Looker upgrades their platform’s machine learning with highly complementary Google Cloud BigQuery ML integration

The integration between Looker and BigQuery ML unites all elements of data scientist workflows, so they can do all their work without having to Leave Looker.

Business intelligence and analytics vendor Looker has announced a new integration with the Google Cloud BigQuery ML [Machine Learning] service that they believe significantly enhances their platform’s value because it adds major functionality to its data science workflows. The announcement was made at the Google Next 2018 conference.

Santa Cruz-based Looker is a Big Data vendor which does its analytics in the database in real time, using a browser. One reason for its efficiency is that it doesn’t store any data itself, so it doesn’t silo data in separate applications.

“The first wave of business intelligence was data warehousing like Oracle and Teradata, and the answers it gave were very accurate and reliable, but they were monolithic stacks, and very inflexible in terms of what you could ask,” said Daniel Mintz, Looker’s Chief Data Evangelist. “The second wave was self service departmental tools, and they provided the ability to ask ad hoc questions, but there was no real governance, they lacked standard metrics, and the data wasn’t as reliable. Now we are in a third wave of BI, which is a revolution under the hood in terms of what database techs can do. It has gotten very cheap and can now provide the reliability of the first wave in the data warehouse along with the second wave ability to ask ad hoc questions.”

Looker has focused in the last several months on data scientists, which had not been a primary focus historically for the company.

“Data scientists have three parts to their workflow,” Mintz said. “The first is data munching, which is where they spend most of their time. The second is using the data to build predictive models. The third is moving the insight obtained from the models back to the business folks who need it. When we looked at that workflow, we decided not to focus on the middle piece. Data scientists have very strong options about tools that they like. However, Looker is very good at the first and third steps – preparing the data and putting it into the hands of the right people. We smooth out the data so the data scientists can leverage the cleaned data and pump it into their workflows, then we put it back into Looker so end users can use it.”

The BiqQuery ML integration allows the first and third steps to be seamlessly linked together, so that the data never has to leave Looker at all. The Google service makes it easier for data scientists to quickly create and deploy models at scale. So the integration means customers can use Looker to run the models directly in BigQuery, and seamlessly move from there to making the analysis from the models available to the organization’s end users.

“BigQuery ML is one of the incredible new databases, and is growing fast among our customer base,” Mintz said. “Our deep partnership with Google had them reach out to us to get this to work seamlessly. What it means is that now, if you use BigQuery ML for that middle stage, nothing ever leaves Looker.”

Looker sees BigQuery ML as a purely complementary offering to their own,

“It’s definitely one plus one equals three,” Mintz said. “They make each other more powerful, and better. Because Looker doesn’t have its own proprietary data store – as most BI tools do – we don’t compete with data warehousing capabilities. BiqQuery ML is the leading edge of what I think will be a new trend  bringing together analytic and machine learning workloads.”

Looker expects that this new integration will further build on their recent momentum.

“We now have over 1400 customers, and are growing fast,” Mintz said. “Our customers run the gamut, from the new like Jet.com and Amazon to the old like Sears and Nordstrom’s. We are also expanding in other markets. 20 per cent of our customers now come from Europe.”