2
Disruption Is Cool Until It Happens to You

Amazon's Big Move

At RJMetrics, by the time we realized we were losing, we had already lost. The same pace of market innovation that made our product a hit had continued under our feet so rapidly that it made us a dinosaur just as fast.

The first domino to fall came from Amazon Web Services (AWS). In late 2012, they announced a new product called Redshift (see Figure 2.1). It was a “fast and powerful, fully managed, petabyte-scale data warehouse service in the cloud.”

Remember that cool custom build of MySQL that we developed at RJMetrics? The one that sat right at the center of our technology offering and served as the central location where data was stored, processed, and analyzed? As it turned out, our data warehouse was the horse-and-buggy version of Redshift's Model T.

A study published by SiSense around that time (see Figure 2.2) summed it up plainly: “On average, Redshift was 500x faster than [traditional databases] for metrics like Daily Revenue, Daily Active Users, and Daily ARPU [average revenue per user].”

Yikes. These kinds of “metric queries”—ones about revenue, user activity, and average revenue per user (ARPU)—were precisely the kinds of metrics you could find in RJMetrics. Now you could get them 500x faster if you stored your data in Redshift instead.

FIGURE 2.1  Amazon Redshift announcement release, November 2012.

FIGURE 2.2  SiSense benchmarking study outputs, 2015.

Note: “RDS Postgres” is a cloud-deployed competitor to MySQL that was comparable to the RJMetrics data warehouse at the time.

We convinced ourselves that this new technology, while remarkable, wouldn't affect us. Why? Because the data warehouse was just one piece of the puzzle. BI software like ours had a lot more to it. Amazon didn't offer any data pipeline software to extract, transform, and load (ETL) data into its warehouse, nor did they offer a reporting layer for building out your analytics and presenting them to users. Those were big gaps to fill. Weren't they?

Well, as it turned out, a 500x performance improvement, combined with the ease of setup that came with AWS cloud deployments, proved to be enough to drive a major paradigm shift in how companies did analytics. In under three years, Redshift went from a new product to “the fastest growing service in the history of AWS” (see Figure 2.3).

Redshift proved most popular among engineers, data scientists, IT leaders, and professionals in the emerging field of “data ops,” all of whom saw tremendous value in keeping control over their company's data. By having a centralized data warehouse located in your AWS cloud and managed by your own ops team, you could exercise unprecedented control over data usage, consistency, cost, and more. It was an exciting proposition for enterprises on countless dimensions.

Cut back to RJMetrics. Here's what we started hearing on sales and renewal calls:

• “Yeah, this looks great, but our data team is telling us that we already have all this data in a data warehouse over at AWS. Why should we also pipe all that same data outside of our cloud over into yours so you can analyze it?”
• “We set up RJMetrics, but the results are just a little different from what our IT people see when they query our Redshift warehouse. Can you do an audit to get those reconciled?”
• “Can we just use your charts and dashboards on top of our warehouse instead?”

I recently caught up with Vijay Subramanian, who was chief analytics officer and head of growth at Rent the Runway. They were a marquee customer of ours at RJMetrics who churned off of our platform in this pivotal era.

“RJMetrics was used to report on our key user and growth metrics, but as we started tracking other parts of the business, we felt the need to pipe it all into a common warehouse where we could manage more of the nuanced business modeling,” he reminisced. “RJMetrics struggled to serve our needs as it was an all-in-one solution.”

FIGURE 2.3  The Register headline, April 2015.

FIGURE 2.4  When we lost the warehouse, we lost our way.

Our data warehouse was a loser (see Figure 2.4). And when we lost the warehouse wars, it literally snapped our value proposition right down the middle. The other two ends of our product just couldn't hold up the product-market fit chasm in the middle.

The Modern Data Stack Is Born

Former Netscape CEO Jim Barksdale famously said that there are “only two ways to make money in business: one is to bundle; the other is unbundle.” Between 2013 and 2016, we experienced a radical unbundling in the way data was moved, processed, analyzed, and consumed by modern businesses. In the process, it forged one of the most powerful and valuable partner ecosystems ever to exist, exposing me to the ecosystem-led growth playbooks that would change my career forever.

Within a year of Redshift's release, the traditional business intelligence approach of an all-in-one analytics stack had been split apart. In its wake, a new generation of companies emerged that took on the rest of the stack piecemeal.

Depending on how you count it, somewhere between five and ten new categories of software and services seemed to emerge overnight. These companies each took over a slice of the old all-in-one stack and, when combined, gave end users a “modern data stack” that could be seamlessly integrated to provide more power, flexibility, and affordability than anything that came before.

Data warehouse innovation didn't stop with Redshift. Google and Microsoft soon answered with BigQuery and Azure Data Warehouse, respectively. Later, Snowflake's powerful data warehousing offering won it a dominant position as one of the fastest growing and highest valued SaaS companies ever with a peak market value of $118 billion.

ETL platforms such as Fivetran and Matillion arrived, whose main purpose was to move data between leading SaaS platforms and the warehouses listed above. In a 2021 funding round, Fivetran was valued at $5.6 billion.

Reporting platforms narrowed their scope to focus on nailing the data modeling and dashboarding capabilities. This is where Looker focused. In the modern data stack era, a wave of new vendors came to market in this space including Looker, Mode, Periscope, Redash, Hex, and Omni. Legacy application players such as Tableau also entered the space with more robust cloud offerings. In 2019, Tableau was acquired by Salesforce for $15.7 billion and Looker was acquired by Google for $2.6 billion.

Other layers surrounding these tools—such as transformation workflow solutions including dbt (“data build tool”)—have also emerged and come to meaningful scale. In a 2022 round of funding, dbt Labs was valued at $4.2 billion.

In addition to software innovation, an entire industry of services businesses, agency partners, and system integrators emerged around this market as well (see Figure 2.5). Especially with enterprise deployments, the value of human experts to pull these pieces together, train teams, and deploy best practices are more important than ever.

Just incredible. So many new markets that they're hard to count, each with new incumbents that didn't exist a decade ago—all valued at billions.

And with that, a new industry emerged with a bang, and RJMetrics would be sold off with a whimper.

The Ecosystem Effect

It's now clear as day how Looker grew to 100x our value in half of the time: they grew as part of an emerging and highly disruptive ecosystem while we flew solo.

“The partner ecosystem was one of our superpowers,” Keenan Rice shared with me in a recent conversation. Rice was a member of Looker's founding team and served as a vice president of Sales, Marketing, and eventually Global Partnerships at the company. “We found our place developing and coleading a powerful ecosystem at a critical time in the larger data sector, and those partner relationships catapulted us to the top of our category.”

Looker didn't just make its partners' products more valuable—its use was predicated on their use. It made co-selling and using partner products an existential requirement for its customers. In turn, the same was true in the opposite direction. They built together, they sold together, and they won together.

Meanwhile, any company still trying to go solo fizzled out. At RJMetrics, what felt like a strategic advantage—we were a one-stop shop, the only thing you would need—ended up being our downfall.

The tale of the modern data stack and its disruptive effects on incumbents such as RJMetrics is not a one-off. In fact, it was just a canary in the coal mine of the entire software industry.

Data has become more structured, portable, and well-suited to product integrations, and in turn the DNA of the software value proposition has been rearranged. Who you partner with, and how, has gone from a curiosity question to a core tenet of how purchasing decisions get made and value gets delivered.

FIGURE 2.5 The...