Feb 22, 2024 · 47m · mad
Future of The Modern Analytics Stack | Tristan Handy, CEO of dbt
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The MAD Podcast, host Matt Turck interviews dbt Labs Co-founder and CEO Tristan Handy to explore the evolution of the modern data stack, the rise of the analytics stack, the integration of generative AI in data engineering, and the operational realities of scaling an open-source business.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 22.6% of the talking time here. How this is scored →
Matt as informed peer 3.8 Guest teaching 5.3 Guest disagreement 0.9 Matt pushing back 1.1
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 6:21 Guest playfully accuses VCs of inflating the ecosystem
Tristan jokingly singles out VCs like Matt as the root cause of category overcrowding, remarking 'Yeah, VCs. It's your fault' when asked what broke the modern data stack.
Hardest push from Matt ▶ 15:36 Host challenges best-of-breed tooling against platform consolidationMatt pushes back on the viability of unbundled tools by citing Databricks and Snowflake expanding functionally into catalogs and governance to capture the entire stack.
Biggest teaching moment ▶ 9:14 Guest explains why the 'Modern Data Stack' definition is technically deadTristan educates the host on how legacy vendors like Tableau adapted to the cloud over eight years, rendering the original technical distinction of modern data stack obsolete.
Matt holds his own ▶ 19:09 Host demonstrates insider knowledge on Reverse ETL market shiftMatt displays deep domain knowledge by noting that Reverse ETL startups were forced to rebrand into Customer Data Platforms, prompting Tristan to ask if Matt was an investor in the space.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome and Podcast Cross-Over Discussion | 2 | 1 | 1 | 0 | Matt and Tristan open with casual banter about a podcast cross-over confusion and Tristan's recent blog post questioning the utility of the modern data stack term. Matt sets an informal tone and references his own annual data landscape report. | |
| Origins and History of the Modern Data Stack (2012–2019) | 2 | 6 | 0 | 0 | Tristan educates the host on the early timeline of the modern data stack, tracing its birth to Amazon Redshift in 2012 and explaining how cloud-native tools like Looker displaced pre-cloud tools like Tableau. | |
| The VC Funding Boom and Category Overcrowding | 4 | 5 | 2 | 1 | Tristan playfully points the finger at venture capitalists for overfunding the category post-Snowflake IPO, leading to market overcrowding. Matt agrees from a VC perspective, acknowledging that infrastructure suddenly became overly hyped. | |
| Why the 'Modern Data Stack' Term Has Lost Utility | 2 | 7 | 2 | 1 | Tristan lays out two key arguments for why the 'modern data stack' term is obsolete: legacy tools have adapted to the cloud, and buyers no longer want to stitch together ten best-of-breed tools. | |
| Shifting to 'The Analytics Stack' and Lessons from Agile | 3 | 6 | 1 | 1 | Matt asks whether the underlying functional methodology remains valid despite the term's decline. Tristan compares the modern data stack movement to Agile methodology, proposing a pivot to the simpler term 'Analytics Stack'. | |
| Platform Consolidation vs. Best-of-Breed Tools | 5 | 6 | 1 | 3 | Matt pushes Tristan on whether major platforms like Databricks and Snowflake are functionally expanding to squeeze out standalone tools. Tristan breaks down hyperscaler incentives, arguing cloud providers care mostly about compute and storage consumption. | |
| Category Evolution and the Case of Reverse ETL | 6 | 6 | 1 | 2 | Matt demonstrates sharp sector knowledge by raising Reverse ETL's transition into Customer Data Platforms (CDPs). Tristan validates this insight and explains the acquisition challenges caused by heavy platform integration work. | |
| Integrating Generative AI into Data Transformation Workflows | 4 | 5 | 1 | 1 | Matt asks if AI is a friend or foe to data infrastructure companies. Tristan explains that while valuation multiples shifted to AI, LLM integration brings pragmatic gains in code authoring, documentation, and regex generation inside dbt. | |
| Software Engineering Principles and the Future of Analytics Engineers | 4 | 6 | 2 | 2 | Matt asks if analytics engineers will become prompt engineers. Tristan reframes the thesis, arguing that data engineering is software engineering and AI will function like an accelerated compilation feedback loop rather than replacing engineers. | |
| Enterprise AI, RAG Architectures, and the Value of Structured Data | 6 | 5 | 0 | 1 | Matt highlights enterprise AI architectures like RAG and vector databases. Tristan agrees, explaining how dbt-curated structured data powers internal AI support agents and improves resolution speeds. | |
| Recent dbt Releases: dbt Mesh and the Semantic Layer | 3 | 6 | 0 | 0 | Matt asks about dbt's recent product launches. Tristan explains multi-project refactoring challenges at scale and how the dbt Semantic Layer standardizes business metric definitions across disparate BI tools. | |
| Future Product Roadmap: Low-Code/No-Code and Data Catalogs | 3 | 5 | 0 | 0 | Tristan outlines dbt's upcoming roadmap, including combining AI with the Semantic Layer, adding low-code visual interfaces, and launching an affordable data catalog experience. | |
| Functional Expansion and Founder Executive Leadership Transitions | 5 | 5 | 1 | 2 | Matt asks about executive team turnover, open source monetization friction, and sales tactics in a tighter economic climate. Tristan reflects transparently on replacing most of his executive team with experienced scale-stage leaders. |
Statements from this episode (22)
Handy: Looker and Mode were superior to pre-cloud Tableau
“I had a strong preference to use tools like mode or looker over a tool like tableau because tableau was pre cloud and it assumed that you were going to be able to download all the data that you needed to operate on into your local cache.”
Handy: The core modern data stack architecture remains a good idea
“When people started using the term modern data stack, there was a there, like fivetran, dbt, mode, looker, et cetera. These are tools that. Were built around an idea of how data technology should work. And it was a, it was, and is a good idea.”
Handy: Venture capital created too many data startups too quickly
“There was just, there were too many companies created too quickly.”
dbt CEO Tristan Handy: The modern data stack concept is dead
“I don't think the modern data stack is a useful idea anymore.”
Handy: CDOs shouldn't build data platforms by integrating nine point products
“If you say I want a modern data platform as a CDO, like the right answer is not go out and buy nine different products and integrate them together.”
Handy: The shift from ETL to ELT will remain true forever
“One of the kind of trends inside of this is the transition from ETL, extract, transform, load to ELT, extract, load and transform. And that seems to anyone who's not in data that Might not seem like a big thing, but in fact, it's like a really significant tran…”
Handy: dbt Labs and Fivetran are not going anywhere
“And so that means that companies like us and companies like Fivetran that are big parts of that, you know, we're not going anywhere.”
Handy asked dbt marketing to adopt 'the analytics stack' label
“My big conclusion is I sent an email to our product marketing, our head of product marketing. I was like, Hey, can we just say the analytics stack? Like this is a set of technologies that works together to do analytics. And it kind of tries to get away from th…”
Handy: Cloud platforms care about compute consumption, not partner revenue
“My read is that people like us, people like Fivetran, while we are critical to their success as platforms, because we drive a ton of consumption. They don't actually really care about the dollars that we make as businesses. Cause they're so tiny relative to th…”
Handy: Hyperscalers build features for RFPs, not to kill independent tools
“When you talk to the folks at the hyperscalers, they will say, we are, we're like making sure that our solution is like covers all these different areas, but it's mostly not. Because we want to compete in these areas. It's mostly because people come to us with…”
Handy: Reverse ETL tool value accrues to sales/marketing, not data teams
“It turns out that the value of a reverse ETL tool does not accrue to the central data team. It accrues primarily to. Sales and marketing teams who get the data.”
Handy: Building in-house is often faster than M&A for data infrastructure
“Had to build, you know, there's like the iceberg, which is 10% above the water and you can see it. And then there's 90% below the water for a data company. The platform is the below the water part. And then the functionality that you build on that platform is …”
Tristan Handy: Software multiple expansion has shifted from modern data stack to AI
“If what you were trying to do is maximize the multiple that your software company was trading at, then AI is the enemy because all the multiple expansion has moved from modern data stack to AI.”
dbt Labs has internal prototypes for AI-generated models, tests, and documentation
“We've already got internal prototypes working of generate me a model that does this or generate me tests for this model or you know, write documentation for, so we've got all of this stuff that is going to be making its way through at some point.”
Tristan Handy: Building a production data system is building a software system
“One of the core beliefs about the, that we have about the profession is that data and software are not that different. And when you're building a production data system, you're building a production software system, which means that we should be taking lessons…”
Tristan Handy: The semantic layer centralizes meaning like data platforms centralize data
“The semantic layer in the same way that like a cloud data platform centralizes data, so you can all use the same data. The semantic layer centralizes meaning. How do you actually analyze that data to produce a particular business metric?”
Handy: dbt will start exploring visual, GUI-based workflows this year
“And so this year will probably be the year where we start playing around in that space.”
Handy: Low-code data tools must read and write underlying code
“You have to get no code, low code experiences that read and write code so that code can go through a get PR process and it can have all the same mature stuff built on top of it as any other code.”
Tristan Handy: dbt had 1,000 company users before raising venture capital
“We waited, waited a long time to raise any venture funding at all. We had a thousand companies using the product before we raised a single cent.”
Tristan Handy: dbt replaced almost its entire executive team within one year
“If you look today the only people who are consistent presence on our executive team from a year ago are me and our CFO and everybody else around the table is, is new.”
Handy: Social media pricing backlash did not cause mass dbt cancellations
“Our business as a result has not really had a hiccup. But if you only existed on Twitter or LinkedIn you would think that you know, our customers desired us on mass and we're every single one of them was livid and canceled, but that just like didn't happen.”