Key results:

• 41% savings through automation: Rabbit’s automated BigQuery Autoscaler dynamically adjusted capacity, cutting waste and lowering costs by 41%.
• Engineering Time Savings: Eliminated need to build and maintain custom dashboards and reports.

About Daangn

Daangn, operator of Karrot, is one of the fastest-growing community-driven marketplaces in Asia, connecting tens of millions of users to neighbors, goods, and local services. At its core, the platform relies on massive-scale data processing: recommendation engines, fraud detection, and platform monitoring all depend on BigQuery. As usage scaled, so did costs — making BigQuery one of Daangn’s largest infrastructure line items.

For the engineering team, the challenge wasn’t only financial. Managing slot allocations, catching anomalies, and reconciling commitments had become a full-time operational burden. They needed a solution that could restore cost predictability and efficiency — without slowing product innovation.

“BigQuery’s built-in autoscaler left us consistently paying for capacity we didn’t use. The minimum 60-second downscaling window was a hidden tax. Rabbit’s Autoscaler keeps BigQuery usage tightly aligned with demand, cutting unnecessary capacity and saving us hundreds of millions of won each month.

The Challenge — Hidden waste without visibility

When Daangn’s engineering team scaled their use of BigQuery, their priority was performance and reliability. Queries had to run fast, and workloads had to scale without friction. On the surface, Google Cloud’s native autoscaler seemed to provide exactly that.

What they didn’t realize at first was the silent inefficiency built into BigQuery’s native autoscaler. Autoscaling happens in fixed 50-slot increments with a minimum 60-second billing. When short, spiky workloads triggered sudden slot demand, the autoscaler would scale up unnecessarily, leaving large portions of those slots unutilized. This created hidden waste that accumulated across the reservation. Rabbit solved this by dynamically adjusting the maximum slot setting in real time. Rabbit ensures that critical, high-priority pipelines always have the power they need, while idle or low-activity periods are handled cost-efficiently.

This hidden tax compounded quickly. With millions of queries running every day, slot waste turned into hundreds of thousands of dollars in unnecessary spend each month. Before Rabbit, Daangn’s engineers weren’t aware of the scale of the waste — it simply went unnoticed. Rabbit made it visible and gave them the tools to act on it.

See how much BigQuery waste your team could cut.

Calculate your waste →

The Solution — Deep cost visibility with automated optimisations

Daangn’s data engineering team brought in Rabbit to replace high-level billing exports with granular, workload-aware cost intelligence. The rollout was seamless — no re-architecting pipelines, no downtime — yet it gave engineers the kind of observability that had been missing since day one.

Rabbit provided query-level cost attribution, automatically mapping every BigQuery job to its actual slot consumption and financial impact. Engineers could finally distinguish exploratory ad-hoc queries from latency-sensitive production pipelines, revealing which workloads were driving disproportionate spend.

The platform also embedded real-time anomaly detection at the pipeline level. Instead of combing through delayed billing data, Daangn’s engineers received proactive alerts on runaway queries, anomalous job retries, or reservation imbalances as they happened — enabling intervention before costs escalated.

Finally, Rabbit’s in-depth cost insights revealed the extent of waste in BigQuery reservations, highlighting significant potential savings. The Daangn team gradually enabled Rabbit’s automated optimization features to manage Autoscaler waste, resulting in a substantial reduction in their monthly costs.

The Results — Sustainable Savings and Engineering Efficiency

The impact of Rabbit was immediate and measurable. Within just the first 23 days of the free PoC, Daangn achieved a 22% reduction of BigQuery spend. What started as an experiment quickly proved to be a structural shift in how the team managed cloud economics.

Rabbit’s BigQuery Autoscaler optimization transformed slot-hour efficiency across three major reservations. By reducing the waste on the reservations, utilization improved by 37–46%, unlocking 41% in recurring monthly savings.

But the financial results were only half the story. Rabbit also freed up a significant amount of engineering hours every month — the equivalent of nearly a full engineering team — by automating anomaly detection, query-level cost attribution, and spend reconciliation. What once consumed days of manual SQL auditing and billing reviews became an automated background process.

For Daangn, Rabbit didn’t just lower the cloud bill — it embedded cost governance into their engineering culture. Every query, every reservation, every pipeline now runs with transparent economics, enabling the team to focus on product innovation while maintaining tight cost control.

Start your 30-day exploration of Rabbit and see how automation can free your team to build.

Start your free trial →