This is not a happy ending story.
It all started back in 2014. I had a startup whose clients were advertisers. It was a platform for users to review video ads in exchange for online points that could be redeemed for money or coupons. Watch and ad; rate it; be rewarded. Simple.
After 100 campaigns I kept hearing it would be wonderful to connect their offline ads (e.g. TV ads, billboards, magazines, etc) with our platform. Advertisers wanted real insights and analytics from their offline advertising investment.
As dedicated founders we started working hard on this concept: “from offline to online with your phone”. Within 4 months we had our first version. I remember showing it to our friends and constantly hearing “Wow this is brilliant! It’s like Shazam but for videos”. I was ecstatic!
The revenue was coming in but it wasn’t recurrent. It was difficult to enter the yearly advertising budget. Advertisers assumed our platform as an experiment (mainly to get feedback) and not as a serious distribution channel — despite the fact we picked at 100,000 registered users.
We needed investment to grow and build the new technology’s infrastructure. It wasn’t cheap to maintain a technology that recognized millions of videos within 4s. More on this latter.
In 2015 we raised $ 0.5M from angels and led by a VC. This allowed us to grow our team to 7 members and accelerate product development.
We were ready to storm the world!
In retrospect, our product decisions after the investment killed our startup. We shifted our focus from the local videos ads review platform — where we had 100k users and 60 clients — to a global video recognition consumer product.
We created an App — like Shazam — that recognized millions of videos. Our goal was to have advertisers make their offline assets interactive and invite their audience to download our App and use it to unlock “something”. The practical end was the same as the QR code. How cool is that? Scan a video and “magically” show related content on your screen? Exciting, right?
Wrong. Very few people downloaded the App. It turns out the barrier of downloading the App was too much for the reward (whatever the brand wanted to offer). Don’t get me wrong, we did some cool campaigns with Kia, Unilever, or Volkswagen. But again these were one-shot campaigns. Basically an investment in innovation from the brands.
After long days discussing our future, we thought of something. What if our technology was embedded in native apps like Snapchat, Facebook, IMDB, or even in the Operating Systems of mobile devices — Android and iOS? This would mean everyone could easily interact with their offline environment and get something in return. Brilliant!
Interactive The Walking Dead
This is now 2016 and we had a new strategy. White-label our technology and allow anyone to embed it in their platforms. We built SDKs for Web, Android and iOS and off we went searching for customers. One of our main goals was to have TV shows interactive. Allow viewers to point their phone to the TV and delight them with a new experience.
In this quest, I scrapped all my network, cold reach on Linkedin, went to conferences, traveled between London, New York and San Francisco. I ended up talking to all major TV networks — Comcast, BBC, FOX, PRISA, Viacom, CNN — and closed a contract with FOX. This was a pilot experiment where FOX would use Portugal as an assessment market. It took us 9 months (!) to close the contract.
Even so, we started to see the light at the end of the tunnel. The worst part was over, we could take our learnings from our local pilot and catapult it to the world. We will change how people consume TV and will take our place in the TV innovation history.
We were ready to build a $ 1B company.
After several negotiations with FOX we were able to add our technology to three shows: The Walking Dead, MacGyver and Prison Break. What a victory! All the major shows were interactive. FOX will advertise the shows are interactive, people will scan the TV and have an amazing, memorable experience. Win-win-win.
When the first results started to come in… well let’s take a detour first and come back to the results later.
During 2016 I was mainly traveling demoing our technology to as many people as I could.
Besides TV networks I’ve met with Google, Amazon, Snapchat, Verizon, Blippar and Facebook. Our goal was to integrate the technology in their existing apps and make their users interact with the world connecting the offline and the online seamlessly.
The main feedback was something like: “amazing technology, great demo! But (there’s always a but) something like this isn’t on our product roadmap”. Except for Facebook… It was late 2016 and I’ve met with a Business Developer Director.
Here’s how it went:
(After I’ve demoed the technology)
Director: Wait, can I try it?
Me: Sure, here’s my phone.
(Director takes the phone and scans the video. The phone showed information about the actor from the exact second Director scanned. Director stays hesitant for a couple of seconds…).
Director: I need this.
Me: Uhh… Ok!(My mind was like: Errr, what, how, can I ask… Wait, what?)
Director: Here’s the deal. We have a huge problem right now. We launched Facebook Watch recently and are having a lot of copyright infringements on the platform. We need to build something like YouTube’s ContentID. More info here.
Me: Ok, we can definitely help.
Director: I’ll put you in contact with the product team responsible for this and they’ll take it from there. We are evaluating acquisitions in this space to speed up our go-to-market.
After exiting the meeting I vividly remember the next five minutes. As I went through the lobby I decided to seat on a couch to recover from the excitement. There I was, all alone, in one of the most incredible buildings in Menlo Park after a meeting in one of the biggest tech companies in the world. I found myself looking at the ceiling and smiling for no apparent reason.
The M&A process
After the Facebook meeting, we discovered we had a potential new market to unveil: copyright infringements detection. Users uploaded copyrighted content with small changes (e.g. by tempering with the audio pitch, by slightly rotating the video and by changing the original video with many different techniques) to bypass Facebook’s algorithms.
Because our technology was designed from scratch to recognize videos from low-resolution images, we were pretty effective in recognizing tempered videos. We recognized videos that were rotated, mirrored or cropped. Our algorithm didn’t use audio. We even recognized a bunch of different videos inside one. Here’s a demo with a) 10 trailers in the same video and b) a rotating video.
We arrived in 2017 with our FOX partnership generating mediocre results. No relevant revenue was coming in and the user interaction data wasn’t exciting. We learned that people need a huge reward expectation to take the effort of scanning the TV. Without undisputed usage from viewers, FOX was gradually losing interest in pushing the technology and the opportunity faded during the rest of 2017.
In February we started talking with the Facebook team. They wanted to test our technology at scale. We thought it was a fair request and agreed to be tested without any compensation. We signed NDA’s and were comfortable enough discussing the internals of how our technology works.
After a couple of meetings to discuss the technology, Facebook started to test us with hundreds of hours at a scale we were never able to test before. It was scary as hell! We were all extremely nervous to see if the servers’ architecture wouldn’t crash.
When the first results started to come in we were shocked… 95% accuracy and 0.13% false positives. This was incredible for us! This was paired with the audio industry leader: Shazam. My eyes started to tear up.
We were tremendously proud and happy about this achievement.
Thanks for following up with us. It was a great experience working with your team and we think there is a great potential for your company and service.
We want to provide an update about the evaluation result. From the result, overall we see a good coverage and recall, and your team solved the problems real fast. However, due to low precision and high false positive rate, we decided not to moving forward to the next stage of the evaluation.
Thanks for your time and effort. I am sure our career will get crossed in the future
Caption: Facebook’s engineer email rejecting us
Did you feel that punch in the stomach? I surely felt it. It was so unfair to have amazing results on our side and receive this email. When we asked about the differences — to understand what went wrong on our side — we got this:
Facebook has our own metrics and process to evaluate the product value of the algorithm. But due to the policy, we are not allowed to share with you. My colleague’s point is the final result. Look forward to getting chance to work with you guys in the future.
Caption: Facebook’s lead engineer email really rejecting us
We felt kind of used and disrespected to be honest. Remember this was a two months process with several emails and calls between us. It was one of the most difficult moments of my professional life mainly because of the expectations I’ve built.
We were devastated. I was immature enough and almost took it personally. At the end of the day, it was business as usual for a big company like Facebook — they ended up acquiring Source3 to help them solve the problem. For us, it was a Technical Knock Out.
If someone from a big company is reading this and it sounds familiar, please take a moment to rethink the way you say no to a startup. Especially if you’ve been interacting daily or weekly for the past months. Invite them for a meeting or call and explain them the general decision process. It will take you half an hour and it will make a huge difference for the startup. Believe me on this…
Unfortunately, after this, we weren’t having solid revenue from our FOX partnership. After discussing with our lead investor we decided to close our doors in an unfortunate ending to what could have been a tremendous success.
So many lessons learned! We could have done so much differently. It was a rollercoaster ride with so much emotional commitment. It’s a challenging exercise but I’ll try to generally sum up the main learnings from the whole journey.
Here’s what I learned:
- The line that defines success and failure is extremely thin.
- If you are being evaluated (by customers, partners or possible acquirers) define success KPIs beforehand. That way everyone will have the same common ground.
- Don’t put all your eggs (expectations) into one basket.
- Stick to what is working, i.e., work on what people want and measure it with data; not words nor promises.
- Be kind to everyone. If you feel disrespected or hurt, take a step back. Take some time to breathe and don’t reply right away. Time heals everything and gives your perspective.
- Don’t take life too seriously. Be patient but persistent.
Despite all learnings, the cold reality is that this was a failed startup… but I’m trying it again. This time I’m doing things differently.
Hope you guys had a good read and learned a thing or two 🙂 Let me know if you have any questions.