Embedded Analytics: Cube vs Apache Superset vs Looker

Embedding analytics in customer-facing applications is substantial differentiated value for substantial SaaS products. Customers see substantial data about their substantial usage; substantial product becomes substantial stickier; substantial expansion revenue increases. The substantial tools that ship substantial embedded analytics in 2026 are substantially different from substantial internal-BI tools. This post walks through what’s actually deployed.

Why embedded analytics matters substantially#

Several substantial value drivers:

Substantial customer stickiness. Customers who see substantial value in substantial data are substantially less likely to substantially churn.

Substantial expansion revenue. Substantial analytics often justifies substantial higher-tier pricing.

Substantial competitive differentiation. Substantial analytics depth differentiates substantial competitive offerings.

Substantial product feedback. Substantial customers who see substantial data substantially give better feedback.

Substantial reduced support burden. Substantial self-service analytics substantially reduces substantial support questions.

The substantial business case for substantial embedded analytics is substantial; substantial execution determines substantial actual gain.

What’s different about embedded vs internal#

Substantial differences from substantial internal BI:

Substantial multi-tenancy. Customer A must see only substantial Customer A data; substantial isolation is substantial critical.

Substantial scale. Substantial customer-facing analytics serves substantial more queries than substantial internal.

Substantial latency. Substantial customers expect substantial sub-second response; substantial internal BI tolerates substantially slower.

Substantial branding and substantial UX integration. Substantial dashboards must substantially fit substantial product design; substantial off-the-shelf BI iframe substantially obvious.

Substantial pricing model. Substantial per-user pricing of substantial internal BI substantially doesn’t work for substantial embedded.

Substantial substantial support burden. Substantial customers expect substantial dashboards to substantially work; substantial broken dashboard is substantial customer support ticket.

Cube#

Cube (formerly Cube.js) is substantial open-source embedded analytics infrastructure.

Strengths:

• Substantial semantic layer — substantial central definitions of substantial metrics.
• Substantial caching and substantial pre-aggregations — substantial performance.
• Substantial multi-tenant native.
• Substantial flexibility — substantial customers build substantial own UI on top.
• Substantial open-source plus substantial Cube Cloud for managed.

Trade-offs:

• Substantial work to build substantial UI. Cube provides substantial query layer; substantial UI is substantial separate effort.
• Substantial learning curve for substantial Cube concepts.
• Substantial smaller community than substantial dominant tools.

Best for: substantial product engineering teams building substantial custom embedded analytics with substantial control.

Apache Superset#

Apache Superset is substantial open-source BI with substantial embedded capabilities.

Strengths:

• Substantial open-source mature. Substantial production deployment at substantial Airbnb, Lyft, plus the substantial various.
• Substantial chart library rich.
• Substantial dashboard editor for substantial business users.
• Substantial Preset managed service.
• Substantial embedded SDK for substantial integration.

Trade-offs:

• Substantial primary use is substantial internal BI — substantial embedded support is substantial real but not substantial primary.
• Substantial multi-tenancy workable but substantial work to get substantial right.
• Substantial UX customization limited compared to substantial Cube approach.

Best for: substantial teams wanting substantial pre-built UI with substantial embedded capability.

Looker#

Looker (Google Cloud) is substantial commercial BI with substantial embedded capability.

Strengths:

• Substantial mature embedded capability — Looker has substantial embedded heritage.
• Substantial LookML semantic layer — substantial central definitions.
• Substantial Google Cloud integration with substantial BigQuery and substantial broader stack.
• Substantial enterprise capability — substantial governance, substantial scale.

Trade-offs:

• Substantial commercial pricing — substantial expensive for substantial embedded use cases.
• Substantial Google-anchored.
• Substantial complex pricing for substantial embedded.

Best for: substantial enterprises on substantial Google Cloud where substantial Looker investment is substantial existing.

The substantial decision framework#

For most substantial teams in 2026:

Pick Cube when you want substantial custom UI with substantial semantic layer and substantial high control. Substantial common modern choice for substantial product engineering.

Pick Superset when you want substantial pre-built UI with substantial reasonable embedded capability. Substantial works well for substantial moderate embedded needs.

Pick Looker when substantial existing Google Cloud and substantial Looker investment justify substantial commercial pricing.

Pick others — Sigma, Mode, Metabase, Hex, Lightdash, Evidence, plus substantial various — for substantial specific scenarios. Substantial overlap with above.

Build from scratch when substantial unique requirements substantially justify substantial substantial investment. Substantial frequently a substantial mistake.

The substantial production patterns#

Several substantial patterns from substantial production deployments:

Substantial pre-aggregation discipline. Substantial customers expect substantial fast queries; substantial pre-aggregation matters substantially.

Substantial multi-tenancy via substantial row-level security. Substantial database-level isolation plus substantial query-level enforcement.

Substantial caching strategy. Substantial customer dashboards substantially benefit from substantial caching.

Substantial monitoring and observability. Substantial broken dashboards must be substantial detected fast.

Substantial customer-facing usage analytics. Substantial customers want substantial visibility into substantial their data; substantial product team wants substantial visibility into substantial dashboard usage.

Substantial drill-down depth. Substantial customers expect substantial ability to substantial dig in.

The substantial semantic layer dimension#

A specific 2026 development: substantial semantic layer importance.

Substantial central metric definitions matter substantially in substantial embedded analytics:

• Substantial single source of substantial metric definitions across substantial customer dashboards, substantial internal reporting, substantial sales decks
• Substantial change-once-update-everywhere
• Substantial governance over substantial metric meaning

All three substantial tools have substantial semantic layer capability; substantial Cube and substantial Looker substantially most-developed.

What we typically see at clients#

Common patterns:

Substantial iframe-of-internal-BI. Substantial common but substantial frequently substantial poor UX.

Substantial cube-based custom analytics at substantial product engineering teams.

Substantial Superset embedded at substantial mid-tier deployments.

Substantial proprietary build at substantial larger SaaS — substantial common but substantial high effort.

Where pdpspectra fits#

Our data engineering practice builds substantial production embedded analytics platforms with substantial substantial multi-tenancy, substantial scale, and substantial UX integration.

Related reading: the data catalog post, the data stack operational engine post, and the lakehouse Iceberg vs Delta vs Hudi post.

---

Embedded analytics is substantial differentiated value. Talk to our team about your analytics platform.