A look at the internal memos of Twitch and Periscope

We are actively investing in consumer video startups. It’s a space we’ve been focused on for a while now. For example, in 2012, we invested early in Twitch following the company’s pivot from JustinTV, which sold to Amazon for approximately $1 billion and we were one of the first seed investors in Periscope, which was acquired by Twitter in 2015 while their product was still in beta.

As we make new investments in the space, I thought it would be worthwhile to see what can be learned from the past. Specifically, knowing our early bets in Periscope and Twitch paid off, I went back to the first investment memos we wrote recommending funding Twitch and Periscope to see how they might shed light on future video opportunities.

I found this trip down memory lane to be particularly enlightening, and thus I am open-sourcing certain elements of both investment recommendations with the permission of the founders. Because I remembered seeing shades of Twitch in Periscope, let’s start with Twitch, an investment I championed in partnership with my colleague David Cowan.

A look into Twitch

What we liked

The team: They were passionate about the business, solid on execution and knew where to focus to build the organization. Emmett Shear was a first-time CEO, but up for the challenge. Although there were gaps in the team, they were off the charts on a number of “hard to hire” attributes: passion, product vision and empathy for their users.

Here's what we wrote in the internal memo:

Team Update

We have spend an increasing amount of time with the core management team since the initial IR update and continue to be impressed with their scrappiness and focus on execution.

Emmett Shear - CEO

Despite being a first time CEO (he was CTO of Justin.tv before spearheading the Twitch initiative) Emmett has impressed with his open leadership style, calm and effective negotiating approach, and his passion for the Twitch product and community. The one thing we hope to see out of Emmett going forward is a deeper focus on implementing process and standards (specifically around partner support) as the team scales. We believe he can be the long term CEO of the business.

Kevin Lin - COO

More than any other executive, Kevin is Emmett's partner in building Twitch. He is a detail-oriented executive whose primary focus is on the day-to-day operations of the business. He is a grounded, solid counterpoint of Emmett's focus on high-level strategic thinking. He handles the financial aspects of the business today but the business will need a CFO pretty quickly.

Jonathan Simpson-Bint - CRO

Jonathan joined Twitch in October 2011 after a career in game content having founded IGN in 1996 and subsequently developing products for publishers including Microsoft, Sony, and Nintendo. He is the prototypical industry insider and is the reason Twitch has had the ear of top brass at Activision, EA, Ubisoft and other tier one publishers. Beyond having a world class rolodex, Jonathan has proved himself to be an excellent salesman and core asset to the Twitch team.

While the remainder of the executive team is strong, specifically in infrastructure and video technology, we believe Twitch will need to augment the group with an experienced ad sales leader and more business development talent as they continue to scale.

With the Amazon acquisition behind us, it’s impressive to look back at the path Twitch travelled. The team killed it: Emmett proved himself to be a terrific CEO, managing the technical complexity of scaling such a large site cost-effectively as well as the business challenges of managing the community, cementing deals with both game publishers and console platforms while building a successful advertising and subscription business. In the process, he hired a strong management team, including a CRO, CFO and General Counsel to help him manage the growth.

It turns out we were wrong about the need for an experienced ad sales leader. Immediately after we invested, Twitch’s CRO Jonny Simpson-Bint took this on and built one of the best ad sales teams around.

Pattern recognition: Twitch, characterized as a two-sided esports marketplace made up of players and spectators/viewers, benefited from powerful network effects that helped grow the video streaming platform over time.

Value Proposition

It is helpful to think of Twitch as a marketplace where gamers use the IP of publishers as "raw materials" to make content that viewers want to consume, and charge those viewers through advertising, subscriptions, PPV, and paid HD. Like many marketplace businesses, Twitch creates value for all constituents by allowing them to transact in a low friction environment and handles all monetization on their behalf. As a result, the company is able to capture the majority of the transaction value.

The idea and market size: As we described it to the firm: “Twitch is based on the non-intuitive idea that watching other people play video games is entertaining.” Clearly, Twitch targeted a very specific audience. Maybe we weren’t their target demographic, but their growth was undeniable. We knew they were on to something and conversations with enthusiastic Twitch partners confirmed that.

We did have a long existential debate internally about whether the overall market size was big enough — and believed this to be the key risk to the investment — but, unfortunately, there weren’t many definitive data points we could point to that informed the debate.

Market

Twitch's value proposition for streamers and publishers, and therefore its business model, are predicated on the idea that there is demand for viewing live digital gaming content. It's obvious from global human interest in sporting events that there can be a large audience for watching games, especially when a portion of that audience has played and/or enjoyed the game. The relevant question for Twitch is whether there is a large enough audience for viewing live digital game content to support a valuable media business around what is being termed "eSports." Since this is an emerging category in the US, there are no incumbents to directly measure the size of the opportunity or how monetizable that viewership will be at scale.

We are encouraged by the explosion of live eSports events in Asia. South Korea has been an early adopter of eSports with dedicated cable channels devoted to it. Events like this at a smaller scale are being held around the world with increasing frequency. In face, eSports leagues on Twitch are beginning to see television audience sizes. This is not surprising when one considers that the 30m players in the Call of Duty community outnumber US residents who played football, basketball, baseball, tennis or golf in the past year. The fact that a single hardcore gaming franchise outnumbers some of the most popular and commercialized US sports in terms of direct engagement is a hopeful sign that latent demand may exist for viewing live game content (just as real demand exists for viewing live sports).

Another encouraging sign is that video on demand (VOD) viewership for streaming game content is 7x the size of the live market at 250m monthly uniques. This space also has a clear leader, YouTube, where gaming is the largest category after music. The majority of YouTube's gaming content comes from its second largest network after VEVO, Machinima, which grew 3x last year and has reached 175m monthly uniques.

In drawing parallels between Twitch and the viral growth of online poker and esports in Asia, we convinced ourselves video games could not only become a spectator sport, but that the market size was large enough for Twitch to become a massive site, which is necessary to have any sort of effective advertising business.

It also felt like a preexisting behavior, as even before the Internet, it’s pretty natural to find a group of friends watching each other play a particularly engaging console game live in the living room. So why wouldn’t this experience eventually migrate online?

Twitch’s metrics pretty much spoke for themselves:

Twitch has experience phenomenal month-over-month growth since January 2011. The company is growing monthly uniques at an average of 9.6% and minutes watched per month by 16.0%. In addition, average daily unique visitors now watch over 75 minutes of content while on the site. Of the 18M monthly uniques, ~40% are in the US, ~40% are in Western Europe, and ~20% are in Asia.

What we got wrong

Twitch’s advertising monetization efforts: When we first invested, Twitch was still in the nascent stages of figuring out its media sales strategy, having tried remnant optimization, rep firms and even direct sales. They were working with one specific partner to bundle Twitch’s inventory with theirs in order to garner premium CPMs:

[redacted]% of the net CPM that Twitch generates. [redacted] sells advertising on Twitch's behalf with [redacted]% of the inventory sold as a premium "e-sports content" generating $[redacted]-$[redacted] revenue with Twitch. Eventually Twitch may build a direct advertising sales team but for now they are happy with [redacted] representation as they package it with GameSpot and other premium gaming content and are in the process of further segmenting the private exchange content so portions of it can be sold as a premium. [redacted] involvement is extremely helpful to Twitch in that they are able to ensure that the inventory stays solidly in the "premium" bucket - enabling them to generating [redacted-redacted]x the CPMs of remnant. Of course this is also a key risk as it's unclear what the value of the inventory would be on a standalone basis. We have become increasingly comfortable with this risk as they grow their audience size to become a top 5 gaming destination and one of the top two site for gaming video - of increasing interest to brand advertisers.

This turned out to be wrong; having a partner rep their inventory actually degraded — not improved — its quality. The better strategy was to sell Twitch on its own. Fortunately, Twitch was able to amicably end the relationship and build their own sales team. By that point, they had grown enough that Twitch was by itself a premium site and didn’t need the help.

Financing needs: Our memo is completely silent on this topic. Given that the team had operated JustinTV profitably for many years, I think we believed they would be able to scale the infrastructure needed to support Twitch in a capital-efficient way.

It turned out that as traffic continued to ramp faster than expected (and the partner-driven advertising efforts underperformed), we needed additional capital to continue scaling while taking advertising in-house. As a result, we raised a Series C one year later. We try to forecast things like future financing needs, so this was a miss on our part.

Competition for broadcasters: While we talked a lot about competitive threats from other players like own3d (now defunct) and YouTube, we missed the key issue here: attempts to poach popular streamers would become a regular (and costly) occurrence. The team did a great job fighting and winning this battle through building the better product and attracting the biggest audience for live game viewing, but it was a constant distraction and more costly than we realized.

Mobile: The only mention of the word mobile in reference to Twitch is in this section:

Twitch is in discussions with all of the largest hardcore gaming publishers including Activision/Blizzard, EA, Ubisoft, and even mobile publishers like ng:moco to integrate the ability to stream to Twitch directly into their games without third-party software or hardware as is required today. The first of these integrations with a small Swedish game developer Paradox.

This shows you how much the world has changed over the past three years (or how asleep I was back in 2012). We just didn’t focus our diligence on Twitch’s plans for mobile. Luckily, the Twitch team was more on top of it than we were and built a highly popular mobile app for viewing (which is consistently ranked in the top 25 in the entertainment category on iOS today) and several SDK-based mobile integrations for broadcasting.

Our conclusion

Would Twitch be big? It was too early to tell, but we were optimistic. The niche was unusual, but we couldn’t deny the traction and potential. For a young company, Twitch exhibited some powerful early indicators of success: a meaningful and growing customer base, a powerful network effect and very strong engagement. We wanted to invest.

We are excited to invest in a fast-growing video community site with a very passionate and engaged audience. Twitch is unequivocally the leader and has strong momentum in what is emerging as a major online video category - beating heavyweights like YouTube and Machinima in the process. In addition, the company is in the process of executing integrations with major game franchises like Call of Duty - which we believe will further accelerate growth and monetization. We strongly recommend this investment.

A look into Periscope

To quote our first investment recommendation from April 2014, we considered Periscope “an audacious concept,” whose time may have come. Recognizing elements of JustinTV (the failed predecessor that spawned Twitch), we felt that now both the technology and the market were ripe for Periscope.

I recommend that 15 Angels invest $225k in the seed round of Periscope, an audacious concept for a viral mobile applications that would allow anyone with the app to request and view live streams from anyone else with the app anywhere in the world - much like the original vision for JustinTV which unfortunately did not benefit from the wide proliferation of smart phones with video cameras and high bandwidth connectivity.

What we liked

The Team: As with Twitch, we respected and liked the founders of Periscope. Both teams were scrappy and focused on executing.

The founders are extremely high caliber. Kayvon Beykpour started TerriblyClever Design as a freshman at Stanford. TerriblyClever allowed universities to design white-labeled mobile applications for accessing campus systems (courseware, campus maps, transcripts, registrations, etc). Blackboard bought the company in 2009 and Kayvon became GM of Blackboard Mobile growing the team to over 100 while continuing as a student at Stanford and graduating in 2010 with a computer science degree. He continued with Blackboard as GM of Mobile until December when he left to found Periscope.

His co-founder in this venture - as in the prior one - is Kayvon's childhood friend Joseph Bernstein. Joseph dropped out of New England School of Photography to pursue TerriblyClever and subsequently became CTO of Blackboard Mobile after the acquisition.

As with Emmett at Twitch, Kayvon, Joe and the team turned out to have very prescient vision for future forms of online social interaction and be strong product designers and scrappy operators. I didn’t have as much time to observe their execution, but from the tremendous growth in Periscope (10 million registered users and 40 years of video watched per day, according to a Q&A Periscope stream that Kayvon held), they’ve clearly managed the dual challenges of generating awareness and scaling the technical infrastructure required to support a massive user load.

Total addressable market: Periscope had broad appeal. Sure, it was audacious, but given the proliferation of social media access on smartphones, it seemed conceivable. And in this case, there wasn’t as big a concern about only appealing to one particular type of user. As Facebook and Twitter have shown, the inclination to share — either broadly or with a limited set of friends — is pretty ubiquitous if given the right environment and context.

The team envisions everything from Arab Spring protests to tech conferences to the line at Bi-rite Creamery to the view from the center field bleachers at AT&T Park during the World Series eventually being requested and streamed on Periscope. Like with any mobile application concept, the challenge will be getting to enough of a critical mass that there are interesting things to watch and users with the app in the right places so that common requests (e.g., view of the Eiffel Tower in Paris) can be met. The team has a few clever ideas for how to make the concept viral including embedding live video feeds into the Facebook timeline and encouraging people to share their stream via social channels.

What we got wrong

When investing in Periscope’s seed round, I expected them to launch the app, take some time to iterate to get to a compelling product and eventually refine their approach to distribution over the following year. I was surprised (and amazed) by how quickly all of this happened.

Within five months of our investment, they were on the radar screen of large Internet companies like Twitter, and had nailed the interaction and distribution model so cleverly that Twitter was interested in an acquisition even before they had launched a product. Had I known this, I certainly wouldn’t have waited to try to lead the Series A.

Our conclusion

Again this was a very early investment — much earlier than in the case of Twitch — but the idea and founding team were so compelling that we invested as much as the team would let us. In fact, when I first circulated this investment memo, my colleagues at Bessemer told me that the idea and timing were so compelling and the team so well-suited to building a great product that I should ask Kayvon to let us invest more. Fortunately, he relented and I was able to more than double our investment.

This is a zero-stage consumer mobile idea and so the concept - at least as envisioned today - is highly likely not to work. The challenges are many, including building a compelling product, figuring out how to engage and reengage users, figuring out how to monetize the audience, and most importantly figuring out a scalable viral growth engine. However, the high caliber of the team along with the potential for the app to grow virally has convinced me that an angel investment here makes sense.

In reviewing these memos, some of the conclusions are obvious.

The team is critical to ultimate success. The timing must be right (e.g., JustinTV wasn’t able to capture the broader Periscope opportunity back in 2007; there weren’t ubiquitous smartphones back then). Having a revenue model that will ultimately fund the operation of the business shouldn’t be an afterthought. But some of the interesting pieces lie in the nuances and patterns.

Both teams thought about the appeal and incentives for all of the various constituents — broadcasters, viewers and game developers/publishers in the case of Twitch, and broadcasters and viewers in Periscope. They thought up front about incentives and rewards (social, financial and otherwise) for being on the platform and how those mechanics differed as the products grew up and the online behavior became more mainstream.

Both teams also thought about the fundamental mechanics of social interactions in the offline world — and how best to bring those into new media in the most natural way possible. This didn’t necessarily mean replicating existing behaviors without any behavior change, as the Internet and mobile affords an opportunity to re-invent and reshape cultural norms.

But each successful social product has to tap into an innate desire to be social in the first place — an obvious statement in hindsight, but not always considered up front. And finally, both teams had tried it before and persisted — with the result being that they were driven to have a bigger impact this time around.