Shared Service Platform (SSP)
AKA perhaps why ByteDance is worth $400B.
The SSP has also brought together other important teams: user-growth teams, which help identify and acquire desired users; content teams, which establish partnerships to acquire new content; analytics teams, which help to develop deeper user insights; and sales teams, which drive monetization. - Harvard Business Review
At the core of XO is a very specific kind of resource optimization problem (other than our whole existence being a slightly complicated math problem, but I'll save that for another time). That resource optimization problem is something like this:
- We buy small companies that can't afford all the resources they need
- If we buy multiple of them, we can afford the resources but we have to share them
- The way that we share them (our alpha) is the thing to optimize
If we get this right, we win. If we get this wrong, our unit economics won't work and we will go bankrupt... Needless to say I think about this problem quite a bit. For a bit more context I worked at a venture studio for a few years and we certainly figured out ways not to do this efficiently.
Every CEO I've ever met (yes even the public company CEOs) LOVE the idea of SPP and doing 101 things all at the same time efficiently. And yet so few of them have effective organizations. Clearly there's an execution gap.
What did ByteDance (TikTok, etc) get right?
Before getting into the weeds, keep in mind that there is a top level (i.e. top down) BOX all these teams must fit in. They had focus.
Harvard Business Review recently profiled ByteDance and claims their org structure is their super power. Or rather the way they do SPP. The key is specialists. This is an unhelpful observation for tiny companies. You can't afford to hire a specialist in market research for example, you have to lean on product (typically) to serve as market research when exploring a new product or feature idea. Obviously if you hired a ninja market researcher, your product person would be thrilled. They'd get the answers they need faster.
Additionally, if you task a product manager to go create a new app, the product manager might come back with a great idea that you'd then have to go build from scratch. At ByteDance, the answer might be to tweak existing tech. This is really the only way ByteDance could have launched 140+ apps in their first 2 years. You're never building from scratch.
Comparison to the way AWS (Amazon) sets teams up.
Amazon is another company that is lauded for its efficiency, but their setup is quite different than ByteDances'.
- (my interpretation) Product teams don't control as much of their own destiny, and that's a good thing. At Amazon, each team is somewhat independent so long as the contracts between services are upheld (i.e. APIs between teams / services are contracts). At ByteDance, teams do not always get to chose the technology, they may be forced (again, in a good way) to use something from another team.
- teams can work on more than one product / project.
- (my interpretation) You're hiring weapons (specialists), not generalists. This is an uncommon approach. As a startup you're often forced to hire generalists. When ByteDance says they have a 'Video Processing Team' for example, you can bet your ass these guys are going to be able to accomplish some wicked video processing shit. This, practically speaking, for anyone who has worked at a large co knows this is certainly not always the case with 'experts' who work on other teams at the same company. (i.e. oh ya call Joe, he's our best 'Video Processing Guy', and you do and Joe is a quack). A simpler example. If you have a specialist in market research on a team, you as a product owner don't need to mess around yourself, you can tap the team to help and they will be (must be) better than you at market research... bc it's their specialty.
Overall these guys cast a wide net within a constrained space (i.e. many different ideas that all revolved around short-form video content) and leaned on specialists for precision and speed. They took learnings from all the apps launched to make the next attempt a little better. It's genius and incredibly difficult to execute on.
Quick Caveat on AI
Machine learning is the all-star SSP opportunity. Having a killer proprietary recommendation algorithm that you can customize for any application built by some real specialists is an incredible advantage. Indeed it is TikTok's secret sauce which is also a repurposed recommendation algorithm that already existed by the time the TikTok team needed it.
How do you benefit?
As XO congeals on our own SPP, I think we take note of a few highlights from ByteDance.
- As soon as we can, start building out SPP teams with specialists.
- Try to buy companies with close / complimentary tech stacks so we can re-use some technology between companies (i.e. shared stripe integration code). Build out our own code to be shared between companies.
- Top down OKRs that cross teams (i.e. no silos) and a flat org chart.
Is This New?
No, not really. The OGs here are Japanese conglomerates. Take Sony for example. They have been doing this for a long time. They make cameras, smart phones, Playstations, TVs. No way they get all this done without some form of shared services..
I also intentionally left out some of the other stuff like hierarchy, OKRs, culture. All super important. All deserve perhaps their own post.
Thanks for reading!