It’s Still Expensive to Build a Great Product

From our CEO

It’s Still Expensive to Build a Great Product

It has become commonplace to claim that the cost of starting a company has declined by an order of magnitude; just this morning Dave McClure took this claim as the starting point for his dazzling essay, Moneyball for Startups.

I think the claim is partially true, but overstated. And I wish we had data, not rhetoric, to settle the dispute. The costs that are supposed to have declined radically are usually for hardware and sales. But that is nothing new. Since I have been involved with startups, hardware has  been essentially free compared to the cost of software development. In fact, Jeff Bezos coined the phrase “hardware is free” in 1997, not 2007.

And the idea that a sales-force is expensive is ridiculous. It may be inefficient or a pain in the neck, and it may slow revenue growth, but a sales-force isn’t expensive. Since sales-people have existed, the terms of their existence have been that they pay for themselves, often within a few months of starting. And the idea that sales-people have disappeared is a myth: Google, Yelp, LinkedIn are filled with sales-people, cleverly hidden in corners of the campus that nobody visits.

What’s really changed is that software development has gotten easier and faster, due to agile development frameworks like Ruby on Rails. This is a genuine difference but it is incremental not disruptive: it still takes time and talent to build a stand-out web application.

Another genuine difference is that marketing costs have declined, though this too is overstated. We all deplore the 1999 dot.com that spent $10 million in venture capital on SuperBowl ads. But the reason we all noticed startups buying SuperBowl ads is that they were anomalies, not the norm. Nobody building the first railroad had to spend much on advertising to persuade people that it was a better way to travel across the country than a covered wagon. Disruptive technologies have always sold themselves.

And we’ve always known that. For ten years at least, the advice on Sequoia Capital’s website has been: “a huge market with customers yearning for a product developed by great engineers requires very little firepower.” For a hundred years, people have said that if you build a better mousetrap, the world will beat a path to your door.

So I still think the hard part of a startup is engineering a great product — — Picnik’s co-founder Jonathan Sposato once told me that a startup’s prime directive is to “build something hard” — and that the primary cost of doing that has always been great hiring great engineers.

Great software engineers, engineers who could easily get a job at Google, Facebook or Twitter, are worth hundreds of thousands of dollars on the open market. They will accept less if they are major equity holders in a great company, but not much less, at least not for very long. A team of five or ten engineers costs one or two million dollars a year.

To Google, a full team is worth much more than that; as soon as the company forms its value is more than the sum of its parts, creating an arbitrage opportunity for early-stage investors. Unsurprisingly, many companies don’t so much build great products as build great teams, hoping that Google will buy the company for the talent, even if it sets aside the product.

This is less a startup than it is a union. Since I’ve long believed engineers should reap the rewards of their work, a union may be a good trend, as long as your goal all along was a job at Google –and a huge signing bonus.

But if you don’t get bought, you have to build a product that is self-evidently better than any of its competitors, so that it can generate millions in revenue. And I still think that takes at least a year or two. Maybe it shouldn’t, but a true pioneer often screws up along the way, and you always encounter knotty problems you hadn’t anticipated.

What that means to me is that startup costs have declined, but not by the orders of magnitude claimed by angel investors. And I still wonder whether one reason startups are spending less money is that at least some are tackling less ambitious projects.

I am not judging that choice, in part because I am not even sure it is a choice. The path that Redfin has taken has not been any more deliberate than our own idea of fun, and it has been driven for years by our realization that no one will buy Redfin if we fail.

This is cold comfort for an investor. Dave McClure would say thanks but no thanks to a Redfin investment, because he is a seed investor who wants to stake only a little money and get a lot back in a very short time-frame. I like his approach, but I don’t think it’s the only approach. When he argues that his type of investing is the best type of investing, and that the alternative will go the way of the dinosaurs, I just want some data, or even a single example.

His article has no examples, even though we can easily think of counter-examples. Ask any casual technology observer to name a great investment and he might say Twitter or AdMob, both VC-funded deals. Maybe this is the wrong question to ask, since people inevitably focus only on the latest valuation, and not on the time and capital it took to get that valuation in the first place. The best way to phrase this question may be: what are the top 20 deals since 2005, as measured by internal rate of return? It seems silly to argue over what strategy makes the most money, when money can always be measured.

If anyone knows the answer to this question, I’d love to hear it.

Show Comments

Real estate redefined