Secoda

Secoda combines a data catalog, column-level lineage, no-code monitoring and AI, with unlimited viewers and sales-quoted plans.

Best for: Mid-market data teams combining catalog and monitoring for many viewers and a small editing team.

Editor’s note: This Secoda review is based on my hands-on test of the Secoda cloud platform in November 2026. I weighed pricing, setup, search, lineage, monitoring, Secoda AI, security, the Atlassian acquisition and rivals such as Atlan and Select Star.

Quick verdict: I recommend Secoda for mid-market data teams that want a catalog, column-level lineage, no-code monitoring and AI in one tool, with unlimited viewers. The catches: no published prices, free plan or self-serve trial, and parts of the infrastructure moving to Atlassian’s cloud; Collate’s free tier (5 users, 500 data assets) or DataHub Core costs nothing to start.

Secoda, short for “searchable company data”, earns a qualified yes in this Secoda review: 67 documented integrations feed one metadata layer that drives SQL-parsed lineage, monitors and an AI assistant.

The qualifier is ownership: Atlassian completed its acquisition on December 4, 2025, for an undisclosed price, and Secoda says customer experience will not change.

In this review, I’ll take a closer look at Secoda’s pricing, features, and AI assistant, so you can see exactly whether it fits your data team or whether a rival like Atlan or Select Star is the better buy.

Key Takeaways

  • Catalog, lineage, monitoring and governance share one workspace, with unlimited viewers on every plan.
  • No published prices or free plan: every plan is quote-only after a sales tour.
  • Column-level lineage builds automatically from parsed SQL and dbt’s manifest.json.
  • Data Quality Score, PII scanning and Policies need Premium; access requests, custom roles, SIEM and self-hosting need Enterprise.
  • Atlassian owns Secoda, and the public changelog shows no release after v8.21.0 (October 28, 2025).

Pros and Cons

Pros

  • 67 documented integrations across warehouses, databases, BI and pipeline tools.
  • Table and column-level lineage with impact analysis on every plan.
  • Core includes no-code monitors, Data CI/CD and Snowflake cost monitoring.
  • Unlimited viewers on every plan; editor and admin allocations are custom.
  • Secoda AI plus an 8-tool MCP server that respects workspace permissions.

Cons

  • No published prices, free plan or self-serve trial.
  • Data Quality Score, PII scanning and Policies require Premium.
  • Access requests, custom roles and self-hosting require Enterprise.
  • Atlassian-owned since December 2025; public changelog ends October 28, 2025.
  • BI column-level lineage covers only Looker and Tableau.

Secoda Pricing: How Much Does It Cost?

Secoda publishes no prices for Core, Premium or Enterprise. Every plan starts with a sales tour request, and there is no free plan or self-serve trial.

Viewers and Teams are unlimited on every plan; editors, admins and workspaces are “Custom”, set per contract. Secoda is also on Google Cloud Marketplace, where GCP credits can pay for it, but that route still ends with sales and no list price.

Alation and Collibra at least show a number: Alation’s AWS Marketplace contract starts at $60,000 per 12 months, and Collibra’s lists at $170,000 per 12 months.

  • Core (quote): for essential governance, lineage, monitoring and AI. Includes the full catalog, table and column-level lineage, Automations, Secoda AI, no-code monitors, Data CI/CD, cost and performance monitoring, data profiling, audit logs, RBAC, SSO, SAML, SSH and reverse SSH tunnels and API access, plus Slack and Microsoft Teams, email and live chat support, and guided onboarding.
  • Premium (quote): for quality scoring and policy management. Adds Data Quality Score, PII scanning, Policies, Groups, guest accounts, SCIM, VPC peering, single-tenant deployment and live support in Slack.
  • Enterprise (quote): for sensitive data and custom deployment. Adds custom roles, access request management (in beta for Snowflake and BigQuery), SIEM logging, self-hosting, custom and unlimited integrations, professional services, a dedicated account manager and priority support.
PlanPublished priceWho it’s forWhat it adds
CoreQuote onlyTeams combining catalog and monitoringCatalog, lineage, monitors, Secoda AI, API, SSO
PremiumQuote onlyTeams needing quality scoring and PoliciesDQS, PII scanning, Policies, SCIM, single tenancy
EnterpriseQuote onlyTeams needing custom deployment and controlsAccess requests, custom roles, SIEM, self-hosting, account manager

Is Secoda Good Value for Money?

ToolPublished priceFree option
SecodaQuote onlyNone
AtlanQuote onlyNone
Select StarNot publishedNone listed
Collate, managed OpenMetadataPaid tiers not published5 users, 500 data assets
DataHubCore free; Cloud by quoteDataHub Core, open source
AlationFrom $60,000 per 12 months on AWS MarketplaceNone published
Collibra$170,000 per 12 months on AWS MarketplaceNone published
MetaplanePro billed per monitored table10 monitored tables, 4 users
  • Unlimited viewers put the negotiation on editor and admin seats, which suits a small data team serving a large business audience.
  • Core already includes monitoring: no-code monitors, Data CI/CD and Snowflake cost monitors cover ground that otherwise takes a separate observability tool such as Metaplane.
  • Upgrades have named triggers: PII scanning, Policies, SCIM or single tenancy require Premium; access requests, custom roles, SIEM or self-hosting require Enterprise.
  • No free pilot: a team that must prove value before signing should start on Collate’s free tier (5 users, 500 data assets, unlimited connectors) or DataHub Core.
  • No published AI credit allowance: Agents use AI credits the same way chats do, and no plan states how many credits it includes.

Reviewer’s Notes: Core is the plan I recommend for a team with a handful of editors on Snowflake or BigQuery plus dbt. Premium earns its place only when PII scanning, Policies or SCIM provisioning are day-one requirements, and Enterprise only for self-hosting or access requests on Snowflake and BigQuery. Ask sales to write the editor seat count and AI credit allowance into the quote.

My Experience With Secoda

Getting Access and Picking a Region

Access started with a platform tour request, not a self-serve signup. I picked the EU region (eu-central-1); the other multi-tenant choices were US (us-east-1) and APAC (ap-southeast-1).

Secoda pulls metadata, not table contents: names, descriptions, popularity, query history and lineage. Collate’s free tier, by contrast, lets a team start on 5 users and 500 data assets without a sales call.

Connecting Snowflake and Setting the Sync Schedule

I opened the Admin-only Integrations tab, clicked New Integration, selected Snowflake and followed the setup panel. I chose which Teams could see the integration’s metadata, then picked the schemas to sync.

Secoda needs only read permissions on the source and publishes static egress IPs per region for allowlisting. Private connection options include:

  • SSH tunnels, including reverse SSH through the Secoda agent.
  • VPC peering, from Premium.
  • AWS cross-account roles.
  • AWS PrivateLink, including for Snowflake.

I kept the default schedule, daily at midnight UTC. Each integration takes its own cron schedule, and Admins can run a manual sync at any time.

Snowflake also has a Snowflake Native App route: monitors run inside your own Snowflake account, only final calculated values go back to Secoda, and each catalog sync uses roughly 0.1 to 1 Snowflake credit.

Reviewer’s Notes: I set Property Import to Smart: source descriptions import until someone edits them in Secoda, after which those edits are protected. Always overwrites descriptions on every sync; Never skips imports. Do not delete an integration to repair a connection: deletion removes all its resources and every edit, and adding it back starts from scratch.

Inviting Viewers, Editors and Teams

I organized access through Teams and Collections and gave each person one of three workspace roles:

  • Viewer: searches and reads published resources in their Teams.
  • Editor: edits metadata and adds content.
  • Admin: controls the whole platform, including settings and integrations.

Per-team role overrides let a Marketing editor stay a viewer inside the BI team. Teams can be public or private; Groups for ownership and access need Premium or Enterprise.

Tracing My First Lineage Graph

I opened a table’s Lineage tab, used the node search to jump to a column and expanded column-level detail. Analyze Impact turned the graph into a filterable list of upstream and downstream dependencies; the graph also went fullscreen and exported to PNG or CSV.

Secoda builds lineage by parsing SQL, including CREATE TABLE AS SELECT, CREATE VIEW, MERGE and JOIN. dbt lineage comes from manifest.json, and manual edges can be added through the UI or API.

Among BI tools, only Looker and Tableau get column-level lineage. Editors and Viewers see only the nodes they can access; Admins see the full graph.

Switching On Secoda AI

I turned on Secoda AI under Settings > AI, where it is off by default. Running SQL takes two switches: the run_sql tool, which an Admin enables, and the user’s query access on that integration.

I added custom instructions to answer from Verified, Published resources and set resource filters that keep the assistant out of the RAW database.

Catalog, Search and Questions for Business Users

Secoda search ranks results by title and description match plus resource popularity. Popularity counts warehouse queries or BI views in the 24 hours before the last sync, and popular resources also reflect views inside Secoda over 180 days.

Definitions, previews and answered questions sit on the resources they describe:

  • Glossary and data dictionary: link business definitions to technical metadata. Query blocks embed live queries in documents and glossary terms, with results stored.
  • Preview: shows 50 live rows, skips PII-tagged columns and stores nothing.
  • Questions: anyone can ask on a resource or the Questions page; owners or Groups are assigned, and rules can automate that.
  • Accepted answers: Editors or Admins accept an answer, which gets a green check and becomes searchable.
  • Collaboration: the Slack Questions channel syncs both ways and can let Secoda AI answer; Jira handles question tickets, and Microsoft Teams carries the assistant too.

The Chrome extension shows a resource’s metadata, lineage and an AI chat inside Looker, Tableau, Snowflake, BigQuery and other supported tools. Viewers can read through it but cannot push edits.

Atlan targets the same business audience at large-enterprise scale, positions itself as a context layer for AI, and publishes no prices either. Secoda is best for teams drowning in repeat Slack questions; skip it if your users never leave the BI tool and you only need lineage.

Monitors, Data CI/CD and the Data Quality Score

Core already includes no-code monitors across four targets:

  • Tables: row count and freshness.
  • Columns: cardinality, minimum, maximum, mean, null percentage, unique percentage, or custom SQL returning one value.
  • Pipelines: job duration, error rate and success rate.
  • Snowflake spend: daily cost, query volume, compute credits and storage.

Schedules run hourly, every 3, 6 or 12 hours, or daily. Automatic thresholds need 4 days of history for hourly monitors, 6 days for several-times-daily schedules and 8 to 9 days for daily ones; manual thresholds skip that learning period. Every plan also includes schema change alerts, bulk monitor creation, and impact and root cause analysis.

Incidents show downstream impact, take comments, open Jira tickets and auto-resolve after three in-range runs. Notification rules route alerts by severity, resource, integration or team to Slack, email or in-app.

Monitoring needs read access to source data, not just metadata, and only Editors and Admins work with monitors. Data CI/CD adds impact analysis to GitHub pull requests.

How the Data Quality Score Works

The Data Quality Score, a Premium and Enterprise feature, scores each table out of 100:

Component Maximum Breakdown
Stewardship 25 Owner 10; tags 10; answered questions 5
Usability 25 Table description 10; column descriptions 10; schema description 5
Freshness 20 Freshness monitor configured 10; passing 10
Accuracy 30 Nullness 10; uniqueness 10; other tests 10

Accuracy pulls from Secoda monitors or from dbt, Monte Carlo and Great Expectations tests. Tables land in Good, Moderate or Poor filters, and an automation can mark Good tables Verified.

50 of the 100 points reward owners, tags, answered questions and descriptions, so DQS measures stewardship as much as correctness: a table can score 50 without a single monitor or test.

Metaplane starts free with 10 monitored tables and 4 users. I shortlist Monte Carlo when deep data observability is the main requirement and Secoda when monitors need to sit beside searchable definitions and lineage in one workspace.

Secoda AI, Agents and the MCP Server

Secoda AI runs on a mix of OpenAI GPT-4o, Google Gemini 2.5 Flash and Lite, and Anthropic Claude Sonnet 4 and Opus 4.

  • Discovery: searches resources and documents and reads lineage and glossary context.
  • Analysis: runs SQL when permitted, builds charts and, since October 2025, prompt-built dashboards, and answers operational questions from monitors, query history, policies and tests.
  • Content: suggests descriptions, glossary terms and monitors; the AI description editor follows the Admin’s custom instructions.

AI Personas carry their own instructions, resource filters and Team access, so each team’s assistant answers from its own scope.

The assistant suggests monitors but cannot create or edit live ones. It also cannot read Policies, access requests, monitor configurations or audit logs, or change users, settings or integration credentials.

Agents are scheduled, read-only prompts with a minimum one-hour interval in cron syntax. Each run becomes a chat and uses AI credits, and Secoda publishes no credit allowance per plan.

  • Default sharing: workspace metadata goes to the model provider.
  • SQL sharing: with run_sql on, a sample of query data goes too.
  • Processing location: OpenAI and Anthropic process metadata in the US, including metadata from EU workspaces.
  • Retention and use: Secoda saves prompts, and its AI terms (updated July 17, 2025) let it use table and column names, query blocks, prompts and completions to improve Secoda AI, with no accuracy guarantee.

Connecting Claude or Cursor Through MCP

The MCP server, shipped in May 2025, exposes 8 tools that respect workspace permissions:

  • search_data_assets
  • search
  • run_sql
  • retrieve_entity
  • entity_lineage
  • glossary
  • get_secoda_docs
  • chart

Setup runs Secoda’s open-source Python server from GitHub with an Admin-generated API token and the regional API address, and both Claude and Cursor connect to it. Select Star also ships an MCP server, so MCP access alone does not separate the two.

Security, Deployment and the Atlassian Roadmap

Secoda offers three deployment options, gated by plan:

  • Multi-tenant cloud: US, EU or APAC hosting.
  • Single-tenant cloud: Premium and Enterprise, with its own VPC and database, a custom URL and upgrades run by Secoda.
  • Self-hosted: Enterprise, via the recommended AWS ECS Terraform module, Docker Compose for trials and testing, or Kubernetes on EKS, GKE or a generic Helm chart.

Self-hosting needs DevOps support from both your team and Secoda, can cost more and gets less frequent image updates than the cloud product; single tenancy on Premium isolates the workspace without that operating load.

  • Security: SOC 2 compliance with details on request, TLS 1.2+ and read-only source permissions.
  • Backups: metadata backed up every 5 minutes and kept for 24 months.
  • Identity: Google or Microsoft SSO and Okta or Azure AD SAML on every plan; SCIM from Premium.
  • Enterprise controls: SIEM logging, custom roles and access requests, still in beta for Snowflake and BigQuery.

What the Atlassian Deal Changes

Under the December 4, 2025 deal, the Secoda team joins Atlassian and parts of the infrastructure will move to Atlassian’s cloud platform, with technical plans and timelines promised in advance.

Atlassian EVP Tiffany To said Secoda will strengthen Rovo, Atlassian’s AI assistant. Secoda’s public changelog shows no release after v8.21.0 on October 28, 2025.

On October 7, 2026, Atlassian announced Insights, an app adding a data catalog, glossary, lineage and quality signals to the Teamwork Graph for paid Atlassian Cloud customers in a phased rollout. The announcement does not mention Secoda.

Get migration and continuity terms in writing before signing a multi-year Secoda contract. Switching to Select Star does not avoid the issue: Snowflake announced its acquisition on November 24, 2025.

How Does Secoda Compare to Competitors?

I shortlist Secoda for mid-market catalog plus monitoring, especially when unlimited viewers matter more than a free entry tier.

  • Atlan: the pick for large enterprises building an AI “context layer” across many teams. Atlan publishes no prices either, so it does not shorten the sales-led buying process. Secoda’s Core already bundles monitors and Data CI/CD, so ask both vendors to quote that same scope.
  • Select Star: my Snowflake-focused pick for automated column-level lineage and semantic model generation for Snowflake Cortex Analyst, with ERDs and an MCP server. It is now part of Snowflake and publishes no prices; pick it when the stack is Snowflake end to end.
  • Collate and OpenMetadata: my budget-first picks. OpenMetadata is open source under Apache 2.0, and managed Collate starts free with 5 users, 500 data assets and unlimited connectors. Premium covers 25 users and 5,000 assets and Enterprise 50 users and 10,000 assets, at unpublished prices. Collate’s free tier is the cheapest way to pilot a catalog before a Secoda sales tour.
  • DataHub: the option for engineering-heavy teams. DataHub Core is free and open source but means running the catalog yourself; managed DataHub Cloud is custom-priced with a 99.5% availability SLA.
  • Alation and Collibra: the enterprise end, with public AWS Marketplace contracts. Alation starts at $60,000 per 12 months and Collibra lists $170,000 per 12 months; shortlist them only if your budget already covers a contract that size.
  • Metaplane and Monte Carlo: my picks when broken data, not findability, is the problem. Metaplane, now part of Datadog, starts free with 10 monitored tables and 4 users; Pro bills per monitored table and allows 12 users plus unlimited viewers. Monte Carlo sells deep observability through Start, Scale, Enterprise and Business Critical tiers.

Choose Secoda over these alternatives when your team needs catalog, lineage, monitoring and AI together, has many readers but few editors, and accepts a negotiated contract.

How I Reviewed Secoda

I set up everything myself: an EU-region workspace, a Snowflake connection and its sync schedule, Teams and roles, a lineage graph and Secoda AI with custom instructions. I weighed Core’s included features against its upgrade gates, especially whether a team gets catalog, lineage and monitoring without Premium, and unlimited viewers against negotiated editor seats.

For business users, I assessed search, Questions, lineage coverage, monitor behavior and the DQS formula. I judged AI data handling and execution permissions separately from feature breadth, and weighed deployment responsibilities and the Atlassian migration. I compared Secoda with Atlan and Select Star on positioning, Collate and DataHub on free entry, and Monte Carlo on observability.

Prices current as of November 2026.

Secoda Review: Should You Buy Secoda for Your Data Catalog?

I recommend Secoda for a mid-market data team combining catalog and monitoring, particularly on Snowflake or BigQuery with dbt and Looker or Tableau. Core already includes the catalog, column-level lineage, monitors, Data CI/CD and Secoda AI. Unlimited viewers suit a few editors serving many business users, and the Slack Questions channel plus the Chrome extension put answers inside Slack and the BI tool.

Start with Core unless a specific requirement forces an upgrade. PII scanning, Policies, SCIM or DQS justify Premium; self-hosting, custom roles or access requests justify Enterprise, not a bigger reader count.

Choose Collate or DataHub Core for a free start, Monte Carlo for deep observability, or Select Star for Snowflake-oriented lineage and semantic models. Skip Secoda if procurement requires a published price or a vendor with a settled independent roadmap.

If you already pay for Atlassian Cloud, ask how Secoda relates to Insights before signing, and get migration and continuity terms into any multi-year contract.

Book a tour with Secoda, list your warehouse and BI tools, and ask for a written quote that states editor seats, AI credit allowance and migration terms.

FAQ

Is Secoda free?

No. Secoda has no free plan or self-serve trial; Core, Premium and Enterprise are quote-only plans that start with a sales tour request. For a free alternative, Collate’s free tier covers 5 users and 500 data assets, and DataHub Core is free and open source.

How much does Secoda cost?

Secoda publishes no prices for Core, Premium or Enterprise. Viewers are unlimited, while editor and admin seats and workspaces are set per contract. For comparison, Alation’s AWS Marketplace contract starts at $60,000 per 12 months and Collibra’s lists at $170,000 per 12 months. Ask Secoda for a written quote that states editor seats and AI credits.

Did Atlassian buy Secoda?

Yes. Atlassian completed the acquisition on December 4, 2025, for an undisclosed price, and the Secoda team joins Atlassian. Secoda said customer experience will not change while parts of its infrastructure move to Atlassian’s cloud platform, with technical plans and timelines published in advance. Its public changelog shows no release after v8.21.0 (October 28, 2025).

Does Secoda read my data?

By default, Secoda pulls metadata, not table contents. Previews, query blocks, profiling, monitors and metrics read source data when permitted, and you can switch them off. Preview shows 50 rows without storing them, while the other features can store results. Enabling Secoda AI’s run_sql tool also sends a data sample to the model provider.

Which AI models does Secoda AI use?

Secoda AI runs on GPT-4o, Gemini 2.5 Flash and Lite, and Claude Sonnet 4 and Opus 4. It is off by default under Settings > AI. Workspace metadata goes to the model provider, plus a query-data sample when run_sql is enabled; running SQL also requires the user’s query access on that integration.

Spotted a wrong price or a missing integration? Send a correction. A human reads every one.

Similar tools

Other tools in the same category, with the same card and the same honest pricing.

data.world

Enterprise catalog

data.world combines a knowledge graph, glossary, lineage, governance automations and a hosted MCP server in an enterprise data catalog.

Visit site

Zeenea

Enterprise catalog

Zeenea, now the Actian Data Intelligence Platform, combines a SaaS catalog, data marketplace and ODCS contracts with metadata collection on your own servers.

Visit site
Custom quote; commercial availability to confirm

Select Star

Modern catalog

A commercial metadata catalog with search, usage context, column-level lineage and AI documentation. Connector coverage varies; current commercial terms need a quote.

Visit site
Free software; self-hosted

Marquez

Open source

Open-source metadata service for collecting, storing, and visualizing OpenLineage events, with job/run history and dataset and column lineage.

Visit site
Free source; self-hosting costs extra

Amundsen

Open source

Archived open-source data discovery catalog with metadata search, ownership context and configurable lineage. Upstream archived September 2026.

Visit site
Free tier; paid plans by quote

Collate

Modern catalog

Managed data catalog built on OpenMetadata, combining discovery, lineage, data quality, governance, and optional AI workflows.

Visit site
Free core; hosting extra

OpenMetadata

Open source

An open-source metadata catalog combining discovery, lineage, data quality and governance, with self-hosted deployment and a separate managed service from Collate.

Visit site

Coalesce

Modern catalog

Column-aware visual transformation for Snowflake, Databricks, BigQuery and Fabric, with Catalog and Quality bundled; free Developer plan, Starter $150 per user per month.

Visit site

DataHub

Open source

Open-source metadata catalog from LinkedIn: column-level lineage, glossary, data contracts and an MCP server; managed DataHub Cloud is quote-only.

Visit site