How One of the Leaders in Onchain Lending Tracks Liquidity and Growth with Dune

Spark is an onchain capital distribution and allocation platform: it attracts capital through Spark Savings and partner channels and deploys it across lending markets, DeFi liquidity venues, and institutional credit, alongside borrowing and stablecoin liquidity infrastructure.

In Q2 2026, the Spark Liquidity Layer allocated an average of $2.56 billion across approved venues, part of $7.96 billion in total value locked. Spark uses analytics to track not just how much capital it manages, but where that capital comes from, how it's deployed, whether liquidity is sufficient, and how much profit remains after funding and operating costs.

Spark relies on Dune as a core blockchain data provider, using Dune's Data Hub to build much of its financial and operational reporting. Its reporting draws on Dune's lending, stablecoin liquidity, and revenue and asset data. Spark's Data Analytics team runs more than 40 dashboards, organized around four areas:

• Protocol operations: points incentives and partner campaigns
• Market intelligence: user activity and market trends
• Risk and liquidity monitoring: large withdrawals, utilization, and liability tracking
• P&L and projections: the single source of truth for Spark's financial reporting

Open Spark Liquidity Layer Dashboard on Dune

Spark's data analytics team used Dune's materialized views to build each of those dashboards, reusing underlying tables across them. But Spark's P&L calculations run through multiple layers of intermediate models, and as those pipelines grew, they needed dependency-aware orchestration to refresh the upstream and downstream tables reliably and in the right order. This is when Spark adopted Dune's dbt Connector, which made those dependencies explicit and let Spark run its complex P&L pipelines at an hourly cadence.

With Dune's dbt Connector, Spark manages its models and their dependency graph in Git. That also makes the full pipeline directly accessible to AI tools, rather than needing to retrieve individual queries one at a time.

“

"Dune's customer success team was instrumental in setting it up, helping configure our pipelines and add automated checks that catch errors before they reach a dashboard." — Walle, Data Analyst, Spark

As a result, Spark estimates that it spends approximately 25% less time on its reporting workflow overall. The gain came mainly from managing dbt models and their dependencies in Git, which lets Spark's team and its AI tools inspect the full model logic and lineage, quickly identify which upstream and downstream models are affected, and make coordinated updates without manually tracing dependencies through individual queries.

“

"With Dune's dbt Connector, we can see the full dependency graph in Git, so when something changes upstream we know exactly what it affects downstream." — Walle, Data Analyst, Spark

A representative example is adding a new product, such as spUSDG, across approximately 25 models. Manually identifying and updating the affected models previously took around 90 minutes. With Cursor or Codex scanning the dbt code and dependencies in Git, the same work now takes approximately 30 minutes — a 67% reduction.

“

"Before, adding a product such as spUSDG across approximately 25 models took around 90 minutes of manual searching and updates. Now Cursor or Codex can scan the pipeline, and I can complete the same work in about 30 minutes." — Walle, Data Analyst, Spark

This AI-accessible pipeline means means Spark's team spends less time rebuilding reports and more time acting on what those reports show: where to deploy capital, whether liquidity is sufficient, and how that affects profitability.

dbt Connector is available to Dune Enterprise customers. Talk to our team to get access.