By Andrei Soroker
Table of Contents
Chapter 1: Beauty Pageant
Chapter 2: Segment for Chat
Chapter 3: Money and Pivot
Chapter 4: Build, Launch, LAUNCH
Chapter 5: Go-to-Market Strategies
Chapter 6: Go-to-Market Strategies: Results
Chapter 7: SEO and Startup Goldilocks Zone
Chapter 8: Zero to 140 Paying Customers In Ten Months
Chapter 9: Shifting Out of First Gear
Startups tend to operate in two parallel worlds. One is where they're crushing it, and the other one is reality. Sometimes they overlap, but that's pretty rare. Unless you're friends with the founders, true stories are few and far between.
2015 was a really interesting year for our startup. I think I'm going to go ahead and call it a good year. We pivoted, built a product from scratch, and are now teaching it how to walk. It's standing up and taking first steps. These are happy days!
While the days are happy, I want to share our story.
Almost a year ago, my co-founder Peter and I went on a few multi-hour walks in and around Palo Alto. We were coming to terms with the failure of our product and the need to pivot.
Our product was Kato—a direct Slack competitor. We spent two years and 1.6 million dollars building it. Our last-ditch effort—a SAML and SCIM-enabled enterprise version wasn't getting much interest, and we were running out of money.
After spending so much time trying to convince companies to use our chat service, we came to the realization that the world of chat is not moving toward unification. Instead, it's getting fragmented, and this fragmentation is accelerating.
Microsoft, Facebook, and Google all offer workspace chat services. Cisco now has two: Jabber and Spark. Atlassian (HipChat parent) is relatively small, but newly-public and poised to survive the apocalypse. Slack just announced a 80M fund.
That's just the visible tip of the iceberg.
Under the surface, there's a vicious battle for a piece of the remaining pie (and some fresh air—venture money) among the ever-increasing cadre of the small and unseen.
Symphony is actually building its own iceberg, nearby—it raised 160M in a bold effort to disrupt the Bloomberg Terminal (among other things).
Kato was a good—great, even—chat service, with a ton of unique and thoughtful features. It was reliable, fast, and to the point. But, it was a participant in a rigged, overcrowded beauty pageant. And to hell with that—we had to pivot.
It was Christmas, the sky was grey, light draught-drizzle was getting stuck in our beards. We were walking from Palo Alto to Atherton.
"Let's build Segment for chat," one of us said, "no UI!"
We brainstormed for hours and eventually came up with a model that seemed workable.
Sameroom is a no-brainer of an idea. It just appears to be impossible to build.
And it is, almost.
The next day we met again, this time with a small whiteboard. We set it up on a park bench. Diagrams were drawn and photos taken. We had a plan.
Having a plan is great, but we also needed money.
We asked Brad Feld, our lead investor, permission to pivot. Permission was granted, and we raised a 2M inside round to build Sameroom. Brad said: "you know how to build very complex technology that can stand up, while being very money-efficient. Let's give it another go."
Foundry Group is one of the best venture capital firms in the world and we're lucky, proud, and grateful to have their support.
We decided that Peter would start with the simplest possible task: connect a Slack channel and a Kato room, so that two teams can chat with each other without leaving their homebase. Once Peter got to work, he pivoted.
I still had to lay off two-thirds of the company.
This sucked, but it comes with the job. If you're a firefighter, sometimes you have to willingly enter a burning building. If you're a CEO, sometimes you have to fire two-thirds of your staff.
We build quickly. Sameroom went into production six weeks following Peter's initial commits. Here they are (
git log --reverse, so top is oldest):
On Sunday, March 1, we flipped the switch. Sameroom supported Kato, Slack, HipChat, IRC, and Campfire. Peter and I stood at a LAUNCH conference booth with this ridiculous poster:
The few people who stopped by were mostly confused. HipChat—the conference sponsor—was sluggish to the point of ruining our demos.
Peter at our LAUNCH conference table on the day we launched Sameroom
On March 4, we ended up on Product Hunt—this was our initial wave of traffic and it brought our first paying customer: a development consultancy that signed up for three licenses (all still active!).
Once we launched, we had to figure out how to get people to try the product. Distribution is the hardest thing to crack. Most first-time founders don't have an instinctive understanding of this—it develops over many months of despair.
We started off by making two assumptions: First, we wouldn't be able to get press. Second, Slack, HipChat, and other "partners" would not tell their customers about us any time soon. Any press, marketing help from "partners", or successful growth hacks, were they to happen, would be treated as a bonus.
We decided to try five go-to-market strategies simulaneously. (This was also when I finally understood the meaning of the term "go-to-market startegy".)
First, we opted to sponsor technical conferences. In exchange, we would ask conference organizers to offer a Sameroom Portal as a way for people to communicate for the duration of a conference. This way, we figured, conference attendees would discover Sameroom, and—hopefully—take it back home with them.
Second, we assumed that our most likely target audience will be the various agencies and consultancies—companies that have to deal with a fair amount of external communication. We planned on reaching out to them over email.
Third, we wanted to probe the co-working space market. Nearly all co-working spaces sell their "super awesome community" as a feature, so we figured we could help them structure communication in a simple way with Sameroom. We'd reach them over email as well.
Fourth, we decided to place ads on Twitter and Spiceworks.
Lastly, we planned on writing a blog post for every integration that was possible on our platform.
We ended up sponsoring five different conferences. All but one—NodeConf in New Zealand—were a failure, for a couple of reasons.
First, we couldn't get the organizers to fully understand what they had to do to support us, or how our system works. All we needed, it seemed, was five minutes of uniterrupted conversation with them, but those five minutes don't exist if you're putting together a conference.
Second, people don't actually want to instant-message at a conference. They socialize. They work. They're hungover. And some, I assume, are paying attention to the talks. We didn't land a single customer from this effort, but the experience was invaluable.
To reach agencies, we built a part-automated, part human-driven lead-mining machine. Over a four-month period, we sent about 9000 cold emails. It's difficult to put the magnitude of failure of this approach into words, as it approaches infinity: we had two signups and zero paying customers in the end.
We built a very similar lead-mining machine for co-working spaces, but on a much smaller scale. The intermediate results were way better than with agencies—great open and response rates, lots of calls. In the end, however, no signups or customers. Most co-working spaces operate on extremely low margins, buy virtually no software, and, in general, struggle along.
Paid advertising was a total flop (one confused signup, zero paying customers), but it lead us to a much deeper understanding of why paid advertising for something like Sameroom would never work.
Imagine, for a moment, that you have a startup that offers an on-demand emergency stain removal service.
Would low-budget paid advertising work? Of course not: very few (close to zero) people need emergency stain removal while browsing through Twitter, Facebook, or Spiceworks. But, there are definitely people who need an emergency stain removal service, now.
How would they find you in their time of need? They would search, for example, for "emergency stain removal san francisco". As long as you're well-represented in search results, they'll be calling. If there is exactly one on-demand emergency stain removal service in the world, you're golden.
This brings us to our final go-to-market strategy: blog posts and the mythical SEO.
As soon as we began publishing blog posts, we started seeing organic traffic. Not a lot, but incredibly targeted—we were catching all those looking for an emergency on-demand stain removal service. And we were the only game in town.
They'd search for a "slack hipchat integration" on Google, skim our blog, sign up, hit our limit, and subscribe.
Writing (and re-writing, constantly) these blog posts is painful, but the strategy totally works. Here's a graph or daily organic visits to sameroom.io from March 1 to now (December 17, 2015).
The increase in December followed Sameroom's appearance in the Slack AppStore and the associated bump in PageRank (this doubled signups).
Peter began describing Sameroom as a startup in a "Goldilocks SEO Zone". All we had to do to get to the (roughly) #1 spot on Google was to write a blog post. We didn't have to engage in a cashpile competition for keywords with other startups. Nobody else wants those keywords.
With each new integration, we add approximately N blog posts, where N is the number of services we support. To track our progress, we use this Google Spreadsheet:
I spent considerable time looking for writing help and trying out different writers, but in the end ended up writing almost everything myself.
Since nearly 100% of our revenue comes from organic signups, I consider writing our blog posts at the top of my list of priorities.
If you're a firefighter, sometimes you have to willingly enter a burning building. If you're a CEO, sometimes you have to write and re-write dozens of blog posts.
140 is not a huge number, and it certainly doesn't translate to interesting recurring revenue, in our case.
But, I think it's kind of amazing that we got here fair and square. We built some technology from scratch for a market that didn't exist. Then followed a concerted effort of trying a variety of go-to-market strategies, identified one that seemed to work, and executed against it.
Our organic, self-serve customers unfortunately aren't going to turn Sameroom into a viable business—there's not enough time for that.
However, they've helped us battle-test the service and discover a very interesting new frontier: big companies where, for a combination of reasons, the abundance of chat services deployed simultaneously creates serious organizational challenges. We can help them.
We're gaining steam. Our small engineering team is absolutely world-class. We have influenced the APIs of Slack, Intercom, HipChat, and Fleep. Instead of discovering new competitors, we're making new friends. We're getting better and faster at building new integrations.
This week, we released two new integrations: with Microsoft Lync 2013 (or Skype for Business) and Cisco Jabber. (They are so new, we haven't announced them yet.)
Both integrations mark the beginning of a new chapter for us, since they both require our customers to install an on-premise software package (the Sameroom Proxy). And we all know that only big companies install on-premise software packages.
The proxy connects on-premise chat services to Sameroom Cloud, which, in turn, connects them to cloud messengers or on-premise chat services behind other firewalls.
These new integrations and our Chatter gateway are uncovering a brave new world both for us and for chat interoperability in the enterprise.
We're excited to start 2016 not from scratch. We're excited for ongoing proof of concept projects with great companies. We're excited to be the sole participant in a beauty pageant with a tremendous audience.
Happy new year!