​Soda Launch Week

In case you have missed it, from June 9 until June 13, Soda made a series of announcements related to their product. As close partners of Soda and avid users of their software in our customers' analytics projects and element61 acceleration kits, we were delighted to see them launch some great new features in June 2025! To ensure you didn't miss anything, we've summarised the key announcements.

Below you'll find each announcement where we dive into (1) what was announced, (2) what this means, and (3) why we think it's great!

NannyML acquisition

What is this

Soda has joined forces with NannyML—an open-source leader in AI performance monitoring—to create a unified, context-aware data quality and observability platform. Combining Soda’s data-quality checks and observability layer with NannyML’s estimation-based model-performance monitoring, the two teams will deliver end-to-end visibility across data ingestion, transformations, model predictions, and automated decisioning, all within a single platform.

What is the impact

• Closing the gap in modern pipelines: Traditional data-quality tools struggle with today’s dynamic, hybrid batch-streaming, real-time, and AI-driven workflows. They generate noise, miss real issues, and can’t trace downstream impact.
• Holistic impact-aware monitoring: By integrating NannyML’s drift-detection and performance-estimation capabilities, the new platform will not only validate data correctness but also assess and alert on the true business and AI-system consequences of data issues.
• AI-native readiness: As organizations deploy continuous-learning models, LLM-powered agents, and real-time personalization, they need tooling built from the ground up for emergence, drift, and feedback loops—far beyond legacy batch checks.

Why do we like this at element61

• Smarter, lower-noise alerts: Advanced algorithms will surface only the anomalies that truly matter, reducing alert fatigue and helping teams focus on high-impact issues.
• Context-aware root-cause tracing: Quickly follow a single alert from a schema drift in your warehouse through prediction shifts in your models to unexpected behaviors in downstream agents or dashboards.
• Unified observability: One pane of glass for your full data-to-decision lifecycle: see what changed, understand why it matters, and know exactly how to fix it.
• Continued open-source support: NannyML remains fully open and supported—now with even tighter integration into Soda’s broader platform.
• Future-proof infrastructure: Whether using batch analytics, streaming features, or orchestrating LLM chains, you’ll have a foundation built to keep your data integrity and AI behaviors aligned as your systems evolve.

Metrics Observability

What is this

Soda Metrics Observability is an AI-powered anomaly detection system that you can deploy in under five minutes to monitor all your datasets at scale. It delivers immediate insights — no lengthy model training — leveraging a proprietary engine that’s 70% more accurate than Prophet-based methods (cfr. Meta), can process billions of rows, requires minimal configuration, and provides explainable visualizations for every detected anomaly.

Image

What is the impact

• Catches the unexpected: Traditional data tests only validate known issues. Observability uncovers unknown unknowns like schema changes, freshness gaps, partition dropouts, and volume shifts without manual rules.
• Shifts left in your workflow: By spotting issues early—right upon deployment and even backfilled over the past year—you reduce firefighting in production and accelerate time to reliable insights.
• Scales with your data: Proven on Databricks by scanning over 1.1 billion rows in under a minute, it meets the performance demands of large, dynamic data environments where speed and accuracy are critical.

Why do we like this at element61

• Lightning-fast setup & results: Connect your data source, enable the scan, and start seeing anomalies immediately — no tuning or training loops.
• Historical context out of the box: Backfill up to 12 months of data automatically, giving you visibility into past trends and anomalies from day one.
• Explainable, feedback-driven insights: Each alert is accompanied by expected ranges, trend history, and impact assessments. You can mark false positives or confirm true issues, and the system learns from your feedback.
• A path to hardening: Use observability to identify weak spots across all datasets, then codify critical checks and data contracts in Soda — transforming brittle pipelines into resilient, self-healing platforms.

Collaborative Data Contracts

What is this

Soda Collaborative Data Contracts is a unified interface and workflow that lets data producers and consumers jointly define, enforce, and evolve shared expectations for data quality. Whether through a low-code UI or YAML, teams work from the same living contract, turning observed anomalies into upstream checks and preventing issues before they reach production.

Image

What is the impact

• Stop issues upstream. Data problems often originate in source systems but are only detected downstream — in dashboards, reports, or ML models — after they’ve caused damage. Collaborative contracts shift quality gates earlier in the pipeline.
• Bridge business and engineering. Traditional contracts are either too technical for non-engineers or too rigid to adapt. By combining visual workflows (UI) with code-backed definitions (YAML), Soda makes contracts usable for both stakeholders and platform teams.
• Maintain alignment at scale. As data platforms become increasingly complex, maintaining alignment between producers and consumers is crucial. A shared UI and versioned contracts eliminate silos, ad-hoc teams chats and emails, and lost context.

Why do we like this at element61

• Prevent cascading failures. Promote actionable anomalies into enforceable checks with a click, so errors are caught before they propagate.
• One source of truth. All teams — business analysts, data engineers, platform operators — see and maintain the same contract, reducing misunderstandings and rework.
• Flexible deployment. Define contracts from scratch or derive them from existing alerts. Manage them via the Soda CLI or UI, schedule enforcement, version changes, and integrate them into CI/CD pipelines.
• Fast, iterative rollout. Get up and running in minutes, evolve contracts as requirements change, and keep data quality aligned with business needs without bottlenecks.

Transparent Pricing

What is this

Soda is rolling out transparent, self-service pricing for its data-quality platform — no gated demos or opaque quotes. You can view rates online and choose from three tiers designed to match your scale and needs:

What is the impact

• Eliminates buying friction: No more guesswork or back-and-forth with sales — teams can spin up real capabilities instantly.
• Aligns cost with usage: You pay only for the datasets you monitor, not per seat, so costs scale predictably as your data footprint grows.
• Empowers every team size: From solo engineers to large enterprises, the model fits your workflow without forcing unnecessary features or fees.

Why do we like this at element61

• Immediate access & real value: Engineers and small teams get full-featured observability and testing in minutes via the Free Plan—perfect for proofs of concept or side projects that need production-grade quality.
• Predictable, usage-based billing: The Team Plan’s flat rate per dataset ensures budgeting transparency even as you add dozens or hundreds of data sources.
• Enterprise-grade controls: Large organizations benefit from advanced governance — versioned contracts, audit trails, role-based access, SSO, and on-prem options— backed by premium support and volume pricing.
• No surprises: Unlimited users and clear upgrade paths mean you can grow your data-quality practice without hidden fees or license negotiations.
   

​Conclusion

We are definitely excited to put these new features into practice and implement data quality management at scale at our customers! If you want to know more, don't hesitate to reach out to our team!