Reverse ETL, BI, and AI with Brian Kotlyar

Allan Wille, CEO at Klipfolio, sat down with Brian Kotlyar, VP of Marketing & Growth at Hightouch, a reverse ETL data activation platform built to help data teams activate the insights already sitting in their warehouses. Their conversation covered the potential of reverse ETL, the real challenges facing modern business intelligence (BI), and what happens when AI meets data analytics.

What is reverse ETL?

Reverse ETL moves processed data from your data warehouse back into the operational tools your teams use every day — CRMs, marketing platforms, customer success tools. Standard ETL pulls raw data in and transforms it for analysis. Reverse ETL pushes the results of that analysis back out where decisions actually get made.

Brian has seen this problem from both sides for most of his career:

"Reverse ETL actually solves a problem that has plagued me for literally the entirety of my career. My first internship working at Staples as a teenager, we were working on customer 360 initiatives and single source of truth type stuff in traditional on-prem data warehouses and struggling with customer identity and how to activate it — all the way to almost 25 years later working at New Relic. What reverse ETL really is about is, as analytics teams and operations teams, we put all this effort into gathering data and processing data to understand our business, understand our customers. But we consistently seem to run into this brick wall of actually using the stuff."

That brick wall — the gap between insight and action — is what Hightouch was built to break through.

Why data activation is so hard

Most data teams are good at collecting and transforming data. The harder problem is getting clean, trusted data into the hands of the people who need it, in the tools they already use.

Brian points to a few reasons this keeps failing:

• Identity resolution is unsolved. Customer identity resolution — stitching together a complete picture of a customer across touchpoints — has been a challenge since the early days of CRM, and it remains one today.
• Warehouses weren't built for activation. They're built for analysis. Getting data back out into operational systems requires a separate layer — which is exactly what reverse ETL provides.
• Operational teams can't wait for data teams. When a sales rep needs enriched account data in their CRM, they need it now, not after a ticket is filed and a pipeline is built.

The state of business intelligence

Brian and Allan also explored where BI is heading. The traditional model — analysts build dashboards, stakeholders read them — is showing its age. The expectation is shifting toward self-serve business intelligence: business users want to ask questions and get answers without waiting in a queue.

The challenge is that self-serve BI only works when the underlying data is trustworthy and well-defined. Inconsistent metrics, conflicting numbers, and unclear definitions undermine confidence fast. That's why data governance semantic consistency has become central to any modern data strategy.

Reverse ETL fits into this picture by ensuring that the metrics defined and trusted in the warehouse are the same ones showing up in downstream tools. One definition. One number. Everywhere.

AI and the future of data analytics

The conversation turned to AI, and Brian was direct: AI doesn't fix bad data. It amplifies it.

For AI to deliver useful answers, it needs structured, well-described, unambiguous data. That means the work of metrics definition data trust, establishing governance, and maintaining consistency isn't going away — it's becoming more important. AI can handle the language layer; the data layer still needs humans to get it right.

Brian sees the biggest opportunity at the intersection of AI and operational data activation: using AI to surface the right insight, to the right person, in the right tool, at the right moment. That's a much higher bar than generating a dashboard. It requires data that's not just accurate but contextualized and activated.

Key takeaways

• Reverse ETL closes the activation gap between analytical insight and operational action.
• Data trust is foundational. Self-serve BI and AI both fail when the underlying data is inconsistent or poorly defined.
• Customer identity remains one of the hardest problems in data, even decades after teams first started working on it.
• AI raises the stakes for data quality, not the reverse. Better-structured data means better AI outputs.