In August of 2022, Heroku announced that they would discontinue free plans.
This would end an era for Heroku as I knew it: the go-to web hosting platform for getting your first app into production. Every bootcamp I knew of used Heroku to demonstrate deployment. Most Rails shops started with Heroku until their apps got too big for the platform’s per-unit prices. I never had a conversation about a Rails deployment in which Heroku was not the default. Immediately, the chatter began about migrating or letting old projects die. Since Heroku already had a reputation for being cost-prohibitive at scale, folks expected the formerly-free options to become exorbitant, or at least out of reach for educational purposes.
Why would Heroku durf themselves like this?
Let’s talk about the sustainability of internet businesses.
You will not like what I have to say.
In what I’ll call the Centralized Software as a Service Industry, it’s actually not that easy to break even.
How did we get here? Well, once upon a time, making money on the internet was much easier. The market was not as saturated, the American1 public not as jaded toward ads, and programmers not as expensive as now.
I’ve talked about this period some; why it was so easy to make something “visionary” twenty five years ago, and why it’s not like that now. The bar for making a “visionary,” market-breaking success story is much, much higher than it was in 1995, when the dot com boom created today’s tech millionaires and billionaires.
Frankly, even the founders who struck it rich in the ’90s can’t clear that bar. Plenty of their later golden child investments have tanked.
But those dudes are sitting on accumulated wealth from the ’90s, and they don’t realize they’re out of touch with tech today. So they “angel invest” in projects that excite them—that probably would have been breakout successes in 1995.
And for a little while, or sometimes a long while (hi Uber), companies can nurse that sweet, sweet 1995 money by just keeping those ’95’ers inspired and excited. This is why a lot of unicorns have kook founders. Kook founders are awesome at keeping ’95ers excited.
But it’s not 1995. You cannot make a bullshit, insecure popsicle stick fort on the internet in 6 days and turn a killing (ebay, 1995).
Operating costs have skyrocketed. Plus, programmers cost a lot of money now. The only thing more expensive than a tech team is an ex-techie turned exec who thinks they should take home the amount that whatever ’95er they’re a fanboy of took home.
Furthermore, it’s harder to make revenue for 3 reasons:
1. No one wants to pay for the product.
People will rip it, they will pirate it, they will hack it and give out codes and you-name-it, to not pay for a thing. The number of people who will use a thing is consistently 100-1000x the number of people who will pay a red cent for it.
But that’s fine, right? No one every had to pay for network television or early sites because they had ad sponsors.
2. No one wants to suffer an ad to NOT have to pay for the product.
They will again hack, and adblock, and you-name-it to not see an ad.
You fucking hate it, but the truth is, ads keep the internet free; without them, the internet is a luxury good. At least so far.
That’s where Heroku is at. It can’t operate without luxuryfying. Lyft and Uber prices have increased, too. Twitter moved that way with paid verification and pushed for putting more ads and promoted posts on the platform.
I should mention that folks’ aversion to ads is not unfounded, because for about 15 years companies have tacked hard into collecting their data, often without their consent, to try to drive up ad revenues by personalizing the ads.
Ad personalization doesn’t actually work that well (some research suggests a 4% lift over search query-based ads), but daaamn do tech companies BELIEVE in it, and so do consumers: consumers THINK they’re more likely to buy if ads are personalized.
Why? Well, part of it is that people THINK they’re being personalized to, even when they’re not. I wrote about that once, too—and also about the intersection of data privacy and search.
But if people don’t want to be the paying customer, and they also don’t want to be the product, what’s left is to force people to buy the internet thing by eliminating viable alternatives. Which companies COULD do, except that much RICHER companies are ALREADY doing it.
3. Folks who will pay, or watch an ad, are already doing it with someone way bigger.
80% of internet-based product distribution—of both digital and physical goods—runs through about four companies who are rich enough to buy anyone they want, and who have so far experienced almost zero antitrust limitations.
“Winning” as a web biz is no longer building a sustainable, profitable company. It’s picking a vertical that one of those four companies wants to dominate and doesn’t yet, and then getting chosen by that company as the market entry ticket that they’re gonna buy.
And that “win,” by the way, is only even really a “win” for the founders and maybe the C-suite at the acquired company. It is VERY frequently absolute hell on the actual team. Helllll on the team. Devs burning out right and left. Breaking down. Leaving.
So, who is profitable in Centralized Software as a Service right now?
1. The Four (Amazon, Apple, Google, Meta-formerly-Facebook)
2. Software whose clients are businesses, not people (think Atlassian or Microsoft). Bigger businesses are better because the actual purchaser person is less invested in the purchase decision.
That’s most of the list, I’m afraid.
A lot of the industry got sold on the idea that an expensive undertaking would pay off, and it hasn’t. Now they gotta either make it work or admit they were catastrophically wrong, fold, and fuck up their entire team’s lives.
And leaders, understandably, don’t want to do that. So they try to come up with literally anything else. Twitter sells to an unhinged billionaire. Heroku boots its unpaid clientele. One Medical sells itself to Amazon.
So what can we do?
I don’t have an easy answer to this question. I wish I could console myself with the knowledge that people much more powerful than me don’t seem to know either, but that’s even more terrifying. If I personally were to try to make a software product right now, I know I’d do the following things:
- Focus explicitly on the needs of someone who the dot com boom categorically ignored, because I believe that this is the only remaining place to find visionary technical ideas.
- Optimize, not for growth or ubiquity or profit, but for long-term sustainability within a specific community local to a place or need.
Those two ideas make rich tech bigwigs vomit in their mouths a little. “But we’re supposed to save the world!” they’ll tell me. Newsflash, boneheads: I know that tech grew up adjacent to the whole 60s hippie world-saving thing, but a) that movement failed pretty miserably, didn’t it? And b) correlation does not equal causation. It’s convenient to buy into the idea that the same project that “saves the world” also gives you access to mansion money, but when we step back and take a look at that premise, “convenient” is about the only thing it has going for it. It’s not realistic. It’s not empirically supported. It’s not even that savory when you expand the circle of people you care about—I mean actually care about, as in you would inconvenience yourself personally for their benefit—beyond yourself and some close consorts. To help people who have it worse than you in such a systemically unjust world almost always extracts logistical, emotional, and financial costs on the helpers. My hope is to talk about this further in a future post, because perhaps it is in removing those barriers that the next sustainable idea lies.
- Americans are uniquely averse, per capita, to clicking on ads. I work for Mozilla, a browser with large constituencies in the U.S. and Germany. Despite the fact that Germany possesses about a quarter of the population that the U.S. does, German constituents have a similar number of aggregate ad clicks as American constituents. To quote one of our VPs, “Germans love clicking on ads.”
If you liked reading this…
I don’t write about money all that much on here (at least not so far), but I did write this piece about the psychology of tech money and this piece about how being high-income doesn’t make you not exploited by exploitation capitalism.
Look; it’s been a dark year.
If you need something less decidedly morose and you think you’d be entertained if I were just angry, then this piece about the way we misuse the idea of a “best practice” in tech might suit your fancy.