about your

19:06

black box computer. I want everything now. Right?

19:09

Everything I can get out of this thing, you should be getting

19:11

all of it. This is the

19:13

principal agent problem. We

19:15

don't have a lot of principal agent

19:17

problems at 30-7 Signals because we are

19:20

principals and agents at the same time.

19:22

So we are allowed by ourselves to

19:25

like play it cool. And you know what? Things

19:28

come easy when you take it easy. Okay,

19:30

last question before we wrap it up. That sounds like a t-shirt.

19:32

We should print that. We're gonna

19:34

embroider that on a pillow. How often

19:37

do you guys sleep on something and

19:39

come back to the original thought?

19:42

Most of the time. Are you typically? Yeah,

19:44

you're typically like right the first time and

19:46

you just need that extra time. You're not

19:48

typically changing the position. The

19:51

sleeping on it basically either confirms

19:53

or like radically denies. So

19:55

it's either like, man, that was actually, you know, I slept

19:57

on that. I don't know. like

20:00

I thought that it were 6% off

20:02

the original guess. It's like yes or

20:04

no. That's what the sleep is. That's

20:06

the sleep is for. And

20:09

again, it comes back to like, it's

20:11

usually probably a yes sometimes it's a no. If it's if

20:13

it's a yes, it's like just just try it. Even if

20:15

it's a no, it's kind of like well, I

20:18

don't I don't know. The no could be we don't know.

20:21

But let's get it easy to try. I think there's nothing

20:23

if it's very easy. This is the one way two way

20:25

thing the basal thing like not really quite

20:27

the same but similar like if it's

20:29

easy to try, let's just try it like this pricing thing

20:31

since we're getting back to this. $2.99 like, I

20:34

don't know. I'm kind of feeling a little

20:36

weird about it. I think we could charge more. Well, maybe

20:38

we could but why don't we start $2.99. Okay, it's an

20:40

easy okay. Then we can

20:42

do then we can do it. We can move.

20:45

Like you can't if you're sitting in deliberation,

20:47

you're not moving, you're treading. And that's okay

20:49

here and there, whatever but we don't do

20:51

it much but it's okay but you just

20:53

can't sit there all the time. So

20:56

I think it's better you're better off just like saying

20:59

yes or no and versus like, yes,

21:01

times 1.3 or something like that. No, just

21:04

it's yes or no. I

21:07

think this is why we have such

21:09

forward momentum because neither Jason or I

21:11

need our gut computer to win

21:13

every deliberation because so much of it just

21:15

gets put to reality and reality will tell

21:17

us whether it's correct or not. So sometimes

21:19

Jason will have an intuition that goes like,

21:22

Oh, I think we should do this and

21:24

I go like, my gut computer actually comes

21:26

with the opposite answer and says like, Yeah,

21:28

I think that's right. Sometimes I'll

21:30

raise the concern other times I'd be like,

21:32

Who cares. So we try it either

21:35

works and I was wrong. Hooray. Or

21:37

it doesn't work and I was right. And

21:40

also hooray because we just tried the next

21:42

thing. There's just there's such a premium that

21:44

we try to place on momentum. So

21:47

forward, try things, let

21:49

reality be the judge. Why are we

21:51

wasting our time arguing about something that

21:53

reality could tell us if we just

21:56

put it to the test? I

21:58

also hear you guys say like, the worst that can happen.

22:01

I feel like you guys use that sometimes in your decision

22:03

making too. Yeah, exactly. Okay,

22:07

well, with that, we're going to wrap it up.

22:09

ReWork is a production of 37 Signals. You

22:12

can find show notes and transcripts on our website

22:14

at 37signals.com/podcast. Full

22:16

video episodes are on YouTube and Twitter. And if you

22:19

have a specific question for Jason or David about a

22:21

better way to work and run your business, leave us

22:23

a voicemail at 708-628-7850. You

22:25

can also text that

22:28

number or send us an email to

22:30

rework at 37signals.com.