What is up with Looker?
The strange story of how Looker became synonymous with Semantic Layers.
Looker started in 2012 but already by 2014, only 2 years later, it was gaining significant market share and mindshare as the go to self service analytics tool. And in 2019, 5 years later, it was acquired by Google for $2.6 billion dollars. An amazing story by any account. Looker’s success was fueled by its LookML tech which allowed engineers to configure a semantic layer. It repopularized the concept of a semantic layer which at the time was losing mindshare.
Not sure what a semantic layer is? Read:
But, when you dig deeper its semantic layer left much to be desired. Barely useful.
Each explorer defined a single fact domain.
No concept of snapshot measures
No ability to create leveled measures
Exclude certain dimensions from the group by or filter or both
Include certain dimensions in the group by
No automatic way to blend data across fact tables
Custom syntax with a long learning curve to define semantic objects (LookML)
Looker’s semantic layer is good at querying a single fact domain at a time. But is that really a hard problem? Does that really enable self service?
At the time there were many tools that could handle much more complex use cases. Even more that could handle the commonly required features listed above.
So what happened?
A Case of the Tail Wagging the Dog
Maybe it was the amazing UX for business users.
No. Obviously, not that.
In fact, enabling business users seem to the secondary use case not the primary. A further surprising fact is that companies were paying a premium for Looker. But why?
Well friends, when the tech team, in this case Data Engineering, is more influential in the buying decision, you end up with Looker.
Kudos to the Looker team on finding this angle though. Enterprise sales is extremely hard, but if you can find the right influencers for your product your chances to close goes up dramatically.
Looker solved many of the problems engineering teams were experiencing with traditional BI tools. These old guards of the self service analytics were exclusively GUI driven. Making standard principles of systems engineering difficult. Namely, version control and automation. This was enough for Looker to eventually end up with 2% of the BI market.
The Way Forward
At Strata we are solving for both. Smart adoption of software engineering principles for engineers and a world class semantic layer for business users.
We can have both!