Chapter 3: Selling the Secret Sauce

The big idea and the big trip

Licensing our tech

The biggest event of the ad tech world was a conference called… Ad:Tech. The New York conference was in November each year. In 2004, it was at the New York Hilton. We couldn’t afford a booth, but I got a pass to the exhibit hall and decided to walk around looking for potential clients. While there were some large established companies, most of the exhibitors were scrappy, performance-driven companies. There were booth babes, ridiculous costumes, and giveaways. As I walked booth to booth, I would ask each company what it did, hoping we could add them to our network as an advertiser or publisher.

As I walked around, I realized there were a surprising number of Israeli companies. At one booth, I noticed that one of the employees was a product manager named Elad Efraim. Figuring he was not a sales guy like so many of the other attendees, I asked him why. He explained that the domestic market in Israel is small, and so most Israeli companies service international markets. In addition, Israel has a bunch of excellent technologists due to their mandatory military service and focus on cyber-security and other intelligence-related technology.

I asked Elad what his company did, and he explained that it was an ad network called Cydoor that operated primarily in Europe, selling performance advertising and working with a broad range of publishers. I immediately recognized the similarity to Right Media. Elad’s role was to manage their ad technology, which was developed by a small team in Tel Aviv, and to help them launch their business in the United States. Since I was effectively in the same job at Right Media, we began comparing notes on how our technology stacks worked. As I explained our real-time predictive algorithm and some of the pricing innovations I had created, Elad started to get excited. He asked me if we had ever considered licensing our technology to other networks.

To be honest, the idea of licensing our technology had never crossed my mind. I had done absolutely no thinking about whether it was a good idea, whether it was possible, what it might entail. Elad said that he was quietly doing an evaluation of a few technology vendors to see if there might be a platform that they could use instead of building a proprietary system from scratch. Falling back on my consulting experience, I asked him if we could do some basic requirements gathering to see if there might be a fit. He said sure, and we agreed to meet the next week at his office to dig in in more detail. He told me Cydoor was planning to make a decision in February, and I should be clear about what I could deliver in that timeframe.

The idea of becoming a technology company was thrilling. As an ad network I could imagine us gaining share and growing the business, but there was no way to conquer the market. If other networks would adopt our technology it would enable the real-time auction I had imagined with Yield Manager. Moreover, a technology company powering multiple networks could achieve a majority of the market. Prior to Right Media, I had run a software company, and I knew what it would take to operate a technology business. There were three challenges to overcome. One, our current ad platform would need to be completely overhauled to support multiple networks. Two, I needed to convince Mike that being in the software business was a good idea. Three, we needed to convince Elad and his company that we were the right partner for them.

I pondered each challenge in turn, and decided to go after Mike first. The next morning, I told him about my meeting with Elad at Ad:Tech. His immediate reaction was to raise his eyebrows. “Our entire company is built on the our superior algorithm. Why would we give our secret sauce to our competitors?”

I explained my reasoning. “They’re primarily a European network. We don’t have any plans to go to Europe. In fact, if we sell them the European traffic that we occasionally get from our US publishers, we’ll increase our media revenue and get the software revenue.”

Mike seemed assuaged by this thought. “OK, let’s say we’re open to going this route. How are you going to convince them to use our tech? And how much work is it going to be to build it?”

I didn’t know the answer to either question yet. I told Mike that I was going to meet with Elad the following week to start gathering requirements, and that I would do some design work over the weekend to figure out what a software platform would look like. He agreed with my plan, and I walked out of his office filled with anticipation.

Linking networks

That weekend was a blur as I designed a completely new system that could handle multiple networks. Some of the changes were fairly simple, like adding a network ID and associating objects to it. All of the existing objects would be part of network 1, Right Media. This required a series of cascading changes through the system, but all relatively straightforward. More complicated was the need for a second datacenter in Europe. Our technical architecture made the assumption that all of the servers and data were in one place. To change this meant digging through more than a year of code to try to understand what would break or slow down if it ran across two locations.

As I worked through the technical details, I started to think about what would happen if the two networks bought and sold an impression from each other. If I did nothing, then the first network would get the impression from the publisher, choose its best advertiser, and then serve the ad. But what if the second network had a significantly better advertiser?

I’ve always had this concept that world-changing ideas come like thunderbolts. Ben Franklin holding up his kite in a rainstorm. Einstein watching luggage flying off the back of a train.

The moment that changed online advertising forever… was just one of many ideas scribbled in my spiral notebook. “Allow first network to resell impressions to second network if it can pay more.” After spending months trying to build a true real-time auction for publishers, my mind was firmly in the land of real-time valuation, trying to find the highest-paying advertiser across multiple networks. It was almost obvious to extend this concept to my multi-network ad server. I could do real-time predictions; I could do real-time auctions; now I could do real-time auctions across multiple networks. I named this feature “linked networks” and continued to design the rest of the system.

The next week I went to Cydoor’s office on 17th street between Fifth and Sixth avenues. Elad met me at the door and gave me the tour.There were offices along one side, some cubicles in the middle, and a conference room in the back. I met Limor Zabari, who ran operations, and Bob Regular, who ran the US business, as well as some of the day to day team. We sat in the conference room and they explained their business to me. They ran 7 billion ads a month, primarily in Europe. Their business had started in what they called “contextual advertising,” which I later learned meant ads that served from browser toolbars or desktop software. They wanted to move toward more legitimate inventory, leading them to consider moving away from their proprietary ad serving technology.

I began to fire questions at them. How did they decide which ad to show on a given impression? How did they track clicks and conversions? Were they able to determine the predicted value of an impression in real time? How many servers did they have and where? As they explained how their technology worked, I filled page after page of my notebook with questions and ideas. One obvious gap for us was their need to handle geographies beyond the US. While all of their employees spoke English, there were differences in how the ad business works in different parts of the world. Most obviously, not everyone operates in US dollars, meaning we would need to support multiple currencies. Some of the differences were small but significant. In Germany and a few other European countries, contracts were booked based on the number of ads shown, not the dollar amount. Since Cydoor was a bigger company, they needed the ability to limit the information shown to certain employees, so that the country manager for Spain couldn’t see the reports for Italy without permission. Different countries had different invoice formats. And on and on.

I proceeded to a demo of our user interface, showing them how we booked ad campaigns, loaded the creatives (the actual ads that would deliver), and monitored the running campaigns through reporting. It was their turn to ask questions and point out gaps or concerns. It seemed like every screen I showed them generated a deep conversation about our methodology. After just a year in the industry building a next-generation ad platform, I was surprised at how much I understood of their business, and how well our technology met their needs. As I left their office, I felt like I had gotten them very interested in working with us. I was exhausted and exhilarated as I walked up Fifth to the Right Media office.

With the additional requirements and insight from Cydoor, Ed and I started building Yield Manager 2.0. Matt began scoping out datacenter options to support a global business. Mike and Aaron focused on operating the ad network since the fourth quarter is the most important of the year for advertisers, who spend aggressively heading into the holidays. My recollection of this time is blurry: days and nights running together as I wrote thousands of lines of code.

In January 2005 I went back to Cydoor’s offices to show them a demo of the new platform. It had many of the features I had promised. It supported multiple networks. It displayed multiple currencies. It supported some basic user roles and permissions. It would run across multiple datacenters including one in Europe. It had this random feature called linked networks where one network advertisers could bid on another network’s inventory.

Bob interrupted me as I tried to go on with my list. “Does that mean we can take advantage of the inventory that Right Media has in Europe?” I nodded. He looked at Limor and Elad, then asked me “How much traffic do you have in Europe?” I told him that we had a few hundred million impressions a month that went largely unmonetized. “So if we are both using your software, our advertisers automatically bid on all of this inventory from quality US publishers. If we generate decent prices, that’s worth a couple of million dollars of incremental revenue for us.” I did some quick math in my head and nodded again. Another look passed between the three of them. I wasn’t sure what was happening, but I knew enough to wait for them to finish thinking about it before I moved on to the rest of my demo.

After the meeting, Elad walked me to the door. Elad had told me the other companies we were competing against: DoubleClick, the leading ad server. Falk and ADTECH, two other leading ad servers, both based in Germany. The existing technology was also a viable option, and the in-house engineers were convinced that they could evolve their own system to meet the growing needs of the business. I asked him how he thought we were doing against the other options. He told me that it was basically crazy to consider using brand-new software from a tiny, unknown company. On the other hand, our software was built specifically for an ad network like Cydoor, whereas the other competitors were built for publishers. In addition, none of the competitors had prediction technology, and it seemed like we could significantly improve their performance. Finally, the idea that they could automatically buy European traffic from us meant that choosing us might actually make them money, whereas the other choices would cost them money. While he was still not sure they could take the risk of working with us, he suggested that a trip to the headquarters in Israel might be the next step.

I was thrilled by the progress I was making with Cydoor. Back at the Right Media office, I wasn’t having as much luck. I asked Mike if I could migrate the Right Media ad network onto the new platform and he laughed. “Business is great! Why would we take the risk of moving onto an untested new platform?” I could see his point, but the vote of no confidence was a reminder of just how much work would need to be done if Cydoor actually said yes.

The next day, Elad emailed and asked if I could join him in Israel in early February. I said yes, then ran into Mike’s office to share the good news. He looked like he had seen a ghost. “Mike, what’s wrong?”

“AOL just told us they’re shutting down all of their spend.”

I gasped. AOL had always been our biggest advertiser, but over the prior six months they had become the dominant spender on our network. Thanks to Mike’s savvy and my algorithms, Right Media had outperformed the other 21 networks that AOL used, and they had consolidated their entire online advertising budget with us. Of the $1.2 million of our revenue in January, AOL was $900,000. Losing them was a disaster. “Why are they shutting us down? Is it because of the Ad.com purchase?” Advertising.com was the largest ad network and our primary independent competitor. AOL had recently purchased them, and we were paranoid that they would convince their new parent company to shift all of the ad spend over.

Mike shook his head. “Nope. It’s a corporate decision. They’re not going to advertise dial-up access any more. We might get a little bit of broadband budget, but the big money is gone forever.” It was the worst possible news. I tried to imagine the impact on our business. Layoffs? What about our increasing server bills? How would we fund the new datacenters and hardware we would need to run the Cydoor business? We looked at each other, both knowing the magnitude of our situation.

As we talked, we realized we had two glimmers of hope. The first was the nature of our auction-based technology. While AOL generally was the highest bidder, there was also a second-highest bidder on every impression. If AOL stopped spending, perhaps the other advertisers would pick up the slack, especially since performance advertisers had limitless budgets as long as you hit their goals. The second area for optimism was the idea of licensing our technology. When the ad network was growing like a weed in 2004, Mike had been skeptical of my idea. Now, it seemed prescient, a way to hedge our bets against the unpredictable ad network business. We agreed that I would fly to Israel and he would do everything in his power to drive revenue in the short term.

Selling the Dream... in Israel

Elad and I landed in Israel to dazzling sunshine. I was tired from the red-eye flight and full of energy, excited to visit the Holy Land. We drove from the airport to the small town of Abu Ghosh, where Elad promised me the best hummus in Israel - which in his humble opinion, meant anywhere in the world.

We walked into the restaurant and I was shocked to see that the majority of the patrons were Arabs. The air was full of Hebrew and Arabic, metal tables laden with pita and hard-boiled egg and the garlicky scent of hummus. I shared my surprise with Elad as we ate. I only knew Israel through the constant stream of news about terrorist bombings and conflict between Israel and Palestinians. The reality that Elad was showing me was different. Abu Ghosh was a primarily Arab village in the heart of Israel. The conflicts in Gaza and the West Bank were real, but so was the integration of Arab communities like this one into mainstream Israeli life.

We drove to our hotel in Tel Aviv and dropped our bags. We then went for a walk along the streets of the older part of town. Older is a relative term in the Holy Land, as Elad explained, but Tel Aviv was a very new city. We passed homes in pastel colors and attractive boutique stores. We walked into an ice cream shop that Elad promised was one of the best. As I happily licked my scoop of chocolate, Elad looked up in alarm. “I forgot to bring my shekels!” The man behind the counter laughed and told him to bring money next time.

After an early dinner, we went to bed, agreeing to meet in the morning for breakfast. When I walked out of the elevator into the breakfast room, I was met by a fantastical spread. There were the classic American basics - eggs, potatoes, cereal - but also hummus, babaganoush, Israeli salad, labaneh, cucumbers, peppers, piles of bread and cheese. At first I was hesitant to eat vegetables for breakfast, but I quickly piled my plate high and came back for seconds. I was fueled up for the big day.

When we finished, we hopped in the car and headed north to the Cydoor offices in Hertz’liya. As we pulled off the highway we drove past the offices of major tech companies like HP and IBM. This was the heart of the Israeli tech scene. We parked and rode up the elevator. I had no idea what to expect. For some reason I had decided that to break the ice, I would wear a gaudy orange and red Hawaiian shirt over my jeans and flip flops.

I walked into the Cydoor office. It was very similar to tech company offices all around the world: wood floors, some cubicles covered in pictures and silly accessories. There were offices around the perimeter. I saw a row of company photos taken at picnics and holiday parties along one wall as we walked back to the conference room. It all felt very familiar. As we entered the conference room, I met Cydoor’s CEO, Iri Zahar. He was tall, sandy-haired, with glasses. His greeting was ebullient. “Brian! I have heard so much about you. Welcome to Israel!”

The next few hours flew by as I walked the Cydoor leadership team through a demo of Yield Manager. The software had continued to improve over the previous weeks, and I had incorporated much of the feedback from Elad and Limor. The team’s initial skepticism slowly melted into grudging appreciation. The commercial leaders were especially intrigued by the linked networks concept. The operations staff admired some of the automation I had built into the reporting system. I knew it was going well.

At lunch time, a few of us drove to the nearby marina. As we pulled into the parking lot, a security guard with an imposing firearm waved us to a halt. He asked Elad a few questions in Hebrew, then grabbed a long pole and began walking around the car. I looked querulously at Elad. “He’s checking the car for bombs. There’s a mirror on that pole, and he can see if anyone has planted anything on the chassis.” I must have looked terrified, because he continued, “I actually feel safer in Israel because we don’t take security for granted.”

Over lunch at a Kosher steakhouse, I was full of questions. Such as, “what makes a steak house Kosher?” (Answer: they don’t put butter on the meat when they cook it). The Cydoor team had tons of questions for me as well. We were building a trusted relationship that would underpin their decision about our technology. The mere fact that I had showed up in Israel was huge. That I was curious and open about their business, their culture, and them as individuals furthered the idea that this wasn’t a transactional relationship. I think they could tell that my interest and curiosity were genuine. I wasn’t there to sell them software; I was there to build a partnership.

In the afternoon, Elad took me to Jerusalem. As we walked through the winding passageways and past the stalls, I felt transported back in time. I turned into an alley to catch a glimpse of a beautiful mosque and Elad grabbed my arm. “You can’t go that way.”

“Why not?”

“See the guys with guns at the end? That’s the Al Aqsa Mosque. Israelis can’t go through there; that’s Palestinian territory.”

We turned the corner and suddenly were were at the Western Wall, the holiest ground of Judaism. After a few minutes taking it in, we walked a couple of blocks and entered one of the most important Christian churches in the world. As we headed back to the car I tried to get my head around the layers of history we had just walked through. Elad drove us up to a college campus on a hill overlooking the old city so I could get it all into perspective. He pointed to the north. “There’s the West Bank. That’s the Palestinian territory. If you look east, you can see the airport; beyond that is Tel Aviv.” Splayed out in front of us was the heart of western civilization. It was a business trip, but it was also a history lesson.

The next morning I woke up to a terrified email from my mother. “Honey, are you ok? I saw the terrorist attack on the news.” In these years after 9/11 and before smart phones, the world felt both bigger and more dangerous. I opened the New York Times website and read a story about a suicide bomber at a night club in Tel Aviv.

I wrote my mom back and told her I was ok, then went to breakfast. “Elad, I read about the suicide bomber. Should I be afraid?”

After breakfast, we took a walk along the Mediterranean. The sun shone gloriously off of the water. Old men strutted up and down the beach beside joggers. The world looked perfect and pristine. A few hundred yards south of the hotel, Elad pointed to our left. “That’s where the bombing happened.” There was a building surrounded by police and military vehicles, looking blackened and worse for the wear. On television and the pictures in the paper, all you could see was an intense crime scene. From the beach, it was just one building out of hundreds within view, not a city on fire but a single incident that wasn’t even interrupting the morning traffic.

As we drove to the Cydoor office, Elad briefed me on what I was walking into. “Iri wants to give the entire company the demo you showed us. He thinks that will help everyone get comfortable with the idea of moving to a new platform.” I asked if there were any gotchas or concerns to worry about. “Well, the engineers are upset, because it means we’ll stop using the system they’ve been building. The Germans are upset that we’re not choosing a German company. I think the commercial teams will have a lot of questions. Don’t worry about it. You’ll be fine.”

The entire company had assembled into their largest conference room to see me speak. There were 60 or 70 people staring expectantly at me as I began my presentation. After the first few screens, I heard someone mumble something in Hebrew. Iri said loudly, “English please. You’re being rude to our guest.” Someone else said something in Hebrew and he turned pale and said “ok ok.” I was pretty sure that the comment wasn’t positive. As I proceeded through the demo, the rumblings from the engineers grew louder, and they began to ask gotcha questions. I did my best to keep my cool. After a bit, a well-coiffed man with a German accent introduced himself “I am Klaus Fritz from the Spain office.” I cocked my head. Did he say Spain? While my brain tried to align his accent and his geography, he began aggressively questioning my credibility. “How big is your company? How many software clients do you have? Have you ever done business in Europe? Do you understand how different things are in Europe?” I fended him off as best as I could.

Finally my interrogation was over, and everyone receded back to their desks. Shaken, I apologized to Elad for my performance. He shook his head. “Don’t worry about it. Everyone needed to have their say. You did exactly what you needed to do.” That night, he and I went to a neighborhood bar for a drink. I mentioned to the bartender how much I loved the local beer, Gold Star, and he told me to take the beer glasses home as a souvenir.

The next morning we drove to Ben Gurion to fly home. Cydoor had given me a letter to show security, written in Hebrew. They claimed that it said “Brian was here on legitimate business, go easy on him” but I was half-convinced it said “We have no idea who this guy is, give him the full experience.” After an hour of being scanned, poked and prodded, with every single item in my bag analyzed for potential weapons, I made it to the gate.

As we flew back to the US, I pondered my trip. In one sense, Israel had been what I expected: terrorist bombs and skeptical clients. In another, it had been an eye-opening contrast: intelligent clients looking for a true partnership; a safe, generous, sparking city unfazed by occasional conflict. I was proud to have learned the most important word in Hebrew: toda (thank you) and the proper pronunciation of hummus, a word that transcends religion.