Don't start an analytics company

Business success is mostly about timing. In technology, an underlying innovation drives a decade-long wave of business outcomes built on that innovation. The internet became generally available in 1991 and drove a wave of Internet 1.0 outcomes that accelerated until the 2001 crash. The iPhone launched in 2007 and powered another wave of outcomes as businesses figured out what was possible with our new pocket platforms.

WebVan, a high-flying grocery delivery startup, famously flamed out in 2001. Instacart was founded in 2013 with exactly the same premise but with the benefit of iPhones so customers can order without having to sit at a computer. It’s been a massive success. Timing.

In 2006, Amazon kicked off the cloud computing boom with the launch of AWS. Since then, virtually all business software has moved from on-premises datacenters to the cloud. That created a massive opportunity for startups to displace incumbents as buyers re-evaluated virtually all of their software products.

One little corner of that massive migration was the data warehousing and business intelligence markets. Six years after the advent of cloud computing, Snowflake and Redshift were launched as the first cloud data warehouses. Their cloud-enabled speed and scalability drove business to abandon their on-premises Vertica and Netezza warehouses and adopt Snowflake and Redshift.

This set up the next dominoes to fall: The tools that sit on top of the warehouse, especially business intelligence tools. Legacy BI products like MicroStrategy and Business Objects had to be deployed on-premises, and were optimized to work around the constraints of on-premises warehouses. To compete, a new generation of cloud BI companies were born. They required zero deployment, took better advantage of Snowflake’s and Redshift’s cloud computing power, and integrated with more modern cloud tooling. 

In classic tech company fashion, the exits of those cloud BI companies came nine years after their founding and followed a power law: The winner got a Cadillac, the runner-up got a set of steak knives, and further exits were even smaller. But most importantly, the reason that this generation of BI startups worked – the reason there were any successful outcomes at all – was that these businesses were created and built during an era of mass customer migration from datacenters to the cloud. Timing.

In many ways Modelbit is a spiritual successor to Periscope: A technical product serving the advanced user segment of a market that’s still small but starting to make real noise. And just like last time, there’s a critical dynamic that’s irresistibly attractive to two founders who are hackers at heart: A big, exciting technical trend. In this case, a research breakthrough has led to the deployment of fleets of cloud GPUs powering the use of machine learning for, well, everything. Timing.

Meanwhile, in the years since Periscope Data, I get a call every month or two from founders starting a business intelligence company and asking for my advice and support. My answer is that the last war has already been fought. The mass migration happened, followed by a wave of consolidation. These days, customers have locked in their cloud BI tools, and those tools are increasingly becoming features of their cloud corporate owners

To those founders, I ask: Why now? What massive technological trend is prying this market loose and making all the buyers flock to new solutions? Why is this a time for Instacart, and not a time for WebVan? It’s a hard question. Clarity about the market moving trends that change our lives is usually only available when it’s all over. But this is your life, and you only get to live these years once. If you don’t at least have a credible hypothesis, don’t start an analytics company.