How We Bootstrapped a Weather Startup
The following is a reprint of my 2010 Hacker Monthly article “Bootstrapping Stormpulse,” which disappeared off the Internet in 2013. I thought it worth re-circulating as the discussion around bootstrapping vs. VC has intensified in recent months, and Stormpulse was the story of both.
You want to start a startup? Great. Bootstrap it? Even better! You are here:
Unfortunately, unlike the path our hero takes from left to right, there’s no formula for how you’re going to get to the princess. By definition, bootstrapping is an any-way-you-can-make-it-happen adventure. And here’s what it feels like:
The platform below you is about to give way! Will you jump in time? Will you land on that little platform? Will you slip? Unfortunately you really don’t have time to calculate all of that. You have to jump now. Did you make it?
I will add: I’m not always thinking “oh man, I eat risk for breakfast! I’m so made for this!” I may have been born with the gears arranged so that when I see something that most people would turn away from, I jump in, but when I get halfway through the air I am still wondering how things are going to land. And beneath it all, I have to say that if I weren’t able to do this in faith that it’s all for a higher purpose in my life, in my character, in my family, in my soul, then I wouldn’t be doing it at all.
I figured it would only take a few weeks, maybe a month or two, and then die, just like all of the other small projects I’ve dabbled in.
A little over 2,200 days ago …
- July 2004. My first child, a daughter, is born.
- September 21st, 2004. Stormpulse germinates in my brain as an organic startup idea:
- Dismayed by the havoc of the 2004 hurricane season and eager to find an idea for a web project, I write: “I’m tired of the cartoon renderings that currently pass as meteorological forecasts and have embarked on a journey to bring information rich interactive displays using hurricane data to the general public.” I figured it would only take a few weeks, maybe a month or two, and then die, just like all of the other small projects I’ve dabbled in.
2004 totals: ~300 hours, $0.00 invested -> $0.00 revenue
- 2005. Still fascinated by this web project I’ve started, I listen to Tom Coates’ “Web of Data” talk at the Future of Web Apps London, which serves as the catalyst for my thinking about what would become Stormpulse’s reason for existence and its design philosophy. I listen to it many times while commuting with my iPod. It sinks deep. Lots of synapses firing around weather data.
- Summer 2005. The most active hurricane season in recorded history. Katrina leaves an indelible mark on the consciousness of hurricane country natives.
2005 totals: ~600 additional hours, $0.00 invested -> $0.00 revenue
- August 2006. I leave my software developer job at McMaster-Carr to move to Florida. Stormpulse, a directory of PHP scripts for scraping data, is the only thing I’ve got going. At this point it’s a bunch of PHP code I’m converting to Python (TurboGears at the time). No Flash component (map) yet. Brad (my manager at McMaster) decides to become a seed investor and my co-founder while remaining at McMaster for now. His seed investment will be used to pay my family’s living expenses for the next year, and remains our largest single cash infusion to this day (roughly $50k out of a total $75k).
2006 totals: ~1,000 additional hours, $50k invested -> $0.00 revenue
- February 2007. Brad leaves his job at McMaster-Carr to work on Stormpulse full-time.
- March 24th, 2007. Brad and I go west to Startup School 2007! Distinct memories: meeting Jessica Livingston (when I first arrived at YC the day before, she asked “Oh no. Are you going to ask if you can hang out here all day?!” [this was funny, no offense taken]), Drew Houston, Justin Kan, Kyle Vogt, Sam Odio, Trevor Blackwell, and Barry Hess (who?!). Inspired by the whole experience. Paul Bucheit’s talk was great. Got to tell him I didn’t like Gmail (ironically I now have 6 Gmail accounts).
- April 25th 2007. Our first 42 visitors. Having finished the first cut of the interactive map (from scratch, no third-party API’s) and main page design, we begin emailing our first demographic — the cutting edge of hurricane tracking and research — NASA, NOAA, et al. The site crashes as a couple dozen people try to access it simultaneously. I learn what it means to close database connections and gladly hand the CTO reins to Brad.
- May 16th, 2007. We attend the Governor’s Hurricane Conference in Ft. Lauderdale, FL. On a shoestring budget we bring a 30” LCD monitor, a laptop, and a few fliers and draw some meager attention to ourselves in the weather industry. Spending lots of time devouring all of Paul Graham’s writings.
- June 25th, 2007. We post a link to Stormpulse on News.YC and get solid, useful criticism on our need to simplify the presentation for the average Joe and also make the site year-round useful. We would spend the next three years doing just that.
- Spring 2007. Turned down by YCombinator for Summer 07. Realize that living in South Florida is a funding problem, but I’m resolved not to relocate.
- June 2007. Initial funds run dry and it’s time to get a day job again. I start working at The Palm Beach Post, which is ironic, given their position as a competitor to Stormpulse (in reality, it wasn’t yet much of a competition with Stormpulse being totally unknown to 99.99999% of the market). What would I learn on the inside? Meanwhile, I learn and begin to adore Django. I email Adrian Holovaty to tell him about Stormpulse and that if I had to do it all over again, I’d use Django. Looking back, I’m really glad we didn’t.
- Summer 2007. Hunting for revenue, Brad and I put together a proposal for the Post to exclusively license our tracking map for embedding on their website. It would have been a $2–3k per month deal. Better than closing up shop, right? Deal fizzles as Editor shows no interest (presumably due to price).
- Winter 2007. With traffic still flat (<100 visits per day on average), I’m pretty close to thinking Stormpulse is dead. Brad starts dabbling in other projects while I step away from Stormpulse and plug away learning Django and geospatial databases at the Post …
2007 totals: 3,000+ additional hours -> 37,640 visits, 18,233 unique visitors, $0.00 revenue.
- February 2008. My second child, a son, is born.
- Spring 2008. We continue making small changes to the site, making it less and less intimidating/confusing (moving away from weather nerd market and focusing more on the mainstream type of user). Starting to serious apply the mantra of “engage beginners while attracting experts.”
- July 2008. EUREKA! What if we offered Stormpulse as a freely-embeddable widget? I pitch the free idea to my managing editor. He says ‘yes!’ (Maybe hearing Chris Anderson at Startup School 2007 was having some kind of lasting effect on us?).
- July 19th, 2008. The Post embeds our map, which means instant brand exposure to tens of thousands of people. We think this could be big.
- August 2008. Dozens of major newspapers and radio stations embed our hurricane-tracking widget, which fits the “ten times better” requirement versus existing solutions. And it’s FREE! All we ask in exchange is that sites link back to Stormpulse as a “hurricane-tracking” site. Major SEO boost ensues.
- August 15th — September 15th, 2008. The Atlantic hurricane season catches fire and our site gets visited 11.2 million times by 3.3 million people (without crashing — thanks Amazon Web Services!). 58% of this traffic is direct, and the top two searches from Google are “stormpulse” and “storm pulse”. In a lazy effort to monetize we add a Tipjoy widget and later a PayPal “Donate” button the site. Tipjoy widget gets some press. PayPal calls to find out what we’re up to (tripped some red flag somewhere apparently).
- September 2nd, 2008, 1:48pm ET. I get an email from a weather producer at CNN just in time to get up from my cube and walk over to a flatscreen where I watch the CNN meteorologist using Stormpulse on live TV. About as surreal as it gets, really.
- September 5th, 2008. My co-worker at the Post tells me that Stormpulse is on TechCrunch. TC referrals make very little marginal difference to our traffic at this point.
- September 18th, 2008. Tim O’Reilly shows Stormpulse on stage as an example of “what might be Web 3.0” at the Web2.0 Expo in NYC. This is getting weird.
- October 2008. The former founder of a telecom company and current Palm Beach billionaire resident courts Stormpulse after Hurricane Ike brings the site to his attention. The idea of a strategic investment and partnership falls apart when we learn he wants 51% of the company in exchange for guaranteed jobs to work on his current financial project. It was educational, but no thank you.
2008 totals: 2,500+ additional hours -> 13.6 million visits, 4.0 million unique visitors, $125.00 AdSense revenue, ~$2k PayPal donations.
- January 2009. I watch the inauguration of Barack Obama with an attendance of 4 million people. It dawns on me that the crowd I’m seeing on CNN is the same number of unique people that have used Stormpulse this year. Mind blown!
- February 24th, 2009. A strategic investor in the weather industry proposes an arrangement wherein they would own 51% of Stormpulse. We consider the idea but place a much higher value on the company than the potential investor and the deal is preemptively killed. We learn that 51% is a popular equity amount among ‘strategic’ investors this early in the game.
- May 2009. I quit my day job to work on Stormpulse full-time and I give the Post my two weeks notice. Before my exit, I discuss the possibility of COX Enterprises investing in Stormpulse, but no agreement is reached.
- June 2009. While feverishly working on premium account features to be launched ASAP, I learn to sell, sell, sell … advertising on Stormpulse on a CPM or CPC basis, which keeps us afloat. All of the leads are coming to our inbox without any prompting, which is … nice.
- July 11th, 2009. We become legitimately freemium by launching our Pro accounts, originally under the name of “Stormpulse Advanced” (now simply Stormpulse Pro). We sell a couple dozen on the first day, which gets us excited. Original price was $3.95 per month or $19.95 for the hurricane season. Most sales were the latter. We have cashflow!
- September 2009. We pitch Stormpulse to a Chicago VC. Our conversion rate is too low for his tastes.
- October 2009. I pitch Stormpulse to the investment arm of a gigantic media enterprise in NYC and find out we’re too early for them. Should have known better going in, but it was educational. Investor said that if I could come back with monthly uniques of say, 100k, it might be interesting. I tell him we already have that, but that doesn’t extend the conversation.
- December 1st, 2009. The 2009 hurricane season officially comes to a close as THE QUIETEST STORM SEASON IN 12 YEARS (interesting timing).
- Late 2009. Looking for funding, we find out that the few angel investors we had hoped would come through … weren’t going to. Stormpulse opportunity doesn’t make sense to them. Why not?! It makes sense to us! Oh well …
2009 totals: 4,000+ hours -> 6.7 million visits, 1.8 million unique visitors, 25% of the way to year-round break-even cost vs. revenue.
In other words …
- January 2010. I test the waters and start looking for contract work. The ideal contract gig (solid hourly pay, remote OK, weather-related [?! — yes, seriously]) opens up and then after three rounds of interviews with glowing endorsements … falls through. This cycle then happens with two more contract opportunities. I’m left with just Stormpulse. Again. Lots of existential questions coming to the surface: What do I think I’m doing and who am I?!
- February 2010. We get introduced to a very well-known early stage VC. Top three on your wishlist kind of thing. Amazing. After a couple phone calls and a face-to-face meeting, we are told via email that we’re getting a seed stage term sheet this week (YES!!!!). Then it becomes the following week. And then the following. And then, after another video conference, it becomes “sorry, but nevermind” (NO!). The main problem: our disparate geography (did I neglect to mention, dear reader, that we’ve built Stormpulse with Brad living in Chicago and me living in South Florida?). The most difficult let down yet. I really thought this was going to pay my/our bills.
- February 2010. We scramble to find a handful of family members willing to invest just enough to stay afloat another few weeks. Brad antes up a final time (at least WE believe in us!). I find out my third child is going to in fact be twins.
- February 2nd, 2010. Using up our funds, we broaden the site to include year-round weather. Some might call this a pivot.
- April 2nd, 2010. With no income, no insurance, three dependents, and twins on the way, I pray for some idea of what to do. The answer I hear is “stop asking for money and you do it.” That’s all I needed to hear. I put the pedal to the metal, and in a major flood of inspiration I stay up all night and manage to design and build a high-resolution, animated radar feature (#1 request at the time) and we add it to our Pro accounts. I email 3,980 registered users and go to bed.
- April 3rd, 2010. I do the math and figure we need to sell an average of at least 4 Pro accounts per day to pay the bills. Incredibly, the Super Radar feature starts driving subscriptions at this rate or better. We’re alive again!
- April 2010. My family temporarily relocates to Atlanta, GA for the birth of the twins. I dive in to the Atlanta tech community and meet Paul Freet, Ben Hill, and the awesome fellows at Georgia Tech / ATDC while attending a couple of Bootstrapper Circle meetings on Wednesday mornings. Meeting other entrepreneurs is a breath of fresh air at the right time.
- April 2010. During a Bootstrapping Circle I mention our business and the fact that the Weather Channel is nearby (I’m secretly hoping for a meeting). It just so happens that one of the founders of the Weather Channel is across the street at the B&N at GaTech. We’re introduced and I have 90 seconds to pitch Stormpulse. A few days later I learn that he’s not really interested in the weather sector anymore.
- May 2010. We pitch Stormpulse to a very well-known early stage VC entrepreneur-in-residence in NYC. No follow-up/interest at this stage.
- June 2010. We completely redesign our product pages and introduce B2B-level plans. We sell one of each package (trying to validate interest and pricing) before launching these plans to the public.
- June 2010 — August 2010. Self-service B2B accounts gaining traction with emergency managers, government agencies, and Fortune 500 companies. Running a single graphical ad on the site with the occasional premium direct sale.
- August 2010. Send in application to AngelList. No bites. I’m pretty sure our nearly empty “Social Proof” form field didn’t help.
- August 2010. A North Carolina millionaire businessman contacts us about introducing us to enterprise clients of his. In exchange for selling our service for an enterprise-worthy price, he would like to split the business in half and own 51% of the enterprise corp. What is it with these people and 51%? We decide we’ve come too far to stop now. No deal.
2010 YTD totals: 3,000+ hours -> 13 million visits, 4 million unique visitors, 3x revenue increase over 2009.
So what now?
We’re taking a step back and seriously evaluating our value proposition in light of what we’ve learned in the last 3 years.
We’re being deliberate about our pricing so we can continue to hold this tiger by the tail.
A couple weeks ago, another founder that’s just heard our story told me he’s shocked at how backwards we’ve done things. I immediately took it as a compliment (which is what he intended). I guess the norm is: team (which often means funding) first, product second.
But I like the way we did it. Don’t wait around. Solve a real problem with a simple product, now. Then talk to others that have the same problem (these people exist, right?) so you can correctly estimate the opportunity. I think there’s a tendency to overestimate existing solutions (competition) and therefore underestimate the opportunity. I never thought that first silly PHP page could become a real business.
The only thing I regret is that we took so long to understand and establish pricing. So in closing may I suggest:
If you’re going to bootstrap:
Product first, value-based pricing second.
And I really think the rest will follow.
Original comments on HackerNews.