24 Months to the First Yes

Every founder remembers their first yes. Not the first meeting, not the first pilot, not the first press release. The first time a customer signs a document that says your thing is worth actual money.

Ours took 24 months. It took three proofs of concept, five cities, an eight-hour meeting that ended with the one person who controlled the budget leaving for the hospital, an incumbent vendor ten times our size working to bury us, three months of contract line reading, and then, with the finish line in sight, four months of total silence. I have started a company since, and I will probably start more, and I can tell you that nothing I have done professionally taught me more than the two years I am about to describe.

The customer was Ma Bell. You know the one: the century-old telecom giant whose name I will leave to your imagination, out of the same respect that keeps every other name out of this story. The company was Cask, the startup I co-founded, back when it was still called Continuuity. Out of respect for the many good people on the other side of the table, I am not going to name any of them, or any of their internal projects. They know who they are, and some of them became friends for life. But the story itself deserves to be told, because somewhere out there is a founder in month nine of their own first pursuit, wondering if they are failing. You are probably not. This is just what it looks like.

It started with an invitation we did not earn so much as inherit. Our investor, Andreessen Horowitz, held a briefing for a visiting delegation from Ma Bell and gave its portfolio companies a few minutes each to present. We were a tiny data-infrastructure startup with a big idea: that building applications on Hadoop should not require a priesthood, and that an abstraction layer, the thing that became CDAP, could let ordinary engineers build real-time data applications without mastering the whole zoo underneath.

In that room was a senior leader from their research organization who saw something in us. I want to be careful here, because without this one person there is no story. Every enterprise deal you have ever read about has someone like him inside the building: a champion who decides, long before the paperwork agrees, that you are right. He scheduled a follow-up demo. Then he kept scheduling things. His tenacity got us our first meeting with the executive who ran their newly formed big data organization, at the Foundry in Palo Alto. Years later he told us, in a readout meeting, that he had known almost from the beginning we were the right fit. Champions see early. That is what makes them champions.

Your first customer is never a company. It is one believer inside a company, spending their credibility on you before you have earned it.

The green light we got was not a deal. It was permission to prove ourselves, and enterprises do not ask you to prove yourself once.

The first proof of concept was a Twitter sentiment application that had to stand up to the full firehose, the entire Twitter stream, in real time. At that point we barely had product to show: CDAP, our Flow engine, our dataset layer, and a lot of conviction. The contractor assigned to evaluate us was openly skeptical, which I have since learned is the best kind of evaluator, because skeptics who convert become your loudest internal advocates. Jon and I spent hours with him. We built the application, handed it over, and let him scale it and break it in their environment. It held. He converted. He became the project's champion inside their walls, which mattered far more than the benchmark.

They came back wanting more, which is what success looks like in an enterprise evaluation: the reward for passing a test is a harder test. The second proof was a real-time geo-fencing application with recommendations, built largely by their own engineer to test whether our platform was usable without us hovering. He got most of the way alone, then asked for help. Jon built the dataset layer for the geo-fencing, I worked with him on the processing, and it shipped. The third proof took Jon to New Jersey to integrate R into the platform, because their scientists lived in R and a platform that ignored that reality was a toy to them.

Then came the readout call: Jon, Tom, and me on our side, the research team on theirs. They walked through the results and called the engagement a success. We were one year in. And here is the sentence I need every founder to sit with: after a year of flawless technical execution, we had revenue of exactly zero dollars from this account, because the people who evaluate technology and the people who buy technology are almost never the same people.

The person who controlled the budget was a legendary figure inside the company, and the path to him ran through a gatekeeper: an architect who styled himself the great man's right hand. Tom and I flew to St Louis to win him over.

I will tell this one honestly, because the sanitized version teaches nothing. We were made to wait. In lobbies, in empty conference rooms, for stretches long enough to memorize the carpet. When we finally got our audience, day one was not a conversation, it was a deposition. We spent almost the entire day defending the premise that an abstraction layer should exist at all. Not our abstraction. Any abstraction. There is a particular kind of exhaustion that comes from justifying your company's reason for existing to someone who has decided, before you spoke, that you should not.

That night, in the hotel, Tom and I made a decision that changed the trajectory of the company: stop arguing, start showing. The next morning we asked for one more session and ran the whole thing live, from nothing. We provisioned a cluster in front of him using Coopr, which at that stage was held together with shell scripts and Derek's magic, installed CDAP onto it, deployed a real application, and ran it, end to end, while he watched.

The room changed. There was, suddenly, excitement in that gloomy office. And here is the part that still makes me smile: the excitement was not for CDAP, the product we had flown there to sell. It was for Coopr, the cluster provisioning tool we considered internal plumbing. He did not care about our beautiful abstraction. He cared that we made a cluster appear out of thin air. We told him, honestly, that it was not even a product yet. It became one, in real part because of that meeting.

We flew to St Louis to sell the product we loved. The market bought the tool we had been embarrassed to show. Listen to the room, not to your roadmap.

What followed was the long middle, the part that never makes it into founding mythology because there is nothing cinematic about it.

There was the big Dallas meeting, seven or eight hours in a room with every stakeholder we needed, where the executive whose budget we were chasing had to leave early because he was unwell. We learned later he had been hospitalized. All that orchestration, and the one decision maker in the room was gone by lunch, and you feel terrible for feeling sorry for yourself because a man is in a hospital and you are worried about a software deal. Founding a company does not exempt you from being human, and it does not exempt anyone else either.

There were the months designing their new cloud platform alongside their architects, my phone buzzing at all hours with questions from the gatekeeper who had become, improbably, my most demanding collaborator. There were six months of nonstop trips to San Francisco for yet another proof of concept, through political crosswinds, technical roadblocks, and an evaluation lead who left the company halfway through. There was the stretch where the account went quiet and Tom and I got on planes and did a lap of the country, St Louis, San Francisco, Dallas, New Jersey, Atlanta, for no reason more sophisticated than to stand in front of people and repeat our value proposition until it stuck. We joked we were one badge short of being employees.

And there was the incumbent. They had chosen a major Hadoop vendor whose CTO had a direct line to the executives we were courting, and that vendor saw us, correctly, as a threat. Every technical choice we had made was suddenly a question raised in a meeting we were not in. We tried, more than once, to make peace and find a way to work alongside them, and were pushed away every time. So we did the only thing a small company can do: we answered every objection, in writing, in person, with receipts, and we kept every stakeholder in the loop with a transparency that bordered on oversharing. When you are small, your only structural advantages are speed and trust. We spent both freely.

The best move of that whole stretch was not defensive. In a meeting at their New Jersey office, Tom floated an idea: take our stream-processing engine and a streaming technology their researchers wanted to open source, and merge them into a single open source project. Their side lit up. Engineers from both companies actually built it, and we announced Tigon together. It gave our champion visibility inside his company, and it told everyone watching that we were there to build with them, not just sell to them. Nothing you say in a pitch earns trust like shipping something together.

Eighteen months in, it came down to two architects, the last technical minds we had to win. One was nearly with us already. The other was the hardest kind of smart, the kind that does not accept a conclusion until he has personally re-derived it from first principles.

Andreas, Tom, and I went into a two-day session with them that I can only describe as an excruciating, thorough, professional grilling. Every layer of the product, every design decision, every failure mode. Midway through the second day, I watched it happen: the harder of the two, in a striped blue shirt, arms folded, stopped cross-examining, went quiet for a moment, and smiled. An actual aha. The questions changed in that instant from "justify yourself" to "help us understand where this goes." I have closed bigger deals since. I have never felt anything quite like watching that particular smile arrive.

Jon and I met the budget executive soon after. He confirmed what his architects had told him and gave the nod to move to purchasing. Twenty months of work, and the yes, when it finally came, took about a minute.

If you have never sold to a giant company, you may think the yes is the end. The yes is when the second gauntlet starts.

Procurement took three months of haggling. Why do you charge this way. Why should we pay for nodes you do not run on. Jon and Tom pushed the pricing conversation all the way to an enterprise-wide license, which for a startup our size was an almost absurd outcome, and then the three of us spent our evenings reading every line of every agreement, and I mean every line, because we kept catching things. The contract structure sprawled to cover internal use, external use, both products, every combination, artifacts of a cloud initiative of theirs that was cancelled mid-procurement. We left it all in the paperwork anyway, because you never know what comes back. They also wrote the deal assuming we were open source, which we were not, yet. We promised to open source within a year, and we kept that promise.

Then, with the master agreement signed on our side and sitting on a desk somewhere on theirs, their corporate development group decided to review the deal and consider whether they wanted to invest in us. And everything stopped. For nearly four months, nothing. No signature, no timeline, no answers. We were inches from the line, fully extended, a startup burning venture money against the silence of a corporation that measures time in fiscal years. Those were the low days. There were evenings I could not have told you whether we had a company.

We accepted some of their terms. We held on others. And one day, roughly 24 months after a borrowed conference room at our investor's office, the signature came back. We had our first paying customer.

Tom had a running joke with two of their leaders that when the deal closed, he would personally drive Philly cheesesteaks to New Jersey. So he loaded up his Denali with cheesesteaks and drove them to the office, and we spent the day not celebrating so much as planning what we would build together next. That is the truest picture of enterprise sales I own: two years of grind, settled over sandwiches somebody drove across state lines because a promise is a promise.

An epilogue first, because it carries its own lesson. After all that, after the enterprise-wide license, they spent almost a year and a half struggling to find the right first use case. I did deep dive after deep dive across the country, sometimes with the same people twice, folding everything we learned back into the product. Eventually they started building a data lake and used CDAP for ingestion, and from that beachhead the platform spread into project after project. The deal was signed in a conference room. The customer was won over the following two years, one use case at a time. And a few years later, Google acquired Cask. The company that story built was the company that got bought.

Here is what I carry from those 24 months, for whoever is out there chasing their first yes right now.

The first customer is the hardest, and it is supposed to be. Nobody wants to be first. First means the reference list is empty and the risk is all theirs. Which means your first customer is not really buying your product. They are buying your persistence, your transparency, and one internal champion's judgment. Price those in.

Face to face is a strategy, not a nicety. Every meaningful turn in this story happened in a room: St Louis, Dallas, the grilling, the road trip. Some of the most valuable intelligence we gathered came from hallway conversations, coffee runs, even being walked to the restroom by someone who talked the whole way. You cannot get escorted to the restroom over email. When the account goes quiet, do not send a deck. Get on the plane.

Listen for the accidental yes. We flew to St Louis to sell CDAP and accidentally discovered that the market was starving for what became Coopr. Your customers will keep telling you what they actually want. It frequently is not the thing on your slide.

The technical win is the halfway point. Labs evaluate; business units buy; procurement negotiates; corporate development can freeze all of it on a whim. We won the technology battle at month 12 and cashed the first check at month 24. If your board is asking why the pilot has not converted, show them this paragraph.

Build with them, not just for them. Tigon did more for trust than any presentation we ever gave. Co-creation converts a vendor evaluation into a shared stake.

And the thing I most want to say, founder to founder: there were low days in those two years that felt bottomless, and high days that felt like flight, and at the time I scored myself daily on which kind of day it was. That was the wrong scoreboard. Looking back, every single one of those days, the lobby waits, the depositions, the silence, the smile in the striped blue shirt, was teaching me how to navigate a market, a customer, a competitor, an organization, myself. On average, in retrospect, every day was a day of learning. That is the only average that mattered.

That one deal taught us more than any before or since, so let me put the whole ledger in one place, every line this essay bolded. I will be honest with you: nothing here guarantees that your pursuit ends where ours did. Not every attempt closes. But one thing has been true every single time I have tried: every attempt teaches you like never before.

• Champions see early. That is what makes them champions.
• Your first customer is never a company. It is one believer inside a company, spending their credibility on you before you have earned it.
• Skeptics who convert become your loudest internal advocates.
• The reward for passing a test is a harder test.
• The people who evaluate technology and the people who buy technology are almost never the same people.
• Stop arguing, start showing.
• Listen to the room, not to your roadmap. The market may buy the tool you were embarrassed to show.
• When you are small, your only structural advantages are speed and trust. Spend both freely.
• Nothing you say in a pitch earns trust like shipping something together.
• Twenty months of work, and the yes, when it finally came, took about a minute.
• The yes is when the second gauntlet starts.
• The technical win is the halfway point.
• Face to face is a strategy, not a nicety. When the account goes quiet, do not send a deck. Get on the plane.
• The deal is signed in a conference room. The customer is won afterward, one use case at a time.
• The first customer is the hardest, and it is supposed to be.
• On average, in retrospect, every day is a day of learning. That is the only average that matters.

Persistence, blood, and sweat are not the romantic version of the founder story. They are the operating manual. The first yes is the hardest one you will ever earn, and it is worth everything, because it is the one that makes all the others possible.

If you are in the middle of your own 24 months right now: keep going. Get on the plane. The room can still turn.