Skip to content

For embedded analytics in a SaaS product, Sigma and Omni both isolate tenants well, but they put the lock in different places. Sigma passes user attributes and teams in a signed JWT and enforces rows with workbook formulas, per-tenant connections, or separate Sigma organizations. Omni applies access filters from its semantic model to every query, with schema-per-tenant and database-per-tenant options. As of November 2026 neither vendor publishes a price, so “pricing at scale” is a negotiation either way; Basedash, ThoughtSpot Embedded, and Metabase are the alternatives worth a look if you want a different shape.

Everything below was checked against Sigma’s and Omni’s official documentation on November 16, 2026. For the general (non-embedded) comparison of the two tools, see our Omni vs Sigma comparison.

Sigma vs Omni for embedded analytics at a glance

Attribute Sigma Omni Basedash
Embed method JWT-signed one-time URL in an iframe Signed URL (standard or 2-step SSO) in an iframe JWT SSO in an iframe, or the @basedash/embed React SDK
Front-end SDK React SDK and JavaScript SDK Server-side signing SDK only, no React component React components for chat, dashboards, insights, automations, full app
Row-level tenant isolation User attributes and teams in the JWT, enforced by RLS formulas or warehouse roles Access filters on model topics, applied to the WHERE clause of every query Separate Basedash org per customer; Postgres RLS via groups; signed filter tokens on shared dashboards
Per-tenant schema or database Swap connection per user with eval_connection_id Dynamic schemas and dynamic environments Separate org and data source per customer
Separate org per customer Yes, Sigma Tenants (GA July 2026) Not documented; entities and entity folders inside one instance Yes, created through the API
Customers build their own content Yes, edits saved as the embed user’s draft Yes, Create Mode with entity folders Yes, full app embedding includes chart and dashboard creation
AI in the embed Embedded Sigma agents in a chat element Embedded Omni Agent, or your own agent via Omni’s MCP server Embedded AI chat (BasedashChat)
Query model Live queries on the warehouse Live queries, results cached 6 hours by default Live queries
SSO / SCIM SAML or OIDC; SCIM (requires SAML) SAML or OIDC; SCIM for Entra, Okta, Rippling SAML 2.0 or OIDC; SCIM (Enterprise)
Deployment Cloud only (AWS, Azure, or GCP) Cloud only (AWS or Azure) Cloud, or self-hosted on Enterprise
Published price No, quote only; secure embedding is a premium feature No, quote only; AI overage $1.00 per credit No for embedding (Enterprise, custom); Startup $1,000/month excludes embedding
Free trial 7 days Via a demo request form; length not published 14 days, no card

Pricing verified November 16, 2026.

How we compared Sigma and Omni

The criteria come from the questions SaaS teams actually ask when they shortlist these two tools:

  1. Multi-tenant isolation. Where is the tenant ID enforced, and can a customer tamper with it?
  2. Developer effort. What does your backend sign, and what does your front end render?
  3. Self-service for customers. Can customers build their own reports without seeing each other’s content?
  4. AI inside the embed. Can customers ask questions in plain language, and who pays for it?
  5. Pricing at scale. What is published, and what will you be negotiating?
  6. Security and deployment. SSO, SCIM, compliance, and hosting options.

Which handles multi-tenant row-level security better, Sigma or Omni?

Both are sound. The difference is whether isolation lives in the BI model or in the embed token plus workbook logic.

Sigma. Your server signs a JWT whose claims can include user_attributes, teams, account_type, and tenant (claims reference). Row-level security is then a formula such as CurrentUserAttributeText("Region") = [Store Region] on a data model or table (row-level security). For stronger separation, Sigma offers three escalations:

  • eval_connection_id swaps the workbook’s connection for a per-tenant connection at query time. It can’t be combined with writeback or OAuth connections.
  • Connection roles can be assigned from user attributes, so the warehouse enforces access (dynamic roles).
  • Sigma Tenants manages users and resources “across multiple Sigma organizations”, so each customer can get its own org (multitenancy).

Column-level security in embeds requires a data source with column-level security configured (restrict access in embeds).

Omni. The tenant ID goes into userAttributes in the signed URL, and Omni’s access filters “apply the values assigned on a user attribute to the WHERE clause of every SQL query a user runs” (embed best practices). Because the filter sits on the model topic, every dashboard, workbook, and AI answer inherits it. Omni documents four tenant patterns:

  • Row level, with default topic access filters.
  • Schema level, with dynamic schemas that point each customer at an identical schema.
  • Database level, with dynamic environments.
  • Model level, with shared extension models.

Field-level control comes from access grants, which can mask values instead of hiding them. Each embed sign-in also deletes user attributes not present in the new token, so stale values don’t linger. Omni’s docs warn plainly that “segmenting data using hidden dashboard filters is not a secure practice.”

Verdict. If your customers share tables keyed by tenant_id, Omni’s model-level filters are harder to forget, because nobody can build a workbook that skips them. If your customers live in separate databases or need fully separate admin spaces, Sigma’s per-user connection swap and separate tenant organizations are more direct. Either way, generate the token server-side and never pass the tenant ID as a visible filter. Our guide to multi-tenant analytics architecture covers the patterns in more depth.

How much developer work does each embed take?

Sigma gives front-end teams more to work with. It publishes @sigmacomputing/react-embed-sdk and a plain JavaScript SDK, plus a Node package for encrypting OAuth tokens (React SDK docs). The host page and the iframe talk through postMessage events such as workbook:variables:update (embed events). Embed users are created automatically from the email in the token. One gate to know about early: secure embedding “is a premium feature” that your account executive has to enable (create a secure embed).

Omni keeps the integration on the server. Its @omni-co/embed package (version 1.2.2 on npm) signs URLs for dashboards, workbooks, the full app, and content discovery; there is no React component, so you render the iframe yourself and listen for postMessage events like dashboard:filter-changed (embed events). Embedding has to be switched on by Omni support. If you built on Omni’s older signing format, note that v0 signing was “unsupported after October 1, 2026, and removed by January 1, 2027” (standard SSO), so upgrade to the v1 payload and SDK 1.0.0 or later.

For a React app, Sigma saves some wrapper code. For a team that prefers to own the iframe, the two are close.

Can customers build their own dashboards inside the embed?

Both allow it, with different guardrails.

  • Sigma lets you allow editing in embedded content. Embed users’ edits stay in their own draft, Live Edit isn’t supported, and URL parameters such as :hide_save_as trim the interface (embed behavior). Themes and multiple custom fonts handle branding.
  • Omni has Create Mode, which lets embed users “create, share, and navigate content as if they were in the native Omni app.” Each customer entity gets an automatically created shared folder and group, so one customer’s reports stay out of another’s view. Themes can replace Omni’s logo and wordmark, and vanity domains are available for embed use cases.

Omni’s entity folders give a cleaner per-customer workspace out of the box. Sigma’s per-tenant organizations do the same at a heavier, org-level granularity.

Which has better AI features for embedded customers?

Both now put an agent inside the embed, and both charge for it in credits.

Sigma lets you embed a Sigma agent in a chat element using the same JWT signing as other embeds. Sigma Assistant can still appear inside an embedded workbook, but the standalone Assistant embed is deprecated and reaches end of support on March 16, 2027 (embed Assistant). Agent messages started consuming credits on November 1, 2026, at 0.1 credit per message; Sigma doesn’t publish the price of a credit (billable usage).

Omni offers two paths (AI chat in embeds): embed the Omni Agent, which comes with the Embed feature, or connect an agent you build to Omni’s MCP server using a token from createEmbedMcpToken, which Omni support must enable. Queries run as the embed user, so access filters still apply. The entityAiCreditLimit parameter caps AI credits per customer at login, and overage is billed at $1.00 per credit (AI usage).

For a team without a full-time data engineer, the setup difference matters more than the chat UI. Omni’s agent answers from the semantic model, so someone has to build and maintain that model first. Sigma’s embedded agent lives in a workbook chat element, and analysts build those workbooks in a spreadsheet-style interface without a separate modeling language.

What does Sigma and Omni pricing look like at scale?

Neither vendor publishes embedded pricing as of November 2026.

  • Sigma: sigmacomputing.com/pricing redirects to a contact form. The embedded product page says pricing “depends on how many people use your embeds and what they need” (Sigma embedded analytics). Licenses come in View, Act, Analyze, and Build tiers (license overview), and agent messages add credit usage.
  • Omni: there is no pricing page (omni.co/pricing returns a 404). The only published number is the $1.00 AI credit overage.

At scale, the cost drivers to pin down in a quote are the same for both: how embed viewers are counted (named users, active users, or capacity), which license tier a customer who edits content needs, how AI credits are priced and capped per tenant, and whether a staging environment costs extra. Our embedded analytics pricing comparison lists 21 platforms, including the ones that do publish prices.

Security, compliance, and data sources

  • Sigma supports SAML or OIDC and SCIM, which requires SAML and isn’t available on trials (SCIM). It runs on AWS, Azure, or GCP, and the region can’t be changed after creation (regions). Sources include Snowflake, BigQuery, Databricks, Redshift, PostgreSQL, MySQL 8+, SQL Server 2022, and Starburst, with ClickHouse and MotherDuck in beta (data sources).
  • Omni supports SAML and OIDC, with SCIM for Entra, Okta, and Rippling (authentication). It states ISO/IEC 27001:2022 certification, annual SOC 2 Type II audits, and HIPAA audits (security). It is hosted on AWS and Azure, with regions in the US, Canada, the EU, the UK, Australia, India, the UAE, and Singapore (cloud regions). Sources include Snowflake, BigQuery, Databricks, Redshift, Postgres, MySQL, SQL Server, ClickHouse, and MotherDuck (connect data).

Neither documents a self-hosted option.

Both vendors also publish comparison pages that argue against each other. Sigma’s page calls Omni a read-only tool that depends on its modeling layer. Omni’s page says Sigma’s AI chat “lacks business context.” Read both as marketing, and test the specific embed flow you need during the trial.

What should Explo customers do?

Omni acquired Explo in October 2025 and said it would migrate the majority of Explo customers to Omni over the following 12 months “before sunsetting the Explo platform” (Omni on Explo). Omni hasn’t posted an exact shutdown date. For an Explo customer, Omni is the default path and the migration is assisted. It also means adopting Omni’s semantic model. If your team never had a data modeler, evaluate Sigma or a tool without a modeling step at the same time, rather than treating the move to Omni as automatic.

Where Basedash fits

Basedash is an AI-native BI tool that embeds AI chat and dashboards into a product. It’s a different shape from both: there’s no semantic model to build first, and the customer-facing experience is centered on asking questions in plain language.

  • Pricing: Embedding is on the Enterprise plan with custom pricing. The Startup plan ($1,000/month plus AI usage, 25 users) does not include embedding (pricing, as of November 2026). 14-day trial, no card.
  • Embed method: The @basedash/embed React SDK ships components for chat, dashboards, insights, automations, the full app, and shared dashboards, plus server helpers for signing tokens. Non-React apps use JWT SSO in an iframe (embedding docs).
  • Tenant isolation: Each customer gets its own Basedash organization, created through the API. A groups claim assigns group memberships on every login. Shared dashboards can lock filters such as tenant_id with a signed token.
  • Row-level security: Database row-level policies through the basedash.groups session variable work on Postgres and Supabase only (row-level security).
  • Security and deployment: SSO with SAML 2.0 or OIDC, SCIM, and self-hosting are on Enterprise.
  • Best for: SaaS teams that want customers to ask questions in plain language, and that are fine with one organization per customer.
  • Not ideal for: Teams that need a governed semantic model with model-level filters on shared tables (Omni), or spreadsheet-style self-service at warehouse scale (Sigma).

If you want published embedded pricing, ThoughtSpot Embedded, Metabase, and Luzmo are covered in our GoodData alternatives comparison.

Which should you choose?

  • Customers share tables keyed by a tenant ID: Omni. Model-level access filters apply to every query, workbook, and AI answer.
  • Each customer has its own database or needs its own admin space: Sigma. Use eval_connection_id or Sigma Tenants.
  • You are embedding in a React app and want components, not raw iframes: Sigma or Basedash. Omni has no React component.
  • No full-time data engineer: Sigma, because analysts can build in workbooks without a model first. Basedash also fits if customers mostly want to ask questions.
  • You want to bring your own AI agent: Omni, through its embed MCP token.
  • You are on Explo today: Start with Omni’s migration offer, and compare Sigma in parallel if a semantic model is a new commitment for your team.
  • On Snowflake: Both query it live; Sigma can also pass OAuth tokens so Snowflake roles apply per user.

Frequently asked questions

Is Sigma or Omni better for multi-tenant row-level security in a SaaS product?

Both enforce tenant isolation server-side. Omni puts access filters on model topics, so a tenant ID from the signed URL is added to every query’s WHERE clause, and it also supports schema-per-tenant and database-per-tenant setups. Sigma passes the tenant through JWT user attributes and teams, enforces rows with RLS formulas or warehouse roles, and can swap connections per user or give each customer its own Sigma organization. Omni is safer when customers share tables; Sigma is more direct when customers have separate databases or need separate admin spaces.

How much does Sigma embedded analytics cost?

Sigma doesn’t publish prices. As of November 2026, its pricing page redirects to a contact form, and its embedded page says pricing depends on how many people use your embeds and what they need. Secure, JWT-signed embedding is a premium feature that an account executive enables. Embedded agent messages consume 0.1 credit each from November 1, 2026, and the price per credit isn’t published. Budget for a negotiated contract and ask how embed viewers are counted.

Does Omni publish pricing for embedded analytics?

No. As of November 2026, omni.co has no pricing page, and embedding is enabled by Omni support as part of a contract. The only public number is AI credit overage at $1.00 per credit beyond the included allotment. Omni’s entityAiCreditLimit embed parameter lets you cap AI credits for each customer, which helps keep AI spend predictable when many tenants use the embedded agent.

Does Omni have a React SDK for embedding?

Not a component library. Omni’s @omni-co/embed npm package is a server-side SDK that signs embed URLs and creates MCP tokens for bring-your-own-agent setups. You render the iframe yourself and use postMessage events to sync filters and page changes. Sigma publishes a React embed SDK and a JavaScript SDK, and Basedash publishes React components for chat, dashboards, and the full app.

Which is easier to set up for a team without a full-time data engineer, Sigma or Omni?

Sigma is usually faster to start because analysts can build workbooks on live warehouse tables in a spreadsheet-style interface without a model. Omni works best once someone has defined topics, joins, and access filters in its semantic model, which is what makes its AI answers and tenant filters consistent. If nobody on the team will own that model, factor that work into the Omni timeline.

What are the alternatives to Sigma for embedded analytics?

Omni is the closest warehouse-native alternative. ThoughtSpot Embedded has a free developer tier and search-first AI. Metabase offers open-source guest embeds and paid tenant isolation. Luzmo publishes a price and has a drag-and-drop editor for customers. Basedash embeds AI chat and dashboards through React components, with one organization per customer. Choose based on where tenant isolation needs to live and whether you need a published price.

What happens to Explo now that Omni owns it?

Omni acquired Explo in October 2025 and said it would migrate most Explo customers to Omni over the next 12 months before sunsetting the Explo platform. No exact shutdown date is published. Explo customers get an assisted path to Omni’s embedded product, which uses Omni’s semantic model and signed-URL embedding, so plan for modeling work as part of the move.

Written by

Rachel van der Lugt avatar

Rachel van der Lugt

Founding Enterprise GTM at Basedash

Rachel van der Lugt is founding enterprise GTM at Basedash, where she leads enterprise go-to-market for an AI-native business intelligence platform. Her work focuses on helping larger teams evaluate, adopt, and roll out Basedash for governed analytics across the organization.

View full author profile →

Basedash lets you build charts, dashboards, and reports in seconds using all your data.