Explorium, a platform that automatically collects external data to improve users’ machine learning models, has raised $75m in Series C funding.
Less than a year after securing $31m, the startup has completed another funding round spurred by unpredictability caused by the pandemic.
Explorium analyses data models, searches its collection of thousands of external datasets and automatically pinpoints the most relevant ones for a user to improve their analytics and machine learning.
During the pandemic, businesses found that their predictive models had become obsolete. In-house historical data could no longer accurately forecast changes or behaviour in the market because of the sudden and dramatic shift to people’s lives.
Businesses had to turn to external data, which they could use to enrich and improve their forecasts. However, a report by Explorium showed that 93% of businesses it surveyed said that finding relevant data took ‘high’ or ‘medium’ effort, while 81% said they had spent at least $100,000 a month on acquiring external data.
Venture capital firm Insight Partners, which led the latest round of investment, called data “the new differentiator”, adding that simply using predictive models is no longer enough.
George Mathew, managing director at Insight Partners, said: “AI and machine learning are already table stakes. Everyone has them. Competitive advantage will depend not just on the quality of your models but on the diversity of data fuelling those models, making Explorium a unique proposition for data scientists and analysts alike.”
In property, Explorium has a client that buys and sells houses, whose in-house pricing model – which included historic pricing data and size – often undervalued properties. By plugging in external datasets, such as air quality ratings, local food ratings and average local business closing times, Explorium said the company raised its margins from below 10% to 14%.
Maor Shlomo, CEO at Explorium, said: “As we saw last year, machine learning models and tools for advanced analytics are only as good as the data behind them. And often that data is not sufficient.
“We’re addressing a business-critical need, guiding data scientists and business leaders to the signals that will help them make better predictions and achieve better business outcomes.