Sunil Nagaraj saw a need for software that identified suitable pairings using algorithms to analyze data on the attributes and preferences of both sides in an exchange. Whether that was a date, a job offer, or a real estate transaction. His initial business concept was that he would be able to license such a matching engine to dating sites such as e-Harmony and Match.

Sunil’s matching engine would draw out a user’s profile and attractiveness to potential partners based on the user’s digital footprint. The matching engine would look at the websites the user visited and bookmarked, the apps they used, and what movies or music they consumed online. His belief was that using this computer-generated information would triangulate a profile that was more accurate and resulted in better matches. At least in comparison with self-reported data where users could exaggerate their responses or even outright lie when creating their profiles.

Online dating was selected to be Triangulate’s first target market. The online dating industry was ready for disruption, it was a $1.2 billion market that had not seen much innovation since e-Harmony launched its advanced algorithm nine years earlier. Sunil figured that inaccurate self-reporting was probably more common in online dating than in other markets. The engine would be capable of identifying suitable matches in other areas such as recruiting, university admissions, and service providers. But the initial market would be online dating.

Sunil’s business was based on three critical assumptions. The first and possibly the most critical was that Triangulate would indeed generate better matches than using self-reported information. The second was that users of online dating sites would see that the engine produced better matches and would be willing to pay a premium for them. And third, that established dating sites would be willing to license the product and offer it as a premium service. Also, that they would be willing to share the income generated with Triangulate.

Before he could raise funds, Sunil would need to validate these assumptions. But before even that, he had to develop his matching algorithm. To do that he needed data, which meant finding folks who were willing to let him track their online activities.

He found 100 volunteers to download tracking software. Unfortunately, the software did not work on most of the volunteer’s computers, derailing this experiment.

The assumption that people would be able to recognize the superior matching capabilities of Triangulate also went largely untested.

The contentious subject of who is going to be in charge proved to be another hiccup for Triangulate. Sunil and his venture partner differed on who was going to be the CEO. Sunil’s co-founder suggested co-CEOs while Sunil, citing the need to make quick decisions, insisted on being the sole CEO. Having other opportunities available to him the co-founder decided to leave.

With blind faith but no validation of his assumptions, Sunil moved to Palo Alto. He recruited two new co-founders and a data scientist. Between them they built the first version of Triangulates matching engine, it was October 2009. This version automated the collection of user data from sites such as Facebook, Twitter, and Netflix.  Now it would be necessary to teach the algorithm to use the data to make suitable recommendations. To get the data required Sunil needed to partner with an existing dating site, but to find a partner he would need to demonstrate his algorithm was working. A classic catch 22.

It was time for Sunil to rethink his strategy and make his first pivot. As well as continuing to pitch his matching engine to VCs and dating sites, Sunil would use Triangulate to launch its own dating site using Facebook, which was growing rapidly at the time. The dating site would allow Triangulate access to the real user data needed to refine the algorithm.

Other dating sites were using Facebook. Zoosk had forty million members and raised over $10 million in venture capital in two years since its founding in 2007. Compared to Zoosk, Triangulate would make far more use of the data available on Facebook.

Triangulate brought its own approach to online dating. As well as using the data gathered from a user’s online presence it allowed a form of social proof in the form of a friend’s endorsement. Sunil thought that this would help the site grow virally as daters would be motivated to sign up friends who could vouch for them.

This was an online version of the wingman tactic that people had long used when approaching a potential romantic partner. It was for this reason that Sunil called his site Wings. It launched in January 2010.

Folks with a Facebook account could use Wings without payment. This allowed them to view other users’ profiles and receive five free recommended matches per day. When a user signed up, they received free digital coins which could be used to buy virtual gifts, additional matches, or the ability to message potential matches. Of course, they could buy additional coins for real money. They could also earn coins by allowing Triangulate to, for example, access their Netflix account.

Now that they had access to data, the team could start to refine the matching engine. They ran a test that showed the engine could predict matches with a high degree of accuracy. Now they had something to show venture capitalists.

The existence of a sophisticated matching engine intrigued potential VC backers. As did the prospect of viral growth for Wings. Established dating sites relied on massive advertising spending to attract subscribers. Customer Acquisition Costs were over $100 per subscriber. Sunil estimated Wings CAC at $45 per user, and each user would have a lifetime value of $145 based on the average life of nine months.

In March 2010 Triangulate closed a $750,000 seed round from a well-respected Silicon Valley VC firm. With the money, the co-founders hired their first employee, a graphic designer.

Once live the team started getting feedback from their customers. A segment they had neglected up until now. They were quick to react, adding features rapidly to the website.

For example in the beginning users’ photos were small as the cofounders felt the user should trust the matching engine rather than defaulting to how a person looked. But users wanted to look at photos. To gauge the level of demand they added a dummy “Photos” button. When users clicked the button they got a “Coming Soon” The button got a lot of clicks. The team added fully featured photo albums within a few days.

Wings found its early users through coverage in the technology press. Facebook added Wings to a dashboard promoting Platform partners. They were also spending about $5000 per month on online ads to attract users. By September they had a user base of thirty-five thousand people. 10,000 of those users were in California. Triangulate concentrated their marketing in the state to increase the chances of a user finding a match in reasonable proximity. The cost of acquiring a customer this way was about $5.

The team also ran a set of ads that provided an incentive in the form of digital coins that could be used on social games. Although this had a very low CAC of about fifty cents, it did not attract the ideal user. Some were not really interested in dating, and only 25% returned to the site after the first week. And they were geographically diverse, those joining from California had a reasonable chance of finding a nearby match. Those in more remote states, such as North Dakota, less so. These users quickly disengaged, and even more damaging gave Wings low satisfaction scores on Facebook. This drove up customer acquisition costs.

To prevent users from uploading inappropriate photos, Sunil used to check all pictures personally. With approximately 1500 pictures being uploaded each day this began a huge time sink.

The team came up with a novel solution. A “Rate Singles” feature that outsourced the task to users who were asked to rate the photos’ attractiveness under the guise of improving their own matches. This proved to be very popular accounting for 20% of Wings’ page views.

By October it was clear that users did not consider Wings’ matching engine produced superior results. So, Triangulate stopped using the engine for that purpose.

Let’s summarize this episode. The business’s original intended unique selling point was to develop a superior matching engine using data gathered from the user’s digital footprint. This would deliver a more accurate picture than one that is self-reported, where is the potential to exaggerate one’s better qualities or to even lie. Online dating was chosen as the first market and the possible needs of the industry most closely matched the engine’s capability. The attempts to validate the three business assumptions, that the engine would deliver better matches, that users would be able to detect the superior results the engine delivered, and that they would be willing to pay for it, went largely untested before the business launched.

After launching Wings it was clear that one of the assumptions, that users identified the improved matching capabilities had been disproven. What would you do next? The next stage in Triangulate’s journey will be posted tomorrow. Please subscribe to ensure you do not miss it.

Leave a comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: