Discover more from benn.substack
The emotionally informed company
The heart has reason—and conviction—that reason does not know.
“Correlation does not imply causation.”
We know. Everyone knows. Google knows. ChatGPT knows. An entire Wikipedia article knows. We know it so well that the catchphrase has become so ubiquitous—among everyone from professional analysts to Twitter reply guys—that we’ve almost become instinctively suspicious of correlations, presuming them to be guilty of being spurious until proven innocent.
I’d argue that it’s an overcorrection. Correlation may not imply causation, but it’s almost always its leading indicator.1 Just as it’d be silly for detectives to ignore a suspected serial killer’s ten-year history of asking Google if police had any new leads on a bunch of unsolved murders, it’s midwitting to dismiss correlations as useless because they haven’t been proven to be causal. Correlations are circumstantial evidence—but circumstantial evidence isn’t inadmissible evidence.2
When you’re running a company, the same is true for how you feel.
For a bunch of reasons—the general usefulness of science over intuition, the decades-long mass movement to be more data-driven, the rising cultural popularity of people like Nate Silver who sold us all on the “magic” of the “quantified universe,”3 sexism4—companies, and society at large, have dismissed emotions as useful inputs for making decisions. They are biased, corrupting, messy human distortions of an unyielding scientific reality. In ways that are both obvious and subtle, and both de jure and de facto, we’ve all made emotions another form of inadmissible evidence.
For example, in Silicon Valley, it’s trendy to use quantitative jargon that disguises emotion and estimation as mathematical fact. We don’t have opinions; we have Bayesian priors and robust latticeworks of mental models. We don’t guess; we make decisions probabilistically. Stories and anecdotes are small samples and Ns of 1. Unquantifiable quantities are nonsensically measured in orders of magnitude. And we use “emotional” as a pejorative. The important thing to know about our feelings, we seem to be saying, is that they aren’t feelings at all.
I think that’s a mistake. There’s a lot of signal in our emotions, even if it’s not scientifically robust or “statistically significant.” And companies and cultures shouldn’t equate their efforts to be more analytical with dismissing how they feel.
Inexplicable, but not imaginary
If you torture a financial model for long enough, you can make it say anything. You’re launching a big rebrand; that’ll increase the conversion rate on your website by a couple percentage points. You have a couple big features coming; you can probably charge higher prices once they’re out. Your business development team is growing; they’ll bring in more leads. And they’ll get more efficient because you’re about to make your first sales enablement hire. Combine all of these improvements together—all modest, on their own—and you can hit the aggressive growth target you’re promising your board.
The Excel file says it’s reasonable. Your heart tells you it’s not. You can’t point to the exact inputs that are off, but you know that something’s wrong with the output. You know that, somewhere in the ostensibly conservative numbers, a bunch of plausible assumptions are compounding into a gut-wrenching result.
As I’ve said before, “startups don’t die because leaders miss these feelings; they die because teams ignore them.” They die because they assume their numbers are right and their emotions are wrong. They die because their instincts don’t add up but their spreadsheets do, and because “in God we trust. All others must bring data.”
If we’re going to run companies with punchy quotes, I’d rather it’d be Blaise Pascal’s: “The heart has its reasons that reason does not know.”5
In a viral post about looking for love, Henrik Karlsson says that the best relationships are often the ones that can’t easily be explained:
When you talk about people you like, or rather when you talk about that thing that happens between you—you have to transform a very complex impression into a string of words. Some relationships can easily be compressed into a compelling string of words. This is usually because they conform to some sort of trope of how romance should look. In my experience, great relationships are harder to compress into a sensible string of words…
This will tilt the playing field against the wonderfully incomprehensible and singular relationships that you should be looking for.
I remember with a cold sweat that I almost turned Johanna [Karlsonn’s now wife] down because I felt confused by my inability to explain what our relationship was and why I liked it.
In other words, analytical thinking and its quantitative cousins have a bias. They favor things we can reason about, quantify, and express in a clear vocabulary. But those aren’t the only things that exist. This is obviously true in love, but it’s also true in how we make decisions.6 Fear, anxiety, excitement—there is something underneath these emotions. Something is making us nervous. Something is putting a pit in our stomach. Something is getting me out of bed early to work on this new project. “If you suddenly fall in love, if you develop a weird compulsion, if you’re incredibly anxious for no reason…there is a reason. You have to look for the reason.”
Ignoring that because we can’t rationalize it—or, more frequently, staying quiet about it because we don’t want to be seen as emotional or biased—is how we walk headfirst into problems that everyone saw coming. It’s how we convince ourselves that our current plan could work even when we know it won’t. It’s how we talk ourselves out of being excited about something we can’t quite explain. It’s how we tell ourselves the lies that Julie Zhuo warns us about:
If I just launch this next feature, customers will love the product.
If I ask the user what they want, they’ll give me the strategy.
If we make this one process tweak, that will fix our execution.
It’s not my fault the customer doesn’t see the awesomeness of our product.
This company does not care about <engineering/design/product/sales/insert-your-role> and that is why no one is listening to me.
Everyone around me knows what they are doing except me.
The quality sucks because we didn’t have enough time.
In all of these examples, we usually know that something’s wrong—or right!—before we can describe exactly what it is. We shouldn’t ignore that spark, regardless of how inexplicable it is. Not every impulse is a breakthrough, and not every moment of panic will save you from a disastrous decision—but nearly every big idea starts with an impulse and nearly every company-saving pivot starts with a panic.
The paradox of emotional disclosure
Several years ago, a Mode executive gave a presentation to the rest of our leadership team to make the case for opening an office in New York City. It was a tight and well-reasoned argument: We had a lot of customers there that could be better served by an in-person presence; we were getting more New York-based leads and face-to-face demos had much higher close rates; we could double our recruiting pool by hiring in New York.
But the pitch has one glaring omission: He never admitted that he wanted to move to New York.
In this instance, it didn't matter that much; we all knew he wanted to move, so his “oversight” became more of a joke than a problem. Still, it was an interesting thought experiment. Had we not known, should he have told us?
Most people’s default answer is probably yes. If he had an ulterior motive to argue for the office, he should tell us, out of some abstract duty to be transparent, or to tell the truth, the whole truth, and nothing but the truth.
But if you push on this further, it gets very weird and confusing. We should base our decision about the office on the business case for it. If he made a compelling argument for that, and we bought it, why do his motivations matter? Moreover, had he told us he wanted to move, it could’ve actually biased us, by making us assume he fudged his numbers or put his thumb on the scale. And that’s an awkwardly damning thing to admit, because it suggests that we’d be fooled by biased analysis unless we knew it was biased—and isn’t the whole point of quantitative analysis that we can judge it on its own merits?7
Or maybe, his partiality was actually good! We often assume that having a personal or “financial interest in something means you've got a conflict and your opinion is somehow ‘tainted,’” but if people are of sufficiently upstanding character, it’s just as likely that that “interest implies knowledge and understanding and leads to good decisions.” Or, per the point Blaise Pascal made four hundred years ago, his urge to move could’ve been driven by some insight that his heart felt but his head couldn’t yet describe. Had he told us about that feeling, we might’ve been able to put it to words, and find the causation in his correlation.8
To make things even weirder, our decision about the office was probably going to be an emotional one anyway!9 Numbers and analysis will only take us so far. The final decision wasn’t going to come from a calculator, but from our CEO, taking an emotional leap of faith.
And yet, we all pretend that both of these facts—that the pitch was emotional, that the decision was emotional—aren’t true. We hide our emotions behind our numerical technobabble, go through two-sided charades of pretending that we all just care about the numbers, and rarely acknowledge we feel something. But doing exactly that—putting our emotions on the table, alongside our graphs and charts—is probably the only way to untie this knot, and have a truly honest conversation about what to do.
There was another Mode exec who used to start his one-on-ones with his reports by telling them how he felt. Most notably, when he was mad, he'd start meetings by saying, “I want you to know that I’m mad at you.”
This is not your typical managerial advice. Today’s textbooks tell managers to act more like a neutral bystander—I observed this, the impact was that. If you confess your emotions to people, the standard guidance implies, they’ll assume you’re irrational. Though most management seminars10 don't explicitly tell people to hide their emotions, they promote a general approach of level-headed detachment, and they certainly don’t recommend introducing emotion into any professional conversation.
In the context of a culture that stigmatizes being emotional (and aggressively so against women and people of color), this advice makes sense. But in a vacuum, would it still be right? If you’re mad, you’re going to make decisions with that anger. You’ll probably look for ways to show people you’re mad, but do it in a professionally deniable way—by saying “we need to talk about this immediately,” by having a stern tone, by using formal and indirect language about observations and business impacts. Would it not be better if we could just say we’re mad? We don’t have to act mad, but we don’t have to pretend that we’re Vulcans either. In tough conversations, people are going to be more worried about our emotions—and spend more time trying to read between the lines to figure them out—than they are whatever substantive thing we’re saying. Hiding those emotions neither eliminates them nor protects people from them—so why pretend that it does?11
As businesses, we should be honest about the decisions we make in the same way. They’re emotional; they’re going to be emotional. We’re going to be emotional. Analytical histrionics don’t change that; they just cover it up.
Informed, not driven
One of the first cyclical corrections to the corporate movement to be more data-driven was a push to be data-informed. Data is useful for generating ideas and validating our intuitions, wise people would say, but we can’t follow it blindly.
Emotions should be the same. The best companies and leaders aren’t driven by emotions, but they don’t ignore them either. They are informed by them, and make use of them—as correlations, as evidence, and as a way to have honest conversations.12
And maybe most importantly of all, emotions give people and companies the courage to be daring. One of the other common criticisms of data-driven decision making was that it would lead us to local maxima. I think that’s true, but not because we’ll mechanically A/B test our way to some suboptimal peak; I think that’s true because analytical cultures are “wary of dubious forces, without and within, that might wreak havoc on their carefully constructed lives.”13 No force is more dubious than a sudden burn of excitement, or a sinking bout of despair. But the next great idea is in these feelings—and data alone will never give us the courage to move on them.
You could write an entire essay about this strip. Is it yet another “fooled by correlation” joke? Is it actually criticizing people who are over-eager to say correlation doesn’t imply causation? Is it a subtle and brilliant troll that’s meant to look like the former, so that people lazily link to it when they want to say correlation doesn’t imply causation, but is in fact making the opposite point?
This is not a legal statement! I’m talking about making decisions about businesses or whatever, not about convicting people of crimes.
NSFW, but, wow, ok, IsNateSilverAWitch.com really took a turn. The funniest part of this is that the new page completely ignores that the site is called IsNateSilverAWitch.com.
No discussion about emotional and empirical reasoning would be complete without at least acknowledging that they're both deeply gendered, and that, regardless of the merits of either, it's hardly a surprise that society elevates the one associated with the more powerful group. Interpreting that correlation is an exercise that’s left to the reader.
In this case, this particular thing seems unlikely; he wanted to move for personal reasons, not because he had a gut feeling that moving to New York would be good for Mode. But you could easily imagine other situations where this applies. If a product manager really wants to build one feature over another, is that some irrational feeling? Or is it their subconscious, lossily synthesizing dozens of customer conversations into a feeling? More often than not, I’d bet on the latter.
h/t to Robert Yi for linking to this interview. Also, as a tangential aside, his post makes a useful point about how making a decision is much harder than evaluating one. Prior to doing something, the universe of options is enormous; after doing it, all you have to do is judge that one thing. Which might explain why action—trying more stuff and seeing what works—may lead to better decisions than analysis.
I’ve been to a number of these, which probably comes as a surprise to everyone who’s ever reported to me.
Some people might say that this isn’t true. If you take a deep breath and act calm, you’ll start to feel calm. Fine, fair. However, you’ll never fully get rid of your emotions—nor, probably, should you, for all of the other reasons mentioned in this post. Furthermore, you can tell people you’re mad without acting mad. I’d even go so far as to argue that telling them you're mad actually makes it easier to act calm, because it relieves you of the urge to show them that you’re mad.
The David Brookses of the world like to talk about a parallel social dynamic that they call “pathologies of passivity.” As they describe it, some combination of wokeness, cancel culture, and participation trophy entitlement is discouraging Gen Z from being bold and taking risks. I don’t know about that, in part because the David Brookses of the world will blame everything on wokeness, cancel culture, and participation trophy entitlement. I could, however, see the quieting of corporate emotions as having somewhat of the same effect, by causing us to lose our conviction in taking chances.