My first iPhone was the 3GS, bought in 2010. The camera was merely okay. Photos taken at night were bad:
The resolution was only 3 megapixels! The camera couldn’t capture HD video, not even 720p! There was no front camera.
From a developer’s point of view, there were no APIs to control any photography settings like exposure. Once you took a photo, you didn’t have any image processing frameworks available like Core Image or vImage. And the hardware wasn’t performant enough for image-processing.
From the user’s point of view, once you took a photo, there were no editors like Lightroom or Snapseed. Let alone video editors like iMovie. If you wanted to share the photo online, there were no apps for services like Picasaweb. Instagram hadn’t even launched.
The trend of using a smartphone as a real camera for people who care about photo quality hadn’t even started, let alone run its course as it has by 2020. Everyone who cared about photo quality was using an SLR or mirrorless camera. I bought mine a couple of years after I bought my iPhone 3GS, and it was a huge upgrade. Smartphones were crappy cameras, to be used only if you didn’t care about photography or you didn’t have your real camera with you.
So this was the state of smartphone photography in the days of the iPhone 3GS.
In this world, there was only one app, Cortex Camera, that used image-processing to take noticeably better photos. You used the camera app that came with the phone. The idea of installing a third-party camera app to take better photos didn’t exist, any more than you install an app on your washing machine to improve how it washes. The phrase computational photography wasn’t even coined, let alone it reaching mainstream by having Apple mention it at WWDC.
After Cortex Camera, another excellent camera app Slow Shutter Cam launched.
The iPhone 3GS was a golden time to launch a camera app.
I entered the market in 2017 with NoctaCam, and pivoted to Futurecam in 2019. But it was too late. Futurecam startup failed.
One important reason was that it was hard for Futurecam to stand out: The builtin camera app was excellent by 2017, and kept getting better every year. And if it wasn’t enough for someone, there were tons of excellent apps.
If you googled how to take a long exposure photo, you’d come up with an excellent tutorial demonstrating the technique using Slow Shutter Cam. This tutorial is a better manual of Slow Shutter Cam than anything the Slow Shutter Cam folks wrote. And it was the ultimate advertisement for the app, showing all its power, and written by a third party, so more credible. It was hard for Futurecam to stand out in such a market.
When I was asked why to use Futurecam instead of these other apps, I came up with reasons, but they were incremental improvements. But you need a 10x improvement to stand out in a crowded field. For example, when Google was built, other search engines like Yahoo existed for years, were an established category, and were used by all web users. So Google had to be an order of magnitude better. Which worked for Google, but I couldn’t make Futurecam 10x better than the already excellent competitors.
Some of my answers to the question of how to use Futurecam were even retroactive justifications.
When I took a step back and asked myself which app I’d use to capture timelapses, it was Skyflow, not Futurecam. For hyperlapses, I’d use Instagram’s Hyperlapse app (Futurecam didn’t have hyperlapses, which many timelapse users wanted). If I were a professional videographer, I’d use Filmic Pro, not NoctaCam. For night photography, I’d use Apple, Huwaei or Google’s night mode, not Futurecam’s. For long exposures, it was a toss-up between Futurecam and Slow Shutter Cam. For light trail, I’d use my mirrorless camera, not a smartphone.
If I myself wouldn’t use a good part of Futurecam, or was indifferent, why would anyone else?
When you’re too late to a market, you end up building a “me too” product.
When I look back, I find that almost all of Futurecam’s competitors launched years before Futurecam:
Slow Shutter Cam: 2010
Filmic Pro: 2011
Cortex Camera: 2012
Cinemagraph Pro: 2013
ProCamera: 2013
Skyflow: 2014
Hyperlapse: 2014
Blendeo was launched after NoctaCam, but the company Flixel has been making mobile photography apps since 2013.
I was too late to this market.
You can’t time when to start a startup in a particular market, because it’s clear only in retrospect when a market has taken off. Even if you could foresee that a certain market will take off in 2023, you can’t plan backwards such as “It will take two years to build this, so I’ll start in 2021.” It’s not a matter of executing a defined plan. It’s not like Google Maps telling you it will take two hours to reach Mysore. Instead, you’re like an explorer in an exotic island where there’s no map and no road — you’re wandering through the jungle looking for treasure. You have to keep experimenting, trying out different technologies, building them yourself, trying different UXs to see what users can understand, different products, different target users, different business models and so on. If you have been playing with this for years, and know it as well as anyone else, you’ll be ready by the time the market takes off.
If trying to foresee when a market will take off doesn’t work, what about being a fast-follower? That is, being quick to see when a market has taken off, and jumping on the bandwagon then? That doesn’t work, either, because it’s clear to outsiders that something has taken off only years after it has:
Put differently, by the time a market is proven to be viable, others are years ahead of you in technology, understanding user needs, having validated dozens of hypotheses, and more. And many markets don’t work out, which is why many startups fail. That’s a risk you have to take to succeed. Startups are rewarded for taking risk, not for choosing to enter a market after it has been proven viable.
As Paul Graham says:
At YC we’re excited when we meet startups working on things that we could imagine know-it-alls on forums dismissing as toys. To us that’s positive evidence an idea is good.
If you’re too early to a market, it can be small. When Youtube was acquired by Google, it was estimated to have only 50 million users. Today, Youtube has 2 billion users. But 50 million users is huuuge for a startup. Even 2 million works. You can build a successful, healthy, profitable business. And explode when the market explodes.
But many markets don’t explode [1]. For example, VR remained a small market. But a small market can still be viable for a startup to be healthily profitable and sustainable.
In any case, don’t worry about being too early to a market. That typically means the market is small, and that’s enough for an early-stage startup [2].
[1] The ones that do attract disproportionate attention.
[2] If anything, early adopters are particularly keen, which is exactly what you want when starting out. I’d rather have a million people beating down my doors to get in than a billion marginally interested people.
Subscribe to Kartick Vaddadi: Tech Advisor to CXOs
https://kartick.consulting
Fun post! I lived through a bit of anecdotal evidence that confirms this: co-founding Google App Engine.
We started it in winter 2005, before AWS or almost any other cloud platform launched. The term "cloud" hadn't even been coined yet. Our inspirations were the make-your-own-social-network platform Ning, shared CGI hosts like Dreamforce, and Salesforce's Force.com, a limited multitenant Java host that was still years ahead of its time.
AWS managed to launch a bit ahead of us, as an a la carte IaaS with VMs, file storage, and task queues. App Engine was a serverless PaaS with built in NoSQL datastore, in memory cache, user accounts, email, and lots more.
Regardless, growth was...slow. We were *very* different from the status quo, so enterprises wouldn't touch us, and we had no good migration path for existing apps, so our early users were mainly hobbyists and new companies that started on us from from scratch. Snapchat and Khan Academy are two well known examples. Thank God Google's management didn't expect anything of us; if we were a startup with angel investors breathing down our necks to hit growth numbers, who knows what would have happened.
Even so, you know the rest of the story. GAE blazed much of the early trail to serverless. It currently has a >$1B run rate and led to Google Cloud, the third largest cloud computing platform in the world. And it was a hell of a lot of fun to build. Too bad we didn't do it as a startup. It probably wouldn't hae survived, but if it did, it might have been a unicorn.