Product-Driven versus Customer-Driven

I have lost track of the number of times I’ve had conversations about product-driven versus customer-driven businesses in recent years. It’s a distinction that just keeps cropping up, and has featured in every consulting gig I’ve had in the last three years, but surprisingly I haven’t found any treatment of it that satisfies me. So this post is partly an attempt to save myself from future repetition.

The distinction is central to many questions people ask in business:

  1. Which kind of business should you build?
  2. Can you transform your business from one kind to the other?
  3. Is one kind provably better than the other?
  4. How can you tell which kind is which?
  5. Which kind suits your personality?
  6. Can you hybridize the two and get the best of both worlds?
  7. Should you listen to customers?

These questions have been discussed for decades, at least since Henry Ford didn’t make clever remarks about faster horses. So why are we having this conversation with increased frequency and urgency these days?  Two words. Steve Jobs. 

But it isn’t just the inspiring dent-in-the-universe life of Jobs that is forcing this conversation, or even the fact of Apple’s exceptional performance in the market during a decade when many businesses were thrashing about in search of a direction. The reason this debate is at the forefront today is that the life and work of Steve Jobs suggested a set of polarizing, absolutist answers to these questions, which have historically attracted hedged answers beginning with it depends. 

Product-Driven Absolutism

The polarizing answers inspired by the life of Steve Jobs, and embraced nearly wholesale by Silicon Valley are, roughly speaking, the following.

  1. Which kind of business should you build? Product-driven.
  2. Can you transform your business from one kind to the other? No, it’s DNA.
  3. Is one kind provably better than the other? Yes; product-driven.
  4. How can you tell which kind is which? There’s an indispensable asshole at the top for product-driven.
  5. Which kind suits your personality? If you’re a winner, product-driven.
  6. Can you hybridize the two and get the best of both worlds? No. You must choose.
  7. Should you listen to customers? No.

These answers seem insane on the face of it. There are obviously dozens of counter-examples to each of these absolutist claims, especially the last one, which sounds like it could not possibly be true. So why is this extreme position, which sounds like a caricature of a business doctrine, and almost unnecessary in its extremism, getting such serious attention? 

Here’s the reason the position is taken seriously: there is now a widespread belief across the economy that we’re not in a business-as-usual regime; that the future is not like the past; that the rules of business in the new game are different; that pre-2000 mainstream business theories validated by many examples must defer to post-2000 Silicon Valley theories validated by few or no examples.

Let’s call this the new-game assumption. 

As a measure of the extent to which people have accepted the new-game assumption, in most conversations about business, very few people bother to contest these absolutist answers directly by citing counter-examples. It is usually easy to introduce a new-economy/old-economy watershed distinction into the conversation without encountering any resistance, and use that distinction to systematically devalue the significance of any old-economy theories or examples. Few attempt to argue that the new economy is really just like the old one.

The counter-argument taken most seriously is one that does not rely on counter-examples at all: just because Steve Jobs appears to have been an asshole, every asshole thinks he can be Steve Jobs. It is a counter-argument that essentially concedes the validity of the product-driven approach, and only attempts to question its causal structure and scalability.

We’ll take a look at the validity of the new-game assumption, but let’s try and figure out what we mean byproduct-driven first. The term is as murky as the term customer-driven is apparently clear.

So that’s the agenda for this post: figure out what product-driven means, then use that to first muddy up and then figure out the real meaning of customer-driven, and finally figure out if we are indeed in a new-game situation where there is no room for the latter.

Settle in. It’s going to be one of those longer posts. Let’s start by distinguishing product-driven from a bunch of things that kinda look like it.

Inside-Out Drives

A few examples — and there are only a few true big examples — of product-driven versus customer-driven businesses are the following. Loosely speaking, product-driven businesses develop visions inside-out starting with internal processes, while customer-driven businesses develop visions outside-in starting with patterns in the marketplace:

  • Apple under Jobs (P) versus Microsoft under Ballmer (C)
  • Pixar (P) versus Dreamworks (C)
  • Google (P) versus Samsung (C)
  • Facebook (P) versus LinkedIn (C)

Two important examples which don’t quite fit are Amazon and Elon Musk’s ventures (Tesla, SpaceX,  Solar City). More about those two when we talk about the validity of the new game assumption. Let’s try and figure out what the clearer examples tell us about product-orientation first.

The first thing to note is that product-driven is not the same as three related kinds of businesses it is sometimes confused with: auteur-driven, personal-itch driven (which could equivalently be described as follow-your-passion businesses) and formalism driven. Like product-driven businesses, all three are driven inside-out by internally generated visions. You could call the whole family inside-out driven. 

But there the similarity stops. The other members of the inside-out family aren’t quite the same as product-driven.

Auteur-Driven Businesses

An auteur-driven business (Steve Jobs is often misclassified as an auteur type) looks something like the realization of the singular vision of a dominant and forceful personality. The term is borrowed from film theory. As Wikipedia describes it:

In spite of—and sometimes even because of—the production of the film as part of an industrial process, the auteur’s creative voice is distinct enough to shine through studio interference and the collective process.

Whether they use the term or not, many observers of business implicitly seem to apply the auteur model to the work of the “assholes at the top,” in order to explain product-driven businesses. 

I am convinced this is a mistake. Auteur-like tendencies on the part of the asshole-at-the-top are necessary but not sufficient. If that’s all you have, you’re more likely to get an idiosyncratically conceived and run business like Richard Branson’s empire, Dyson’s little universe of gadgets and Dean Kamen’s Segway. Auteur-driven businesses certainly create entertaining spectacles, but rarely create top-tier businesses in terms of size, inevitable-seeming historical impact and profitability. The instantly recognizable signature style in auteur-driven businesses is associated more strongly with the individual than the product.

Though devotees of the cinematic arts might not like the thought, building a major business is a significantly more complex proposition than creating a major movie (as an aside: that’s one of the reasons the seductive ‘Hollywood model’ of industrial organization — involving studios and teams of free agents that form and disband easily — is not easy to port elsewhere: it’s not powerful enough).

The assumption that a creative voice can “shine through” an “industrial…collective process” no matter how complex the process, is not generally justifiable. Especially if the primary effect of the creative voice is to cause that industrial collective process to take shape in the first place (which is one way to define innovation — crucially, auteurs in the cinematic sense don’t necessarily invent new camera technologies or innovate film-production processes).

Itch-Driven/Passion-Driven Businesses

The second confusion is between product-driven and personal-itch/passion driven (the two are not quite the same, but for our purposes, they are similar enough that we can treat them together).  Many programmers and other kinds of hands-on builder-engineers in particular, are prone to this particular confusion.

Everybody reacts to the bureaucratic tedium of customer-listening committee processes by developing a private itch to just build something. Makers differ from most in that they can actually scratch such itches without waiting for or convincing others to join them. Their scratching may or may not resolve whatever design-by-committee gridlock drove them to frustration in the first place, but it will definitely create movement that will at least carry the itch-scratcher out of the frustrating situation as a side-effect.

But itch-scratching isn’t product-orientation. There is no good reason to believe that an itch represents a “dent in the universe” level of opportunity.  It may merely be a break-out-of-gridlock level of opportunity or a personal ticket out of frustration.

Itch-scratching is a kind of signal-and-noise hack: you respond to contentious, noisy and unreliable signals about what might be marketable by building something that responds to a very clean and reliable signal — your own. Any other equally clear signal will also do. Listening to a spouse or a child can also give birth to an itch-scratching product. The fact that the least-noisy, most-available signal is often your own does not necessarily mean the itch-scratching products have to be about you in some deeply personal way. Itch-scratching is the product-development equivalent of what market researchers call convenience sampling: you can’t figure out how to carefully construct the right kind of sample population so you sample what’s easily available instead.

Formalism-Driven Businesses

The third kind of business that can get mistaken for product-driven is the formalism-driven business. Formalisms — TRIZ or Christopher Alexender’s theories of patterns or Design Thinking (the capitalized phrase in the sense commercially popularized by IDEO) — differ from other inside-out vision-generation processes in a crucial way: they do not need an asshole-at-the-top.

In fact, the formalisms are implicitly based on the assumption that the asshole-at-the-top is dispensable, and that whatever they bring to the party can be codified into a set of processes. Getting the assholes out (or at least reining them in) is often the raison d’etre for formalist approaches to businesses.

So you could say that formalism-driven businesses try to replace an asshole-at-the-top with a bureaucracy designed to either keep them out or hem them in. A prophet either cast out or imprisoned by a priesthood and church.

Without going too deeply into it, I’ll just state my belief: formalism-driven businesses don’t really work at all. While auteur-driven and itch/passion driven businesses can and do work, and achieve more modest outcomes than true product-driven businesses, formalism-driven businesses cannot work because the assholes-at-the-top are in fact indispensable. I have never encountered a working and profitable example of a formalism-driven business. A little poking usually reveals that it is either not working (the formal process is a theatrical sham covering up the real process) or is not profitable.

So if product-driven businesses aren’t auteur-driven, itch/passion-driven or formalism-driven, perhaps they are just businesses that bolt on a different kind of “listening to customers” to itch-scratching or auteur tendencies?

The Just-Add-Grown-Ups Fallacy

A common characteristic of inside-out drives is that they can seem playful, irresponsibly disconnected from the real world, and capricious. Even the formalist ones, which often involve what one might call rigorous play — brainstorming exercises in rooms full of bean-bag chairs and such. So an entire tradition of innovation models has emerged based on the idea that to make inside-out drives workable, you just need to add adult supervision in the form of either non-playful people or accountability processes.

I call this the just add grown-ups fallacy.

I am not yet certain it is a fallacy, but I am 99% sure. It’s a fallacy because what is needed is a way to keep adult-supervision out for as long as possible (preferably until sustained and predictable profitability at the very least), not to add it as soon as possible.

Sometimes the fallacy leads to adding bean-counters armed with Excel models to the party or even making them the hosts. Sometimes it means adding an ex-military leader to add “discipline.” But the single most common way to add a grown-up element is to add a “listening to customers” element, in the form of a bolted-on customer-orientation process (which is not the same as a design formalism, but often comes paired with one).

The most polarizing and controversial position taken by product-driven absolutists is that they do not want to listen to customers. Just like children absorbed in play do not want to listen to adults. We see this in Mad Men when the firm attempts to add focus groups to the creative process to figure out what customers like. Don Draper responds with “they’ll like what I tell them to like.”

To understand why this is actually not a silly position, consider the idea of the Lean Startup, which is the most recent attempt to hybridize product-driven and customer-driven thinking.

In the Lean Startup model, practitioners hope that by changing who they listen to and how, they can fit a voice-of-customer forcing function without compromising a fundamentally product-driven process. So instead of focus groups and surveys targeting core customers (which by definition implies targeting an existing market with a definable core), Lean Startups look for atypical and often marginal “early evangelist” customers, who are both visionary about their needs, and so poorly served by existing solutions that they are actively looking for, and improvising, alternative solutions. The definition of the “early evangelist” in the process clearly creates a structural bias for detecting disruption opportunities over new-market opportunities, though true believers seem to insist it is equally applicable to both.

The general idea is to to find such early evangelists, and refine a product hypothesis through careful and detailed “customer development” in close collaboration with them. This process then, is supposed to drive product development. The input, Lean Startup types are careful to clarify, isn’t supposed to shape the whole vision. Just some subset of features that the product visionary/auteur decides is in need of customer-in-the-loop development.

What is that subset of features? It turns out that the features that end up in the subset in practice are those that represent adoption-failure risk along an assumed customer-acquisition path. The most extreme example offered is the vaporware landing page: here you test the very first stage of an acquisition pipeline (turning exploratory Googling into captive curiosity about a specific product).

So product development begins not with a limited realization of an auteur vision, but a progressive realization of the subset in need of adoption-failure derisking — the Minimum Viable Product (MVP).

The tension in the approach is obvious: there is no obvious way to prevent the progressive refinement of the MVP from tugging the product in directions that make progress towards the True Auteur North (TAN — an acronym for which I’ll hereby claim seignorage) increasingly difficult. There are various other minor problems, but this is the main one.

In the first partitioning of features into ones to be tested via MVP and ones to be deferred, a zone of lurking demons is created. This by the way, is not unique to Lean Startups. Any pairing of an inside-out drive and an outside-in drive will create such a zone of demons somewhere along the way. Call it a sort of visioning-process debt, by analogy with technical debt (deferred work known to be necessary in software development).

True Auteur North

Not only might the MVP random-walk to places from which the TAN is unreachable, it might also lead down a dead-end with no market. There is no law of the universe that tugs the process out of local-minimum traps and towards some large basin of attraction around a market-defining product that an auteur intuits is present. A law must be bolted on.

The Lean Startup model tries to resolve the tension by introducing the notion of a pivot. The pivot is the strategic reorientation action by which accumulating visioning-process debt is periodically paid down, just as refactoring is the process by which accumulating technical debt is periodically paid down. Unfortunately, there the analogy ends. While there is a vast repository of programmer knowledge about refactoring techniques, there is no comparable metis around pivoting and paying off visioning-process debt.

Various theorists of the movement seem to disagree even on what a pivot is though. Some seem to think it is an OODA-like reorientation move that follows every round of MVP development and testing, while others think it is a periodic reining in of the drifting random walk, either towards the existing TAN or a new TAN.

Either way, at this point, I think Lean Startups seem to lose their way, both theoretically and in practice.

The problem is with the very attempt to have the product-driven cake and eat the customer-driven one too. This pits conflicting forces against each other. Unless the project is incredibly lucky or informed by godly levels of foresight, inside-out drives and outside-in drives eventually have a falling out. The lurking demons emerge.

In the case of Lean Startups, the outside-in drive is to lower operating costs by lowering market risk. There is reason to suspect this may not actually work: it could be that processes that derisk front-end development costs simply create an adverse selection pressure. One that makes projects with the same total risk, but with more of it back-loaded in later development stages, seem more attractive at early stages, when competing paper-napkin ideas are being considered for development. So the Lean Startup process picks those problems that allow the tin-can to be kicked as far down as possible. Or in terms of visioning-process debt, the process favors problems where the lurking demons will emerge as late as possible.

Not a bad idea actually, if viewed as a way to buy time for other options (primarily financial ones) to open up, rather than a way to control costs.

Be that as it may, if we stipulate that the process might actually work, sooner or later it will run up against the logic of the inside-out drive. At the “crucial pivot” you will either have an auteur dictatorially throwing out the process in favor of one that offers a post-hoc justification for what he/she wants to do anyway, or you have a bureaucratic revolt, with the auteur being ousted in favor of a pure formalism-driven process.

I believe the continued (and widely noted; this isn’t just my opinion) failure of pure Lean Startup thinking to open up huge new markets is more than evidence that the particular model has its limitations. It is evidence that there is no middle ground between product-driven and customer-driven and that any pairing of product-driven and customer-driven drives will break down in an analogous way. The twain shall never meet. I haven’t yet made a strong case for this broader claim: in particular, there are two significant potential candidates for hybrids: Amazon and Elon Musk’s businesses. We’ll nail this broader no-middle-ground theory down properly when we get to the new game assumption.

The reason people even expect that kind of outcome from hybrid processes is this: product-driven businesses that succeed in huge ways often cause disruption of existing markets as a side effect, by happening to serve the under-served segments of those markets better. But that is neither their main effect, nor (more importantly) the intent driving them. The fact that the smartphone is disrupting the taxicab industry does not imply that it was conceived as a taxicab-industry disruptor. Or to make it a more focused argument, even Uber does not see itself primarily as a taxicab disruptor, but as a futuristic logistics business. The taxicab battle is just a side-show on the path to bigger things.

In other words, hybrid outcomes in big successes (some disruption of existing markets, some creation of new markets) does not imply that the model delivering the successes has to be hybrid. Pure-paradigm models can deliver hybrid impacts.

In fact disruption qua disruption cannot be the goal of a product-driven business. It can only be a side-effect of creating a new market. Whether there are any models that allow you to put a disruption goal front-and-center is still an open question. After watching proposed formulas for disruption (especially “self-disruption”) come and go for a decade, I am inclined to believe there is no such guaranteed formula for disruption either. The only rule-of-thumb available is to not attempt it from within the incumbent you’re trying to disrupt. That’s a bit like trying to choke yourself to death. You’ll pass out before you kill yourself.

But back to product-driven businesses and the new markets they create, which is more interesting to think about than disruption.

Selling Structure

Let’s summarize the argument so far. Product-driven business models are not the same as other inside-out driven models (auteur, itch/vision, formalism) and are not hybrids created by adding grown-up elements to those other kinds of business models, be they finance, operational discipline or customer-input elements. Attempts at hybrid models tend to die in an unlivable no-man’s land between product and customer-driven territories.

Now that we know what product-driven is not, let’s consider what it is. I propose the following definition: product-driven businesses sell structure. Specifically, experience structure.

To understand what that means, consider that old sales saw popular with customer-driven businesses: you only sell two things, happiness and solutions to problems. That is a very sales-driven posture, and customer-driven businesses also tend to be very sales driven (though not always).

To sell happiness is to fuel an existing desire-satisfying experience; an established craving. Such as eating ice-cream. To sell solutions to problems is to presuppose the existence of a broken experience. Situations are not viewed as “problems” unless there is an implicit expectation defining what “works fine” looks like. That creates a context within which a “problem” can be stated.

This view makes the most sense if you define “customer” not as a person or organization, but the way I did in an old 2009 post: as “a novel and stable pattern of human behavior.”

Customer-driven businesses fundamentally assume the existence of customers (stable patterns of behavior) with clear desires or problems. It may take you a while to identify and understand them, but fundamentally you accept the stable behaviors you find and simply try to feed them better. This is just another way of saying you accept an existing structure of experience.

You accept that people shop this way. You accept that households make decisions about buying vacation packages and health insurance that way. You accept that businesses make buying decisions using this process.

You accept habits. You assume that people don’t change. In fact you bet on their not changing.

So customer-driven businesses do not (indeed cannot) create customers in the Druckerian sense. They can only steal them from other existing businesses. There is a deep relationship between this create/steal distinction and the make money/take money distinction I talked about before: only expansionary government spending and new-wealth creators “make” money in any sense of the word. Everybody else takes it from somebody else. Only new-market creating businesses create customers. Everybody else steals them, possibly removing some inefficiencies along the way in zero-sum or negative-sum ways (stealing by removing inefficiencies and competing on cost is negative-sum in a sense: you win a market by shrinking it). The process leaves the universe of stable patterns of behavior the same or slightly shrunken. The assumption that people don’t change becomes a self-fulfilling prophecy.

This means customer-driven businesses are not true wealth-creating businesses. They are wealth redistribution businesses. Their DNA is deeply mercantilist rather than Schumpeterian/innovative, so to speak, because they operate by moving wealth (in the form of customers) around and driving down the costs of servicing them. But let’s set that macroeconomic angle aside for now.

By contrast, product-driven businesses expand markets in an absolute sense, create customers, and therefore wealth. They must be marketing-driven in a deep way: they “make people want things” rather than “make things people want.” This means they create and satisfy new needs and establish new patterns of behavior.

This act of creation can only work if potential customers are in a state of anomie: there exist untapped regimes in their universe of behavior that are marked by undefined restless energy and undirected curiosity (sometimes experienced as philosophical yearning). This creates a certain level of anxiety, and a regime of under-developed behaviors (often  bucketed under “play” or “hobby” behaviors, characterized by amateurishness, which is another way of saying pre-economic).

These are non-specific proto-customers (again in the customers as “pattern of behavior” sense — behavioral stem cells if you will) who are in a state of yearning for structure. Until they find structure, they cannot pursue either happiness through the structures or solutions to problems with the structure. Philosophically, meaning — in the form of the structure of habit — precedes both happiness and problem-solving when it comes to products and customers.

Only children can endure extended periods of undirected, improvised play without experiencing anxiety. Adults — at least ones with adult responsibilities like making a living or providing for others — inevitably start seeking ways to structure their anomic behaviors, often by seeking out hobbies and pursuing interests with an amateur level of skill. In the workplace, this structure-seeking constitutes much of work culture beyond formal business processes. This “play at work” goes beyond things like watercooler/happy hour/party culture to things like”networking,” vendors taking purchasing agents out to lunch, inside-information channels, elevator pitches, repositories of hacks known only to a particular group of engineers and so on.

We can now put these pieces together with the asshole at the top.

Authoritah as Structure Precipitating Irritant

Product-driven businesses create customers — novel and stable patterns of behavior — by selling structure into anomie. They do this by creating new products and services that structure and refine anomic behaviors into more complex behaviors with larger scopes.

An anomic situation is full of dissolved potential from which many kinds of structure could emerge. Anomie is the primordial soup from which new markets emerge. Some stimulus — an irritant — is needed is to precipitate some structure out of this potential.

The counter-intuitive thing about such precipitation is this: we need it precisely because there is no necessary trajectory the system must follow to exit its restless high-potential state, towards a high-realization state. There was no force of necessity that funneled the universe of goofing-off-at-work behaviors into Facebook and Twitter use. In another universe, the Internet might have invaded the workplace along a different tack.

This is why product-driven businesses tend to reverse yet another customer-driven business mantra: sell aspirin, not vitamins (i.e., sell to painful, immediate concerns that must necessarily be addressed, not to painless long-term concerns that perhaps don’t need to be). That wisdom, which is practically a Golden Rule for customer-driven businesses, is precisely what product-driven businesses don’t do.

A “vitamin” is a sufficient but not necessary stimulus (such as a sand grain in an oyster seeding a pearl or a dust particle in a supersaturated solution seeding a crystal) that precipitates some structure out of a potential field of anomic (structure-seeking) behaviors. This ultimately is the reason why an asshole-at-the-top is necessary, and is necessarily an asshole.

Given a situation with nothing dictating a particular behavior as a necessity, most people will choose in relatively indifferent ways, or experience anxiety at having to make a choice at all. But there are some to whom a particular sufficient path seems absolutely necessary. 

This is why product-driven is a more appropriate term than auteur-driven or authoritarian or genius-driven. The authoritah-figure feels compelled by a sense of absolute necessity that the world has to be a certain way. He or she needs auteur-like personality traits to act on this compulsion, but being susceptible to this sense of necessity is an independent trait. Without the auteur trait as an outlet, it can drive them insane.

In a way, unlike the self-consciously influential auteur, the authoritah-figure is as helpless as the ones they dominate: they are (or are convinced they are) instruments of “what technology wants” (sometimes called the technological imperative) rather than what they might themselves want as ordinary humans.

These are the personality types I profiled in my post, The Exercise of Authoritah (with Eric Cartman of South Park serving as the archetype). To quote myself:

Authoritah works its magic primarily by wearing down the collectee. It becomes so exhausting to penetrate the relentless barrage of bullshit, cajoling, threats, flattery, intimidation, pandering and bullying that you end up yielding. When this happens, you’ve been assimilated into a reality distortion field of the sort attributed to Steve Jobs and charismatic leaders of kool-aid cults who drive their followers to suicide.

This insistence that others play their way explains the distinction between an auteur and an authoritah-figure. The auteur only seeks to get his way within a limited scope, and seeks only to play by himself. As a result of the limited scope, auteurs are as likely to create fantasy universes as they are to make dents in the real one. As self-centered figures, they don’t necessarily want others to play their way. Only to behave in ways (which may be non-playful) that support their play. The work of auteurs seems characteristic of them, but not necessary or inevitable (do any Dyson products seem inevitable or necessary in some cosmic sense? What about the Segway? Or the businesses that operate under Branson’s Virgin brand?)

By contrast, the authoritah-figure not only wants to play, but wants others to play his way. And within the fullest possible scope for that kind of play within the real universe, rather than within a limited scope in a universe that might be pure fantasy. That’s where the dent in the universe motivation of Steve Jobs comes from.

The Creation of Customers, Redux

So how are customers created out of the two ingredients (a potential field of anomic behaviors and an authoritah-figure serving as irritant?). The process that ensues should be unpredictable in the early stages (you don’t know what kind of irritant will kick it off, and where within the potential field), but get more predictable as it evolves (the decidedly irregular shape of the particular sand grain has progressively diminishing influence on the shape of the much more regular pearl that forms around it).

I won’t get deeply into this since I’ve written about this at length before.

Product-driven businesses precipitate structure and create customers by stealthily extending (see The Milo Criterion) and metaphorically overloading  existing behaviors(see Welcome to the Future Nauseous), hacking imitation tendencies, creating social proof, and channeling group dynamics (see Ubiquity Illusions and the Chicken-Egg Problem).

The early path dependency/initial-conditions dependency means product-driven businesses create customers in an absolute sense that cannot be gamed. You cannot get from “early evangelist customers” in pain trying to improvise their way around Blackberry problems, to the iPhone, via a codified process. There is an a priori synthetic element that is neither discoverable by outside-in processes, nor negotiable within a codified inside-out/outside-in dialectic once discovered.

Neither will any random irritant do (that’s the reason simply being an asshole doesn’t make you Steve Jobs), which makes design-formalism-driven processes fundamentally flawed: the processes have to grow around the specific character of the unique irritant that works. Because that character is central to being an instrument of the technological imperative (or if you prefer, technology doesn’t just choose random assholes to get what it wants).

A final word on customer-creation: anomic behaviors are by definition under-developed, with plenty of room to grow. They are the opposite of highly refined and ingrained behaviors built around mature tools and technologies, with little room to grow.

Play is the prototypical behavior to build on, not refined skills or ingrained habits.

So a product-driven business must build on anomic behaviors that have found a certain amount of initial exploratory expression, have room to grow into a much richer space of structured behaviors, but lack the structure to grow and develop. The product becomes a trellis for the development and refinement of the new behaviors, giving birth to new needs and creating new customers (and producers).

Customer-Driven Businesses

Hopefully, clarifying the nature of a product-driven business has usefully muddied the apparently clear definition of a customer-driven one.

The apparently clear definition is this: a customer-driven business is one that uses a set of codified listening processes to figure out what people want, and then builds it.

Here’s why the definition is only apparently clear. To “listen” to a customer, defined as a novel and stable pattern of behavior, is to engage in pattern recognition. But a pattern that is stable is by definition not novel except under two conditions: you created the stability, or it is merely novel to you. 

Product-driven businesses create the stability and therefore the novelty.

Customer-driven businesses discover novel-to-them patterns. Such stable but undetected patterns can only exist in two places: inside the closely guarded core of incumbent markets, where competitors are trying to obscure your understanding of patterns OR in the interstices and margins of existing markets where nobody previously thought to look, or looked but didn’t care about what they saw.

If you can figure out who your competitor’s customers are, in the sense of “stable patterns of behavior” you can potentially one-up them by serving them better. You do this by recognizing hidden patterns from incomplete information. This is attacking a core.

If you can look in an interstitial niche where nobody has paid serious attention before, you might find a foothold on the margins. You do this by recognizing neglected patterns. If you also possess an appropriate lever (some pattern of falling costs or new technological capabilities), you might be able to make the marginal niche a new core and force surrounding old cores to shrink. That’s classical disruption.

Note that in neither case are you creating customers. You are stealing them or reprioritizing the importance of different segments. In the latter case, pattern-detection is necessary but not sufficient. Technological leverage to force the reordering of segments by new priorities is also necessary. You can’t just force a neglected marginal niche to grow at the expense of a large core by detecting a pattern and staring hopefully at it (this necessary element of leverage seems to be missing in the Lean Startup model). At best, you’ll be able to carve off a small chunk. To make that chunk grow, you need a source of leverage.

The big, macro-level question we’ve arrived at is the nature of the balance between the two regimes of business models.  Are we in a world where there’s room for both (with or without a no-man’s-land in between), or are we in a world where product-driven businesses represent vast and expanding opportunities while customer-driven ones represent limited and shrinking opportunities?

In other words, is the new-game assumption valid?

The New-Game Assumption

Here’s a piece of delicious irony. The only person I’ve heard seriously and credibly arguing (I don’t recall where) that everything hasn’t changed is Jeff Bezos, who has perhaps done more to create a new game than just about anyone else, Steve Jobs included. To add to the irony, he is also the most credible contemporary champion of being “customer-driven.”

In its (very low-profile) public posture, Amazon comes across as an extremely customer-driven company. There is that legendary empty chair standing in for the customer at Bezos’ meetings (notice however that it is empty). There are his legendary forwarded customer emails with ‘?’ in the subject line, guaranteed to drive his satraps into panic.

Unlike other new-economy businesses like Google or Facebook, Amazon looks like it is primarily a disruptor rather than a new-market creator. Its corporate identity is built around successful attacks on physical retailing, starting with books. New markets pioneered by Amazon, such as e-readers and cloud computing, seem like sideshows relative to the main act of all-out disruption attacks on traditional retail.

I believe this is purely a case of mistaken identity, gleefully encouraged by Amazon with its myth-making. Here’s the thing: what determines whether new-market effects or disruption effects are the side-effect is not a function of relative size of those impacts, but where the True Auteur North (TAN) liesAnd there are unmistakable signs that for Amazon, the TAN lies in new-market directions. Despite the massive scale of the disruptive impact on traditional sectors like retail, that is not the main act. The main act is still new-market customer-creating efforts like the Kindle, AWS and whatever else the company is working on. In a way, Amazon acts like it is just getting started on its real mission.

The second potential example of successful hybridization is Elon Musk. As with Amazon, his work seems to disrupt existing categories rather than create new ones (space launches, automobiles and the power grid). His products even look conventional, with much of the innovation happening at back-end infrastructure loci, and aimed typically at driving down costs radically or driving immature technologies to maturity. As an additional sign, his one true new-market idea, the Hyperloop, which would create an entire new category of transportation if it were built and proven workable, is one he doesn’t seem to be interested in pursuing personally.

Here again, the reliable indicator is his firm and publicly stated belief in reasoning from first principles as a way to think about technological potential, over analogical thinking. As I noted in the last section, any process of “listening to customers” is ultimately some form of pattern recognition — the classic sort of analogical thinking.

Reasoning from first principles is the antithesis of pattern recognition. It is also a specialized variety of seeking True North. Unlike Steve Jobs, who seemed to find a sense of True North in arbitrary places (such as an insistence on pure geometric forms), Elon Musk seems to have a preference for a particular kind of True North: the kind derived from fundamental limits in physics. Musk’s True Norths fit the formula, if there is no fundamental physics reason why something cannot be done, it can be done and we just have to think it through. 

It is an interesting intellectual orientation that is the opposite of the common one. Most of us don’t believe it is possible to do something we haven’t seen done unless somebody presents a concrete case for how it could be done. Musk seems to operate on the basis of non-constructive proofs: if there is no proof it cannot be done, it must be possible. This leads to looking at existing stable realities (including stable behviors or “customers”) as being stable not because they are held in place by fundamental limits, but because nobody has thought to destablize them in interesting ways.

So Musk too, is product-driven, not customer-driven.

It’s not conclusive, but I think we have a robust case that there is no middle ground and no way to hybridize customer-driven and product-driven thinking.

Which brings us to our final question: is product-orientation better at an absolute level, and poised for a winner-take-all victory?

Big-Bang or Cyclic Economy?

Yes and no.

I do think the evidence is in that it is absolutely better at this time in history, but it remains to be seen whether it will persist. 

The industrial economy is gradually seizing up in a gridlock of entrenched industry structures. Stealing customers from competitors and disrupting from internal margins (as opposed to disrupting from a position in a new market) is increasingly difficult and expensive. Margins are shrinking and revenues are either flatlining or trending downwards, creating inexorable pressure on business models and gradually squeezing out all the room to maneuver. Listening to customers and detecting patterns is getting harder and harder, even with tools borrowed from the new economy.

On the other side of no-man’s land, the new economy just seems to keep growing, and things keep getting easier to do, rather than harder.

It is even clear from my brief 3-year history as a consultant selling vitamins to both sides: you have to work far harder, and under more severe constraints, to sell any kind of services (vitamin or aspirin) into the old economy. In the new economy, whatever the size of the client, decisions are made quickly and money flows quickly.

But the interesting question is whether product-driven businesses will continue to be absolutely better into the future. This is not a trivial extrapolation of the current condition, though a lot of people seem to think it is. To argue that product-driven businesses are absolutely better and will prevail for the foreseeable future, you have to argue for an endlessly expanding universe of anomic human behaviors, yearning for imposed structure. You have to argue that the opportunity there is much greater than in the universe of stable behaviors waiting to be disrupted or served better. That the new economy will subsume the old rather than be “adopted” by it.

There are two ways this could be true. The first is that we could be at the start of a period of unbridled growth in new market opportunities opened up by the Internet — an entire new space of anomic behaviors with different growth physics. No matter how fast we run to create new markets, the horizon will recede faster and there will always be bigger markets beyond. This is the boundless-frontier version of product-orientation, based on an assumption that there has been a secular change in the very nature of economic growth.

The other way this could be true is that we are still in business-as-usual conditions, but at an extreme of a long-term cycle that is beyond the reach of living memory. Carlota Perez’ theory of 50-70 year cycles of installation and deployment of new technology eras is an example of this kind of argument. It’s business as usual, but the last time anyone experienced such exceptional conditions favoring product-driven thinking was the 1930s.

Note that in this cyclic model, we have to be careful about separating corporate life-stage effects from corporate generational effects. Every business, as it ages from startup to mature business, seems to inevitably go through a polarity flip from more product-driven to more customer-driven, usually triggered by the departure of the founding asshole-at-the-top. This effect can largely be ignored: the DNA-level product or customer orientation of a business is a function of the state of the long-term cycle of which it is a part. Infant businesses that are born at the wrong time in a cycle face higher risk of infant mortality. Early generation companies of a technological era are more product-oriented. Later generations are more customer-oriented. At the end of a cycle, markets sustained by that cycle shrink to a fraction of their peak size.

In the boundless-frontier model, product-driven businesses may have gone in and out of fashion in the past, but are here to stay for good in the future. Because Internet. In the cyclic model, we’ll be back in a dominantly customer-driven phase in another 30-50 years as this era matures. It is not yet clear to me which of the two is more likely, or  whether there is an element of both. In both cases, however, the space of anomic behaviors grows, racing against new demands on attention, due to increasing automation of things that used to absorb attention.

Of course, neither model addresses the question of sustainability and the evolution of the resource base under different growth or collapse scenarios, but it seems to me that that is an orthogonal axis. In (say) a collapse scenario triggered by global warming or some other resource shock, responses could still be “customer-driven” (UN committees commissioning global surveys and the World Bank offering cheap loans for solutions to committee-defined statements of problems) or “product-driven” (someone pulling an Elon Musk on a critical resource problem and rethinking limits from first principles). Collapse or gradual shrinkage scenarios with population declines need not also be regressions to earlier technological eras.

We’ll leave that axis for another day. Let’s wrap up by restating the answers to the opening questions. This time with some telegraphic justification.

  1. Which kind of business should you build? Product-driven, because it is far harder to find money to build customer-driven ones today in the old economy than product-driven ones in the new economy.
  2. Can you transform your business from one kind to the other? No, it’s DNA. Product-driven businesses evolve their core processes around what technology wants, as channeled by the authoritah-figure at the top, and oriented towards the True Auteur North (TAN). Customer-driven businesses are built around pattern recognition processes devoted to stealing customers rather than creating them. The only path between them is a near-death experience.
  3. Is one kind provably better than the other? Yes; product-driven. For now, and while it remains easier and cheaper to find anomic proto-markets to structure than to steal customers or disrupt existing markets.
  4. How can you tell which kind is which? There’s an indispensable asshole at the top for product-driven. These individuals are indispensable because they are the irreplaceable channels through which the force of “what technology wants” expresses itself. They are assholes to others because they feel controlled by forces larger than themselves and incapable of compromising the way nice people do. If the entire company seems to be aligned to the TAN of this individual, and the business looks like it exists to maximize the leverage available to him or her, it is a product-driven business.
  5. Which kind suits your personality? If you’re a winner, product-driven. Unfortunately, being customer-driven by temperament — being good at listening and pattern-recognition, and having high empathy — puts you on the wrong side of history in 2014. Being product-driven — wanting to make things, a preference for first principles, and a desire to explore uncharted territories — puts you on the right side of history.
  6. Can you hybridize the two and get the best of both worlds? No. You must choose. The middle is a no-man’s-land where any business model will eventually get torn apart by the attempt to conform to both inside-out and outside-in visions.
  7. Should you listen to customers? No. If you’re product-driven they don’t exist yet, so you actually can’t until you create them first. You can listen to ones that do exist, but chances are low that you can steal them from their businesses that created them or currently own them, or turn their frustration at being under-served and marginalized into a huge business.

Acknowledgement: This article represents the synthesis of a great many conversations with a great many people over the last three years, many of whom I cannot cite for various reasons, and I owe them a debt of gratitude. That does not mean, of course, that they agree with all or even most of what I’ve said.  

Random Announcement #1: There is a meetup of the recently formed London Ribbonfarm Readers meetup tomorrow (Friday the 25th) organized by Ollie Glass. Seems like they had a fun time last time, so if you’re based in London, try to make it.

Random Announcement #2: I’ll be speaking at ALM Chicago next Thursday (May 1). Come by and say hello if you’re attending. 

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Venkatesh Rao

Venkat is the founder and editor-in-chief of ribbonfarm. Follow him on Twitter


  1. +10. Confirms my own biases and strong negative reaction for anything associated with lean startup, MVP, and a general lack of focus on data (how you acquire it, what’s built on top of it, how you use it. Data and networks win).

    Have evolved my thinking into concept of MVP+. =Minimum viable platform or minimum viable full-stack product where data and networks matters most.

  2. I’m sorry, I just couldn’t read it all, and I do appreciate your going into detail and sharing these great conversations. I’ll try again. But here’s an analogy: a concise and to the point article would have been a customer-driven one. Sharing this in parts (say several articles over time), would have been both product & customer-driven one.

    Here’s my two cents: what’s needed today is neither a product-driven nor customer-driven business model. Businesses need to do what Mintzberg once said (in a article on organigraphs) and that I sum up as organizing and focusing based on the essentiels. You can use lean, agile, CMMI, I really don’t care what methods, frameworks, processes, you want to use to do this, but every entreprise (non-profit or for profit, that is to say any entity trying to accomplish a PURPOSE) has to balance looking inward and outward at three areas (multiple nouns used for the sake of demontrating that I’m not limited to any set term):
    Customers/Clients/who I serve
    Resources/Employees/what I need to serve them
    Products/Services/Offering/what I do to serve them

    and over this three areas is PURPOSE, why are we here people?

    So you need to be a “trifecta” company (customer-resource-product-oriented company or a CRP company), but I kinda like trifecta, reminds me that I need to win in those three areas if I’m going to make it as a company. And the best CRP companies do what they do in line with their PURPOSE. Raison-être.

    that’s it.

  3. I love how you use more words that I have to look up in another browser tab than any writer ever and then casually slip in “Because Internet.”

  4. Venkat, I peimise I will read this post fully in time, but wanted to share a quick insight. When you talked about product-driven companies with the a**hole (aka irreverent) on top I couldn’t help but think of one my favorite companies that does this extremely well. 37 Signals. Extremely product driven, and with not one but two irreverent guys on top.

  5. Although I’ve recently begun to follow Ribbon farm, I just cannot believe that I completed reading this post entirely in one go! Will go through it once more to mine out more nuggets. At my first reading, I suspect if what you point out as “precipitating structures” can be expressed alternatively through “defining context”. To clarify, when I say “defining context”, I look at the process in whole that is generated in which behavior, ecosystems reorder and align themselves with the new context.

  6. On mobile, apologies.

    When costumer is defined as “stable patterns of human behavior”, then your analysis is spot on. In many ways, that’s a useful definition, particularly when it comes to the early stages of a company.

    When you define customers as “individuals”, however, it becomes possible to create new customers by helping more people adopt these patterns of human behavior.

    Take early Facebook. The “social networking” customers(as patterns) already existed, but they expanded the total set of customers(as people) by finding the right way to establish those patterns within an under served market (college students). Now Facebook Zero is doing the same thing, but with developing nations / dumbphone users.

    I feel that there is a synthesis here, but saying that customer-focused businesses are only a zero/negative-sum game seems to ignore customer creation as (pattern) adoption vs (pattern) creation.

  7. Mike Leach says

    The zero-sum analogy was thought provoking. I’ve always viewed Uber as displacing and disrupting the taxi industry by modernizing the IT infrastructure involved in a common service transaction. There is no evidence of a new product, that I can see. People are leaving Taxis for Uber.

    Given the number of “Uber for X” businesses on AngelList, the pattern is evidently repeatable to a large number of service industries. It seems difficult for Uber to become a horizontal platform at this point. So, why wouldn’t this be customer-focused? (and what is necessarily wrong with that? it’s a great service…. or did I incorrectly interpret the article’s point?)

    • It is non-zero-sum to the extent that it introduces new patterns of mobility that are not obvious 1:1 substitutes for existing patterns. Are there A-B movement patterns due to Uber that would not have happened at all if choice set had been restricted to {drive yourself, rent a car, take a taxi, take public transportation}? I think there are. I can think of cases where I took trips/made travel arrangements that I wouldn’t have if Uber hadn’t existed… other modes would have been too inconvenient. There are also indirect effects: dollar saved on travel is dollar available for other things.

      That’s just the base case. Other emerging non-zero-sum use cases in delivery services, freed up spaces that would otherwise go to taxi stands etc.

      A more mature example is online labor marketplaces. Sites like odesk are more than just cheaper, more frictionless ways to hire temp labor. You can use labor in new ways. For example, quickly A/B test two workers on same test task and pick the better one. Or use mturk to build more complex patterns of cross-validation and redundancy in sourcing work…

  8. I’m not 100% convinced with the above, after reading Jim Morgan’s article on “Product Focus = Customer Focus”

    Any comments and thoughts?

    • The process described in the linked article is pretty solidly “customer-driven” under this article’s framework.

      A key line is: “To become product focused, companies must understand, better than anyone else, the critical product/service attributes that truly matter to their customer.”

      The articles are using different semantics for what ‘product focused’ means. In Venkat’s dichotomy, the focus on ‘attributes that truly matter’ to (implied to be existing) customers, means that the product produced will be well anchored in the existing ways of thinking about the space.

      Upon release, a product-driven product will likely (perhaps even necessarily) express attributes that *don’t* currently matter to its eventual customers, because those potential future customers can’t evaluate or perhaps even conceptualize those attributes.

      Most of the rest of the linked article could apply to teams building customer-driven products or product-driven products. The difference in some ways is just epistemological: Where do the core ideas come from? Learning about an existing market? or trying to create a new one?

  9. The doomed empath in me wants there to be the possibility of working for a company that doesn’t have an asshole at the top but still makes money.

    Could a formalism or methodology be created that attempts to reason about a theoretical technological imperative and provide useful insights without requiring an actual asshole/prophet to force the future upon us? Or perhaps a system that bestows asshole powers in a limited way, the way democracies and limited monarchies are more stable long term than absolute monarchies, i.e. a indispensable asshole on the top.

    You acknowledged that you were glossing over the notion that formalism driven businesses don’t work, but can you speak to whether they fundamentally can’t work? or if all the previously tried ones just didn’t actually work?

    Great article, and congrats on getting it out. Sounds like it’s been stewing in the noodle for quite some time.

  10. Greg Nelson says

    I like the analysis, but I have to argue that you can do both. In fact, I think it’s imperative to do both and all of the examples you listed actually DO do both. The problem I see is your definition of customer. Like Adrian Ryan mentions, I think that it is not sufficient. I also think it only really covers a small extreme minority subset of customers.

    Even using your definition, here are a few logical tests to your arguments: How is Tesla creating customers? And similarly, how did Toyota create customers with the Prius, or Ford with the model T. So which one of those is product driven versus customer driven?
    Also, with Amazon and the kindle, can that really be defined as product centric customer creating/market creating (especially considering it wasn’t even close to the first), or is it possible that it stemmed from the idea of removing a pain (cost of shipping, delay, size/weight of books)? I see the kindle as much less a product in search of a new customer and instead, a product designed around an existing customer (and existing technology) that is provided as a loss leader (in an effort to lower the barrier to purchase an e-book)

    I think the difference between being product focused and being customer focused is the difference between knowing your customer so innately you can define (and when you cannot define, use an “I’ll know it when I see it” approach) what they want, and having to ask them (directly or indirectly) and interpret the data. You can’t really learn an entirely new direction from customer surveys, but you can, and should, bring them into consideration for design decisions. Being customer focused, doesn’t mean you have to show your product to a customer, it just means you use information about them to make informed decisions as to what constitutes the MVP, and what’s not worth extra developement time. with that definition, Assholes at the top CAN be customer focused and not even realize.

    I think people are quick to assign revolutionary tags to things that are largely successful, even though they are almost always evolutionary.

    • I’d say that individuals are customers of a specific product only to the extent that they have learned behaviors relevant only to that product. Those are the behaviors created by that product.

      So Tesla induces some unique new behaviors, but also exploits established commodity behaviors inherited from the car industry, which are no longer owned by any one product.

      Interesting that you focus on being first. I think first to market is meaningless if the market is being created. Pre-Kindle readers never got to “market” in a sense. Except the uninteresting one of fringe experimentation. The Kindle succeeded to the extent that it created new behaviors: impulse buying of books while mobile, differences in how people read (for example, use of inline dictionaries, alternative ways to highlight and take notes compared to paper).

      Revolutionary/evolutionary is a distraction to some extent, and an artifact of scale. On a sufficiently broad scale, everything looks evolutionary (that was my part of my point in the Future Nausea article). On a sufficiently narrow one, everything looks revolutionary. It is much more useful to ask: what behaviors are new here? How portable are the behaviors to contexts outside the new product/service? What are people doing that they didn’t do before? How much is that novelty worth economically?

  11. As usual, superb analysis. Product-driven companies that create new markets attract copy-cats / competitors which eventually leads to the new industry falling into an old-world paradigm of a customer-driven business. Do you think the survival of product-driven companies is contingent on regularly creating new markets? If yes, are we looking at product-driven companies that survive only as long as the uncompromising asshole at the top?

  12. This post has altered how I think about product development. Until reading this, I considered myself to be a true believer in lean startup methodology. You’ve convinced me that it is fundamentally flawed. Now that I’ve been stripped of some of the guiding principles that I apply daily to ship products, I have some questions for you.

    It seems possible to have some of the product-driven personality traits (a preference for first principles, a desire to explore uncharted territories) while not having a necessary characteristic: being an asshole. And I believe that drastically changing one’s assholeness is probably impossible. In My Age of Anxiety, Scott Stossel explains that people have different baseline levels of autonomic arousal. Some people, like con men, aren’t bothered much by seeing negative emotions in other people. Other people are blessed/cursed with extreme sensitivity to others reactions. If you’re wired with a non-low level of autonomic arousal, you’re doomed to not be an asshole.

    My primary question is this: as a product builder who is determined to realize their visions for restructuring experiences, what’s the optimal way to play a non-asshole hand? I see two primary options.

    (1) Concede that, because you’re not an asshole, you should hitch your wagon to an inspirational product-driven asshole. I.e., try to work for the next Steve Jobs. With this option, your energy can contribute to the highest world impact, yet the TAN is not one that you own.

    (2) Lead a company to the best of your ability in spite of your non-asshole condition. Maybe this means you won’t build the next Apple, but perhaps you can accomplish something less ambitious but still significant.

    For either approach, we need a methodology that replaces lean startup methodology in making the TAN actionable. What are the principles building for building products under the leadership of a product-driven asshole? How did Uber or Amazon choose which markets to incidentally disrupt on their way to the real TAN? For the non-asshole leader who strives to build a product-driven company, what are workarounds will yield the best possible results (even if their ultimate potential is capped)?

    And one other question: what if you’re building a technological solution for an old economy market (e.g., manufacturing)? Are you subject the rules of the new economy or the old economy, or is it its own game? Maybe there is opportunity here for a non-asshole to enact serious change.

    • I’ve never really met a proper exception to the asshole-at-top rule for big-impact companies. If you run into one, make sure you ask him/her and everybody around them how it works. If I knew, I’d have put it in the article. If you turn into one, you’re welcome to guest blog here about how you did it and prove me wrong :)

      Old industries: if you want to disrupt using new technologies, same rules apply. If not, you’re more of a diplomat/elder-statesman/internal-succession type leader who is there to keep the peace, not wage war. You can absorb sustaining innovations/solutions that don’t threaten industry structure or power calculus. That’s just basic build/buy decision-making for clear needs.

      If you’re trying to absorb new disruptive solutions, I’d say the asshole rule applies.

  13. Thanks for the response. To clarify, I’m not trying to suggest that there is an exception to the asshole-at-top role for big-impact companies. My questions have to do with how, as a non-asshole, to maximize your personal impact by either (1) working for an asshole at a big-impact company, or (2) leading a lesser-impact company.