Snowflake went public this week, and in a sign of the broader ecosystem evolving around data warehousing, a startup that has come up with an entirely new approach to modeling warehoused data is announcing funding. Narrator – who uses an 11-column ordering model instead of a standard star schema to organize data for modeling and analysis – has received a Series A round of $ 6.2 million that it plans to use to help users for a to start and build self-service version of its product.
The financing is led by Initialized Capital along with other investments from Flybridge Capital Partners and Y Combinator – where the startup was in a 201
Narrative has been around for three years, but its first phase was to provide modeling and analysis directly to companies as consulting firms and to help companies bring together different, structured data sources from marketing, CRM, support desks and internal databases in order to act as Unit to work whole. As advisors, the company’s CEO Ahmed Elsamadisi used an earlier version of the tool now being launched and said he and others each asked questions “single-handedly for eight large companies” during deep dive analysis performed by someone else.
Now that it’s confirmed to work, the new self-service release is designed to provide data scientists and analysts with a simplified way to organize data so that queries that are referred to as actionable analysis in a story-like format – or “narrative” like that Company names them – can be created quickly and consistently across this data – hours instead of weeks. (Below is a demo of how the company’s data leader, Brittany Davis, works.)
(The price for the new Data-as-a-Service is also available in SaaS tiers, with a free tier for the first 5 million lines of data and a subsequent tiering of prices based on data lines, user numbers and narratives in use.)
Elsamadisi, co-founder of the startup with Matt Star, Cedric Dussud, and Michael Nason, said that data analysts have long lived with the problems of modeling star schemes (and thus the associated format of the snowflake scheme), which can be summarized as “layers of dependencies , lack of source of truth, mismatched numbers, and endless maintenance, “he said.
“When you create a lot of tables out of much complex SQL, at its core you have a growing house of cards that has to keep hiring more people to make sure it doesn’t break down.”
(We) work experience
While working as a senior data scientist at WeWork – yes, he told me, it might not actually be a technology company, but it had “technology at its core” – he had a breakthrough in realizing data restructuring to get around these issues.
Before that, things were difficult on the data front. WeWork had 700 tables that his team managed using a star schema approach, covering 85 systems and 13,000 objects. The data includes information about building acquisitions, customer flows through those buildings, how things are changing and how customers might change, marketing and social networking activities, etc. that are in line with the rapidly growing company of the company grow. All of that meant a mess at the end of the data.
“Data analysts would not be able to do their job,” he said. “It turned out we couldn’t answer basic questions about sales. Nothing matched and everything was taking too long. “
The team had 45 people but had to implement a hierarchy for answering questions because there were so many and insufficient time to search through and answer them all. “And we had every data tool out there,” he added. “My team hated everything they did.”
This narrator’s single-table column model Uses, he said, “had been theorized in the past” but not found.
The spark, he said, was to think of data structured the way we ask questions, where – as he described it – each element of data can be bridged together and then used to answer multiple questions as well.
“The main difference is that we are using a time series table to replace all of your data models,” Elsamadisi explained. “This is not a new idea, but it has always been considered impossible. In short, we’re solving the same problem as most data companies to make it easier to get the data you want. However, we are the only company that is solving this problem by innovating the data modeling approach at the lowest level. Honestly, that’s why our solution works so well. We have rebuilt the data foundation instead of trying to improve on a flawed foundation. “
Narrator calls the composite table that contains all of your data reformatted to fit into its 11 column structure, the activity stream.
Elsamadisi said it takes about 30 minutes to use Narrator for the first time and about a month to learn how to use it thoroughly. “But after that, you don’t go back to SQL, it’s so much faster,” he added.
Narrator’s original market was to provide services to other tech companies, and startups in particular. However, there are plans to open these up to a much larger number of industries. In a move that could help in the long term, the company also plans to make some of its core components open source so that third parties can push products through the framework faster.
Speaking of competitors, he says that, in essence, they are the tools he and other data scientists have always used, even though “we’re violating a best practice approach (star schema), not a company.” Airflow, DBT, Looker’s LookML, Chartios Visual SQL, and Tableau Prep are all ways to create and use a traditional star schema, he added. “We are similar to these companies and we try to make creating the tables required for BI, reporting, and analysis as simple and efficient as possible. However, these companies are limited by the traditional star schema approach.”
So far the evidence has been in the data. Narrator says that cCompanies average about 20 transformations (the unit used to answer questions) compared to hundreds in a star schema, and that these are trThe information averages 22 lines compared to more than 1000 lines in conventional modeling. For those learning how to use it, the average time to create a report or run an analysis is four minutes, compared to weeks for traditional data modeling.
“The narrator has the potential to set a new standard for data,” said Jen Wolf, COO of Initialized Capital and Partner and new board member of Narrator, in a statement. “We were amazed at the quality and speed with which Narrator delivered analysis on their product. We are confident that data analysis will be taught in the future once the world experiences Narrator. “