D23's State of Managed Apache Superset: A Q1 2026 Review

Where We Started and Why It Matters

When we launched D23, the premise was straightforward but underserved: teams building with data needed a managed Apache Superset platform that didn’t require them to become infrastructure engineers. The market had alternatives—Looker, Tableau, Power BI—but they came with licensing costs that scaled aggressively and architectural constraints that made embedding analytics or building API-first BI nearly impossible without significant engineering overhead.

Apache Superset, the open-source foundation we build on, had solved the core visualization and exploration problem. But running Superset at scale demanded DevOps expertise, database tuning, security hardening, and the kind of operational burden that pulled engineering teams away from shipping product features. We built D23 to eliminate that friction.

Q1 2026 marks a critical inflection point. We’ve shipped meaningful features, seen clear patterns in what customers actually need, and refined our roadmap based on real production workloads. This review covers what we’ve built, what we’ve learned, and what’s coming next.

Q1 2026 Shipped Features: The Concrete Wins

Text-to-SQL and AI-Powered Query Generation

Our biggest Q1 release was native text-to-SQL integration using LLM-powered query generation. This wasn’t a checkbox feature—it was a response to the single most common customer request: reducing the barrier for non-technical users to explore data without writing SQL.

Here’s what this actually does. A business analyst types a question in natural language—“What’s our churn rate by cohort for Q1?”—and the system generates the appropriate SQL, executes it, and returns results. The LLM understands your schema, respects column naming conventions, and learns from corrections. After a few interactions, accuracy improves substantially.

We’ve integrated this with dbt workflows so that dbt metadata flows directly into the LLM context. That means the AI understands not just table structure but semantic meaning—the difference between a fact table and a dimension, what a metric actually represents, and how joins should work. This reduces hallucinations and speeds up time-to-insight.

Metrics from early adopters: average time-to-first-query dropped from 15 minutes to under 2 minutes. Queries per user per month increased 3–4x. Support tickets for “how do I write this query” essentially vanished.

MCP Server for Analytics

We released an MCP (Model Context Protocol) server that lets Claude, ChatGPT, and other AI assistants query your Superset instance directly. This sounds incremental, but the implications are significant.

Instead of copying data out of dashboards into Slack or email, analysts can ask Claude questions about their analytics inside their workflow. “Summarize last week’s revenue trends” or “Flag any metrics that moved more than 10% month-over-month” execute directly against your live data. The AI assistant sees real-time results and responds with context.

For data teams, this is a force multiplier. Analysts spend less time running ad-hoc queries and more time interpreting results. For engineering teams embedding analytics, it means end-users can get answers without leaving their tools.

We’ve documented the MCP server integration in our platform guides, and adoption has been faster than expected. Several customers have built internal AI agents that use this server to automate routine reporting and anomaly detection.

API-First Analytics Enhancements

We expanded our REST API to support programmatic dashboard creation, query scheduling, and result streaming. This was driven by customers who wanted to embed charts and tables directly into their products without building custom visualization layers.

One concrete example: a SaaS platform used D23’s API to embed a customer success dashboard into their product. Their customers see real-time KPIs without leaving the product interface. The platform team doesn’t maintain custom dashboarding code—they configure it once in D23 and serve it via API. Maintenance burden dropped 60%.

We also added webhooks for dashboard state changes and query completions, enabling event-driven workflows. A data team can trigger alerts, sync results to data warehouses, or kick off downstream processes based on dashboard updates.

Performance Optimizations and Query Caching

Q1 included substantial work on query latency. We implemented intelligent caching strategies that understand query patterns and pre-warm results for common dashboards. For teams running dashboards across thousands of users, this matters enormously.

Benchmarks from production deployments: median query latency improved 35–45% without touching underlying database infrastructure. For heavy dashboards with 8–12 charts, load time dropped from 12 seconds to 4–5 seconds. That’s the difference between a tool people use and one they avoid.

We also optimized the Superset codebase itself—reducing container size, improving memory efficiency, and streamlining the rendering pipeline. Deployments are faster, scaling is more predictable, and operational costs are lower.

What Customers Are Actually Asking For

Support conversations and customer interviews reveal consistent patterns. These aren’t edge cases—they’re themes across dozens of conversations.

Embedded Analytics and White-Labeling

A significant portion of our customer base wants to embed D23 dashboards into their own products and brand them as their own. They don’t want their end-users to see “Powered by D23”—they want seamless integration.

We’ve shipped white-label support, but the demand goes deeper. Customers want:

• Custom CSS and theming that matches their product design
• SSO integration with their identity provider so users authenticate once
• Row-level security that respects their application’s permission model
• Embedded drill-down and interactivity without page reloads

This is the core use case for embedded BI. Teams building product analytics, customer success dashboards, or internal tools need analytics that feels native to their application. We’re investing heavily here in Q2.

Cost Predictability and Scaling

Every conversation with a CTO or head of data includes cost. They’ve been burned by Looker or Tableau licensing that scaled with user count, query volume, or data size. They want to understand what D23 costs at 100 users, 1,000 users, 10,000 users.

We’ve moved to transparent, usage-based pricing that scales with query volume and data size, not seat count. A team can give analytics access to 5,000 users and pay for the compute they actually consume. This is fundamentally different from traditional BI licensing.

Customers are asking for deeper visibility into cost drivers—which dashboards are expensive, which users are heavy consumers, how to optimize queries to reduce costs. We’re building cost attribution and optimization tools.

Advanced Data Governance and Lineage

As teams scale analytics, governance becomes critical. Who can see what data? How do we enforce data quality? What’s the lineage of a metric—which tables, transformations, and assumptions feed into it?

We’re integrating with data catalogs and lineage tools. Customers using dbt can see the full transformation path from raw data to dashboard. Access controls can be tied to data classifications. Audit logs track who queried what and when.

This appeals strongly to regulated industries and enterprises managing sensitive data. It also appeals to data teams that want to prevent accidental exposure of PII or confidential business metrics.

AI-Assisted Dashboarding and Recommendations

Beyond text-to-SQL, customers want AI to help them build better dashboards. “Given this dataset, what visualizations would be most useful?” or “These two metrics are highly correlated—should we add a chart showing the relationship?”

We’re experimenting with generative dashboard design—the system suggests layouts, chart types, and metrics based on data characteristics and usage patterns. Early feedback is positive, though we’re being careful not to generate noise. The goal is to accelerate dashboard creation for non-technical users, not to replace human judgment.

Multi-Tenant and SaaS-Ready Deployments

Customers building SaaS products want D23 to handle multi-tenancy natively. Isolated data, isolated dashboards, isolated configurations—all within a single D23 instance. This reduces operational overhead and deployment complexity.

We’ve shipped multi-tenant support, but the demand for advanced features—cross-tenant analytics, shared datasets with tenant-specific filters, unified billing—is driving Q2 and Q3 roadmap priorities.

The Competitive Landscape and Where D23 Fits

It’s worth being direct about where we sit relative to alternatives.

Looker and Tableau remain the enterprise standard. They have massive feature sets, deep integrations, and decades of market presence. They’re also expensive, complex to deploy, and require significant expertise to maintain. If you have unlimited budget and deep BI expertise, they work. If you’re a scale-up or mid-market team, the cost and complexity often outweigh the benefits.

Power BI is the cloud-native alternative, tightly integrated with Microsoft’s ecosystem. If your organization is already deep in Azure and Microsoft products, it’s the natural choice. If you’re not, the licensing model and architectural constraints become liabilities.

Metabase is lightweight and approachable, but it’s designed for exploration, not for embedded analytics or API-first BI. It’s a good fit for small teams with simple needs.

Mode and Hex are excellent for collaborative analytics and notebook-style exploration. They’re not designed for embedded BI or for teams that need to serve analytics to thousands of end-users.

Preset, the commercial offering built on Apache Superset, is the most direct competitor. Preset is excellent if you want a managed Superset instance and are comfortable with their deployment model and pricing. We differ in architecture—we prioritize API-first design, embedded analytics, and AI integration more heavily. We also position ourselves as a consulting partner, not just a platform vendor.

D23 fits best for teams that need:

• Embedded analytics in their product
• API-first BI for programmatic access
• Transparent, usage-based pricing that doesn’t penalize user growth
• AI-powered query generation and exploration
• The flexibility of open-source Superset without the operational burden

If that’s your situation, we’re the right choice. If you need the breadth of Tableau or the Microsoft integration of Power BI, we’re not.

Technical Depth: How We’re Building This

For engineering leaders evaluating D23, here’s the technical reality.

Architecture Decisions

We run managed Superset on Kubernetes, with auto-scaling based on query load. The metadata store is PostgreSQL. The query cache is Redis. We support multiple database backends—Postgres, Snowflake, BigQuery, Redshift, and others.

This is intentionally boring. We’re not trying to reinvent the wheel. We’re using proven, battle-tested components and focusing on integration and optimization.

The interesting work happens in the application layer. We’ve extended Superset with:

• Custom authentication providers (SAML, OAuth, custom OIDC)
• Row-level security enforcement through database query rewriting
• Cost tracking and attribution
• Advanced caching strategies that understand query patterns
• LLM integration for text-to-SQL and query generation
• MCP server for AI assistant integration
• Programmatic dashboard and dataset creation via REST API

We’ve also hardened Superset for production. Security patches are applied immediately. Vulnerabilities are tracked and remediated. We maintain our own fork of Superset with production-grade enhancements, and we contribute improvements back to the Apache project.

Data Consulting as a Core Service

We’re not just a platform vendor. We have a data consulting practice that helps teams design schemas, optimize queries, and build effective analytics strategies.

This is intentional. Many teams adopting Superset don’t have deep BI expertise. They need help designing dashboards, defining metrics, and structuring data for analytics. We provide that.

Consulting revenue is meaningful, but more importantly, it keeps us grounded in real customer problems. We see what works and what doesn’t. We see common mistakes and anti-patterns. That feedback directly shapes product priorities.

Roadmap and What’s Next

Q2 2026: Embedded Analytics and White-Labeling

We’re shipping comprehensive white-label support—custom themes, branded login pages, embedded drill-down without page reloads, and seamless SSO. The goal is for end-users to never see D23 branding unless they explicitly choose to show it.

We’re also expanding row-level security (RLS) to be more granular and easier to configure. Teams should be able to define RLS rules in their data warehouse and have D23 enforce them automatically.

Q2 2026: Cost Attribution and Optimization

We’re building dashboards that show which dashboards, users, and queries are expensive. We’re adding query optimization suggestions—“This query could be 5x faster if you added an index” or “This dashboard is fetching data it doesn’t display.”

The goal is to help teams optimize spend without sacrificing functionality. Many teams will discover that a few dashboards drive 80% of costs, and small optimizations yield big savings.

Q3 2026: Advanced Governance and Lineage

We’re integrating with data lineage tools and metadata catalogs. The vision is that a user can click on a metric in a dashboard and see the entire transformation path—which tables, which dbt models, which transformations, which assumptions.

We’re also building governance workflows—approval processes for new dashboards, data quality checks before data is served, audit trails for sensitive queries.

Q3 2026: Generative Dashboard Design

We’re shipping AI-assisted dashboard creation. The system analyzes a dataset and suggests useful visualizations, layouts, and metrics. Users can accept suggestions, modify them, or start from scratch. The goal is to accelerate dashboard creation for non-technical users.

Q4 2026 and Beyond: Advanced Analytics and Forecasting

We’re exploring integration with forecasting and anomaly detection libraries. The vision is that D23 doesn’t just show you what happened—it helps you predict what will happen and alerts you when things deviate from expectations.

This is early-stage, but the demand is clear. Teams want AI-powered analytics, not just dashboards.

Real Customer Outcomes

Here’s what we’re seeing in production.

A SaaS platform embedded D23 into their product to give customers visibility into their data. Result: 40% increase in customer retention, because customers could see ROI. The platform team spent 2 weeks configuring D23 and zero weeks maintaining custom dashboarding code.

A private equity firm uses D23 to track KPIs across portfolio companies. They standardized on a set of metrics, embedded D23 dashboards into their investment management system, and reduced reporting time from 3 weeks to 3 days. They also discovered that three portfolio companies had similar operational inefficiencies, leading to $2M in cost savings.

A scale-up data team migrated from Looker to D23 and reduced annual BI spend by 70% while increasing dashboard count by 3x. They also gained the ability to embed analytics into their product, which they couldn’t do with Looker without significant engineering work.

A venture capital firm uses D23’s AI features to generate monthly performance summaries for their LPs. Instead of spending a week compiling reports, they run a prompt and get a polished narrative with charts and key metrics.

These aren’t theoretical. These are customers we work with every day.

The Broader Context: Why This Matters Now

We’re shipping D23 at an inflection point in the BI market.

Traditional BI vendors (Looker, Tableau, Power BI) are excellent at exploration and reporting, but they’re not built for embedded analytics or API-first use cases. They’re also expensive and complex.

Open-source BI tools like Apache Superset solve the core problem—visualization and exploration—but they require operational expertise to run in production.

The market is bifurcating. On one side, enterprises with deep BI expertise and unlimited budgets will continue using Looker and Tableau. On the other side, teams that need modern, flexible, cost-effective analytics are moving to open-source tools. The question is: who helps them run those tools in production?

That’s where D23 fits. We’re the bridge between the flexibility of open-source and the reliability of managed services.

We’re also shipping at a moment when AI is fundamentally changing how people interact with data. Text-to-SQL, generative dashboarding, and AI-assisted analytics are no longer nice-to-have—they’re becoming table stakes. Teams expect to ask questions in natural language and get answers. They expect AI to help them explore data faster and understand patterns more deeply.

Traditional BI vendors are adding AI features, but they’re bolted on. We’re building AI into the core product because it’s what customers actually need.

Looking Ahead: What Success Looks Like

For D23, success means:

• Teams that would have chosen Looker or Tableau now choose D23 because it’s cheaper, more flexible, and easier to embed
• Teams building SaaS products can add analytics to their product in weeks, not months
• Data teams spend less time managing infrastructure and more time solving business problems
• AI-powered analytics becomes the default way people explore data, not a special feature
• Open-source BI becomes the standard for modern analytics, not a niche choice

We’re early in this journey. Q1 2026 is a checkpoint, not the destination. But the direction is clear, the customer demand is real, and the team is executing.

How to Get Started

If you’re evaluating D23 or managed Apache Superset, here’s what to consider.

First, assess your use case. Are you building embedded analytics? Do you need API-first BI? Are you cost-sensitive? Do you need AI-powered query generation? If you answered yes to any of these, D23 is worth a conversation.

Second, understand the operational trade-off. Managed Superset means you don’t run Kubernetes clusters or manage database connections. You trade some operational overhead for some flexibility in customization. For most teams, that trade is worth it.

Third, consider the consulting component. If your team is new to analytics or BI, having access to experienced data engineers who can help design schemas, optimize queries, and build effective dashboards is valuable. We include consulting as part of our service.

Fourth, check the integrations. Does D23 work with your data warehouse? Your identity provider? Your data catalog? We support most common tools, but it’s worth verifying.

If you’re ready to explore, visit D23 to learn more about our platform, see documentation, and schedule a conversation with our team.

Conclusion: The State of Managed Analytics

Q1 2026 is a good moment to reflect on where managed Apache Superset has landed.

It’s no longer a niche choice. Teams at scale-ups, mid-market companies, and even some enterprises are choosing open-source BI over traditional vendors. The reasons are clear: cost, flexibility, and the ability to embed analytics into products.

D23 exists to make that choice easier. We’ve built a platform that handles the operational complexity, ships modern features like text-to-SQL and MCP integration, and provides the consulting support teams need to succeed.

Q1 was about shipping core features and validating product-market fit. Q2–Q4 will be about scaling that fit—supporting more use cases, more integrations, and more customers.

If you’re evaluating analytics platforms, don’t assume the traditional vendors are your only choice. Open-source BI, managed well, is often the better answer. And if you’re looking for a team that can help you run it, we’re here.

For more information on Apache Superset capabilities and best practices, you can explore Apache Superset’s official documentation, review G2 user reviews of Apache Superset, or check out TrustRadius Apache Superset reviews to see how other teams are deploying it. You can also read The New Stack’s coverage of Apache Superset for deeper context on the platform’s role in modern data infrastructure, or review InfoWorld’s Apache Superset review for a detailed technical assessment. Additionally, PeerSpot Apache Superset reviews provide comparisons with Tableau and Power BI, and the Preset blog on Apache Superset covers managed deployments and best practices.