Data strands, divided
Since its inception, Mode has been for data scientists and analysts. As former data scientists, my cofounders and I started Mode with the intention of building the product we wanted for ourselves. The vision expanded, but the analyst-first philosophy remained. “By analysts, for analysts” became Mode’s informal tagline.
It paid off: Tens of thousands of analysts across 9,000 companies rely on Mode for access to data. More importantly, Mode is the preferred choice for data teams. As Denis Zgonjanin of Shopify told us, “Mode is really the best first and foundational step a data team could make.“
This insistence on building for data scientists, however, required tradeoffs. Because we won’t build anything that slows data scientists down, and because the modeling languages, heavyweight extracts, and cumbersome configurations required to make code-free exploration possible do just that, Mode hasn’t been the best option for non-technical stakeholders.
A similar tradeoff emerged in the market. When asked an urgent question, data teams prefer products like Mode that prioritize speed and in-depth analysis. When building dashboards and reporting suites, teams prioritize analytical “coverage,” the degree to which anyone can slice, dice, and drill into a well-mapped set of data.
And so, in many organizations, the two major strands of analytical work - BI for coverage, and data science for depth and speed - twist and dance around each other, but remain separate.
Two years ago, we started researching an idea for bringing these two worlds together. If we could close a technological gap, we felt that we could enable data scientists to work in Mode just as quickly as they do today, but ship results that provide the same coverage as a BI tool. As we began building, even greater promise emerged: The same technology could also speed data scientists up.
Today, we’re proud to roll out the first piece of technology that makes this possible: Helix.
Here’s how it works: Every query result in Mode is immediately loaded into an on-demand data engine that drives Mode’s visual analytics tools. For data scientists, instead of doing last-mile aggregations and pivots in queries, we can now do those calculations on the fly against curated, granular data into Helix.
It’s less work for more results, faster.
When we share our work, stakeholders have access to the same visual tools on top of the same granular datasets. Rather than each query answering one narrow question, results can be explored and extended by everyone. In other words, it’s coverage, without the setup cost.
Moreover, to avoid the inevitable back-and-forth it takes to answer follow-up questions, data scientists often try to anticipate what we’ll be asked next. With Helix, we can be data scientists, not fortune tellers. Because stakeholders can extend their results, they can immediately get the answers they need and we can focus on new questions.
We didn’t develop Helix just to help our customers make decisions faster; we also made it to help us build faster. With this foundation in place, we’re now building new ways to help more people answer more questions in less time with less effort. We’re excited to get to work on what’s next.