There are countless “feel good” stories of startups that miraculously go from just weeks of cash left, to explosively profitable; this is a story of what it’s like to still be in the midst of that struggle. Just in case you’re wondering what it’s like.
It took a lot of guts, humility, and a healthy dose of uncertainty, frankly, to decide whether to write this intimately transparent story about our startup.
But we need to get something out of the way first. Apart from the final sentences of this story, this is not a feel good read full of puppies, poppies, and pixie dust; it’s not another Hollywood story of a startup’s spectacular success against all odds, blossoming from banana slug to billion dollar unicorn overnight. Not even slightly.
No, for that, you need to read this, one of the most powerful, uplifting, and damned near tear-inducingly inspiring pieces (by Josh Pigford) on that most dreaded of startup scenarios, running out of cash; a scenario so foreboding, so ominous, it’s spoken about with averted gazes, hushed whispers, and delicate body language. It’s the sort of reality startup founders dread, if not consciously, then certainly subconsciously: the sort of thing that wakes them in the middle of the night with cold sweats, pounding hearts, and the sort of gripping fear laced with insecurity that paralyzes to the core.
If you can’t tell yet, Josh’s story really moved me. (Stop and go read his article now if you haven’t yet. Seriously.) Not just because of the transparency with which he wrote of Baremetric’s harrowing adventures (note: Twibble is a customer of Baremetrics, and we can’t recommend it strongly enough), but because of the degree to which it resonated with our own hardships at Twibble. Well, everything but the happy ending, anyway: we’re still somewhere in the second act of this three act story.
And that’s why I’m sharing our story with you, now.
How it all started
Twibble was an accident. An amazing, impossible blessing of an accident that we spun out of our previous startup, for which we’d raised $700k and that, roughly 18 months later, had failed in spectacular fashion.
Venturocket was a jobs/talent matching site. As desperation began to set in during its twilight months — so, Spring 2014 — our little team of four — myself, two co-founder engineers, and our designer—began fervently cranking out one idea after another.
While some were at least thematically related to Venturocket’s raison d’être (a sort of “Quora for crowd-sourced résumé feeback”) — others, er, weren’t (an anonymous QA site; you can guess how that went).
(Looking back on it today, it’s plain to see how and why Venturocket failed — too much, too hard, too complicated, etc; but we honestly still do not understand why our crowd-sourced résumé feedback product, CrowdVitae, failed. It really worked just like Quora: people uploaded PDFs, DOCs, or JPGs of their résumés, while others would grade them, and even leave constructive comments at the bottom. The more feedback you gave, the more you’d receive. And yet, somehow, it just never caught on.)
One of our myriad skunkworks projects, however, wasn’t actually intended to be a standalone project, it was meant to be tool for ourselves to use. Desperate to save Venturocket, we decided to try something new: we wanted to tweet all new job listings (and anonymously, all job seekers, too) from an RSS feed.
There was just one small problem: all existing services were either too expensive, too complicated, or just not pleasant to use: ugly UI, poor UX, or just plain boring.
So one fateful afternoon in April 2014, our designer Ryan Lum suggested, in no ambiguous terms, that we just build our own. As one engineer was busy working with CrowdVitae, the task fell to our other one, while Ryan Photoshopped together a mockup of the design.
Several days later, Twibble was born and we were using it internally, successfully tweeting all new job listings (with embedded company logos) and job seekers (not with photos, obviously).
We were so pleased with what we’d built, and it was proving so awesome for propagating our job listings and job seekers, that just hours later, we decided to launch Twibble publicly: we spent all of five minutes deciding between two competing product names, settled on Twibble, and shortly thereafter, purchased the twibble.io domain name.
Mostly just for shits and giggles.
We honestly had no idea whether it was going to catch on; we certainly couldn’t afford to market it; and in any event, how could such a niche product really grow sufficiently enough to maintain four guys’ salaries?
Well… it couldn’t.
Yes, we’d managed to get water.org and the UN Food Programme to try us out for a while (that was awesome); and we’d managed to impress the guys over at Buffer enough that they wrote the first public mention of Twibble, in July, just three months after we’d launched, and one month before activating Stripe to collect payments from customers. And, precisely one year later, to the day — hah! I just noticed that! — we were even mentioned on Forbes, in addition to several other smaller publications besides.
But still, it wasn’t enough.
By October 2014, Venturocket was forced to close shop. One by one, we started taking down all its services, cancelling subscriptions to the various ancillary services we’d been using, and finally, on or around Halloween 2014, shut down its servers for good.
But not before a last minute decision to move Twibble to its own dedicated dev servers, and Ryan and I agreed to keep the thing alive. Somehow.
Oct 2014 — Feb 2015: Twibble flies on autopilot, a veritable ghost ship adrift amongst the stars….
Ok, that was a needlessly dramatic section title. But it’s apt. And accurate.
Now is probably a good time to explain that from October 2014 until about February 2015, Twibble was completely on autopilot with no development and minimal maintenance (read: restarting servers): apart from a few generous offers to stabilize the site now and then, both engineers were long gone, and Twibble was little more than a ghost ship, miraculously able to stay afloat despite a dizzying array of holes that would make Swiss cheese jealous, and no crew to right or steady her course should anything go wrong.
What had kept us going during this time was that, incredibly, from the first day we flipped the switch on our Stripe account — August 24, 2014 — we had started seeing some cash flow. I mean, it wasn’t much, obviously, but … somehow, there was cash! (We weren’t ever able to say that about Venturocket.)
Finally, some time in February, we were lucky enough to start working with an incredible development team who were willing to jump on board for some equity and minimal pay. So we battened down the hatches, set royals and stunsails, and laid in a new course!
Our destination was somewhat vague — second star to the right, if I remember correctly — but the point is we were once again under way. More or less, anyway. “Now all we need[ed was] a little Energon, and a lot of luck,” to quote the great Optimus Prime. (You have seen The Transformers: The Movie from 1986, right?)
Feb 2015 — Feb 2016: Twibble holds her course… not much happens
So the first half of the chart below reflects what happened during this time: as you can see, it wasn’t much. (If that chart, and the subsequent ones below look vaguely familiar, it’s thanks to Josh’s Baremetrics.)
The sort of MRR growth you do NOT want to see. Dollar values excluded due to our embarrassment. :)
Yes, we were very proud of ourselves — and why not? In addition to Buffer and Forbes, several others like LifeHacker, SocialMediaToday, and ShoutMeLoud had all somehow discovered us too; no idea how. The point is, we were getting (some) decent press without even trying; nevertheless, by and large, growth was only slightly faster than tectonic drift.
(Interesting Aside #1: By the way, in case you’re wondering what that chasm is in April 2015, that’s when we basically had to take everything down for a few weeks and essentially rebuild Twibble from the ground up. In a word? ElasticSearch sucked. Don’t use it. Ever. Even if you’re an ES Jedi Knight, nobody else is, so you’ll never be able to find someone to help you, let alone, to take over. So, we migrated almost everything from ES to MySQL. That sucked.)
We had a few cool customers sign up during this time — Hallmark was a big one; Voice of America another — but sadly, we didn’t know how to monetize any of this correctly.
Our mistake was simple, even if the solution still eludes us today:
First, we (foolishly? naïvely?) thought our customer base would be a mix of individuals and companies. Now we realize our customer base is almost entirely corporate, with very few individuals at all. And so, just like that, we realized we were, for all intents and purposes, a B2B company. Who knew?
Second, and because of the first, we’d gotten our pricing all wrong: when a company like Hallmark is paying you just $15 per month, you’re an idiot. Clearly, we’re still an idiot: because while our new corporate plan dubbed Twibble Infinite starts at $99/month, well, it hasn’t done very well yet. So we need to work on that.
On the flip side, 2015 nevertheless ended up being the happiest year of my life as my wife and I were married in August. So, there was that, mercifully.
Febuary 2016: Twibble gets its dedicated, but not yet full-time engineer
In February I hopped on Turkish Airlines Flight 80 from SFO-IST, en route to Sofia, Bulgaria to meet with our development team, Dreamix, in a beautiful new 777–300; gorgeous plane. Bid my wife farewell, took off in an epic rainstorm (doing the rare east-bound departure on RWY 10L due to the shifted winds rather than the conventional 28R), and suffered brutally disabling jet lag during my impossibly short four night trip. (Note to self: never cross 10 timezones in less than a week again. It actually hurt; never felt more destroyed or fragile in my life.)
Arriving in Sofia, I met with Veliko Minkov who, I’d been informed, would soon be taking over exclusive duties of further Twibble maintenance and further development. We hit it off great, and he very quickly proved his superlative skills, effectively cranking out the work product of at least two-and-a-half engineers:
2016 was pretty cool. Relatively speaking.
Check out that small up-tick in February 2016: that’s when we did something altogether unthinkable, and raised prices on most of our customers by a staggering fifty per cent — with their permission.
It should be noted that his extraordinary work efficiency is a bit of a cheat, because, you see, he doesn’t actually sleep; rather, he takes little cat naps throughout a 24 period. So really, it’s not fair to compare him with anybody else. In case it isn’t abundantly clear, I’ve never been so impressed or honored to work with someone in my life. The guy is impossibly brilliant, and simply doesn’t believe in no-win scenarios. My favorite conversation with him:
“So Marc, [some thing we were working on] simply can’t be done. It’s impossible.” Pause. “But don’t worry, I’ve figured out a way we can do it.”
That was the first time he’d displayed is unwavering determination. It was hardly the last.
So if you’ll look again at that graph up above, you’ll note another (relatively) large uptick around October. Actually, the growth began in September when we rolled out an entirely new homepage (since updated yet again a couple weeks ago). I wish there were an easy way to show you the old versions of the site; they weren’t bad by any means, but, well, to assign numbers to what I’m talking about, we went from about 10% signup rates, to a whopping 20% recently.
The interesting thing, though, is how we went about building this new homepage. I’m not a developer; and as should hopefully be abundantly clear by now, there was no way Veliko could tackle both the never ending torrent of back-end development while simultaneously rebuilding our homepage. So in an act of desperation, I just said that I’d do it.
I’d used Squarespace to build our wedding website; a few sites for friends; and even a couple for myself in the past, and I’d grown quite competent with it, even if it was “the worst, best” CMS I’d ever used. After a week or so of work then, the New Twibble was live, almost immediately bringing with it a dramatic increase in signups; and again, a couple weeks ago, I updated it even more dramatically, thanks to a beautiful mockup provided by Ryan which I was able, more or less, to translate to Squarespace. It even has a cool video banner. Oh yeah.
(Incidentally, this is why you may have noticed twibble.io redirecting, not so inconspicuously, to home.twibble.io: our homepage lives on Squarespace, while signing up for and logging into Twibble puts you on our own RackSpace servers. It’s a hack job for sure, but hey, it works!)
So that was September 2016. Then October came round, and suddenly, we saw usage growth and revenues start surging. Surely this couldn’t really be entirely thanks to our new homepage?
Indeed, surely not.
This was perhaps the most bittersweet moment in Twibble’s young life, when we experienced, secondhand, what it must be like for a startup to close its doors prematurely: one of our competitors, the well known and respected Twitterfeed, announced that it was closing down. And so, just like that, we received the largest influx of new customers we’d ever received in October and November.
As I recounted here, Twitterfeed’s sole founder and engineer, mario menti, was extremely magnanimous and reached out most kindly after what I’d written; it wasn’t long after that he flew all the way from London to meet with me and talk about his exciting adventures, all the while advising and guiding me as best he could manage, and wishing us all the best. It was a touching, and deeply humbling moment for me. Not to mention, a terrifying one.
February 2017: Twibble gets its first full-time engineer
For most of the rest of 2016, things weren’t very exciting, except that we did land iHeartMedia — yes, the corporate entity behind beloved iHeartRadio — in November. So that was pretty awesome.
But as you can see from the graph above, we more or less started to plateau. This was pretty sucky for three main reasons:
Sucky reason #1: Our salaries remained sucky.
Sucky reason #2: We couldn’t hire other engineers (though, to Josh’s point in his article linked to above, perhaps that’s ok?)
Sucky reason #3: It’s exhausting. Working your ass off for so long — in my case, more than six years if I go back to the founding of our first startup, Venturocket — with marginal success takes its toll. But somehow, though I’ve been immensely frustrated at times, I’ve never really slumped into discouragement. And as my dad always cautioned me during my first year of UCLA whenever I bombed a final exam — a not altogether unusual event — the former is fine, while the latter should never be allowed into your psyche.
Now is probably a good time for another aside; but no, you don’t need to read this to maintain continuity. In fact, as far as continuity is concerned, you’d best not read it.
Somewhat interesting Aside #2, if you care about my weird background: I was supposed to be a lawyer after my absurdly Van Wilderesque, circuitous undergraduate years at UCLA which spanned astrophysics, math, chemistry, and bio, before finally settling on a BA in Econ. Actually, I am a lawyer. I graduated law school in 2008 and passed the July 2008 California bar exam on my first go. So what I mean, of course, is that I was supposed to be a practicing lawyer. Thing is though, I always knew I wanted to work with startups; so, I figured I’d do corporate law. That would have been cool; in fact, it would still be cool, if I’m honest. But when the recession hit in 2008, every job offer I had just vanished. Suddenly I was flopping about like a fish in a glass of chocolate milk, and my entrepreneurial life survival instincts kicked in: I tried to start my own law practice while applying to damn near every law firm between San Francisco and San Diego. (There’s probably still a few sheets of mostly dissolved résumés that blew out my car window driving onto an onramp to the 73 freeway down in OC.) And while neither endeavor — my law firm nor applying to jobs — was particularly fruitful — read: actually bore zero fruit whatsoever — it did catalyze the first thoughts of what eventually coalesced in Venturocket: surely there had to be a better way to connect job seekers with jobs relevant for their skills? Indeed, surely there must! And thus Venturocket was born in January 2011 after two months of development; and the rest, as they say, is history….
So then. Finally, after a year of mostly intermittent yet, by and large, mostly successful development — read: net gains — Veliko was at last ready and able to join Twibble as its first full-time engineer.
Think about that for a second; like, seriously, let that sink in. Twibble first went live in April 2014, and we flipped the switch on collecting payments via Stripe in August 2014. Between October 2014 and February 2015 then, Twibble had no engineers. None. It was completely, entirely, floating — more or less — adrift amongst the interwebs. From February 2015 till February 2017, we finally landed some engineering talent that performed arguably the most crucial upgrades to Twibble’s systems thus far; but it wasn’t until just last month that we finally — finally! — had a full-time, completely dedicated engineer. So that’s a staggering three years, nearly, this April, that Twibble was held together by little more than a patchwork of string, glue, and bits of hope and fairy dust.
March 2017 and beyond…
So now here’s that graph again, but this time, fully annotated, with daily resolution rather than monthly:
Yup. It’s been an adventure. Full of curvy arrows.
So here we are then. Present day, when we’re now eagerly working towards the imminent launch of the most-requested feature of all, Facebook support to complement our RSS-to-Twitter functionality. At last.
This is the point at which you’re probably expecting a happy, uplifting, inspiring conclusion to this otherwise heretofore dreadfully dreary, depressing, and altogether uninspiring story. Like, perhaps we suddenly started surging with customer and revenue growth two weeks ago. Or we landed a $5M Series A round.
No. Afraid not.
But how about this, then: I made a point, earlier, to explain that I was not, in fact, discouraged by any of this. Indeed, this has been the most astonishing experience and challenge of my life, and frankly, I couldn’t be more proud of what we’ve accomplished, with so little, and against all odds. That we were able to salvage Twibble from the burning, charred husk of what was once Venturocket; that we were able to get cashflow positive from day one; that we continue to push, persevere, and persist in our mission; that we usually score #1 on Google when searching “rss to twitter;” and that we’ve managed to score such large, respected customers; for me to be anything but proud, would be a deeply insulting disgrace to everything for which we’ve worked so hard these past three years, and for which, I continue to be most grateful.
That, and the tremendous humility and honor with which I live every day knowing that Veliko took a leap of faith to join me on this harrowing adventure, that is something for which I will forever be grateful. Well, that, and having the sincere privilege and good fortune to now call him “my friend.” That alone has made this entire experience worth more than any surge in MRR we could ever possibly hope for, more than any chunk of VC investment.
So there’s your happy, uplifting, and, dare I say, inspiring conclusion. Not dollar signs. Not hockey stick growth. No feel good retrospective of a plane narrowly pulling out of a nosedive just feet above the ground. No grand reveal that, as of last month, we finally managed to right the ship and start growing, no longer carrying a $1200 monthly balance in the bank.
Nope. We’re still struggling, trying to find our footing, and we’re still carrying that $1200 monthly balance.
We’re also carrying hope. Lots of it.
Check back with me again in a few months… maybe then you’ll get your Hollywood ending.
Meanwhile, let me know in the comments if you’re interested in joining our team and want the unique satisfaction of helping us to grow Twibble to the next stage.