PEC.com
Component cost framework

Service catalogue cost in 2026: from spreadsheet to six figures

The service catalogue is the foundational data layer of any IDP. The cost of the catalogue UI is usually small; the cost of keeping its data fresh is usually larger and persists across every approach.

Spreadsheet (under 30 services)
$0-$20k / yr
Genuinely free at small scale. Stops being viable above 50 services.
Backstage or commercial
$30k-$200k / yr
Subscription plus 0.3 to 0.5 FTE on data freshness. The most common choice at scale.
Data freshness work
0.3-0.6 FTE
The often-underestimated cost line. Persists across every approach.

What the service catalogue is for

The service catalogue is the central registry of every service in your organisation. It answers the questions that engineers ask multiple times a day: who owns this service, where is its runbook, what does it depend on, what is its on-call rotation, what is its current health, what is its documentation, what scorecards apply to it.

The catalogue is the foundational data layer of an Internal Developer Platform. Almost every other capability of the IDP (scorecards, golden paths, self-service actions, FinOps integration, security dashboards) depends on the catalogue being accurate and complete. A catalogue with stale data poisons everything downstream; a catalogue with fresh data unlocks much of what makes a platform worth building.

Three approaches to delivering a catalogue

There are three distinct approaches, with different cost shapes:

  • Spreadsheet or wiki page. Manually maintained. Viable under 30 services. Cost: $0 to $20k a year of platform-engineer time on the quarterly review.
  • Open-source catalogue (typically Backstage). Self-hosted or hosted. The most common choice for organisations that have outgrown the spreadsheet. Cost: $0 to $120k a year subscription plus 0.3 to 0.5 FTE of platform-engineer time.
  • Commercial catalogue (Port, Cortex, OpsLevel, Compass). Subscription with stronger out-of-the-box features. Cost: $20k to $200k a year subscription plus similar platform-engineer time.

Across all three, the headline subscription cost (zero for spreadsheet, $30k to $200k for catalogue products) is usually the smaller line. The often-underestimated line is the data freshness work, which persists regardless of which approach you choose.

The data-freshness cost line

Data freshness is the work that keeps the catalogue's information accurate as services are created, deprecated, ownership changes, dependencies shift, and runbooks move. The common patterns:

  • Automated entity sync from Git. A nightly or hourly job that scans repository manifests (catalog-info.yaml in Backstage convention, or a similar file) and updates the catalogue with new and changed services. Build cost: 2 to 4 engineer-weeks. Ongoing maintenance: 0.05 to 0.1 FTE.
  • Automated ownership lookup. Integration with a CMDB, identity provider, or HRIS to map services to current owners as engineers change teams. Build cost: 2 to 6 engineer-weeks. Ongoing maintenance: 0.05 to 0.1 FTE.
  • Automated dependency mapping. Integration with observability data (service mesh, distributed tracing) to populate service dependencies in the catalogue. Build cost: 4 to 8 engineer-weeks. Ongoing maintenance: 0.1 to 0.2 FTE.
  • Periodic ownership review. Quarterly or twice-yearly review with engineering managers to confirm ownership data is current. Manual cost: 1 to 2 engineer-days per cycle for the platform team, plus a few hours from each engineering manager.
  • Anomaly detection. Alerts when catalogue data drifts (services with no owner, services with stale runbook links, services missing required fields). Build cost: 1 to 2 engineer-weeks. Ongoing: marginal.

A typical mid-sized organisation invests 0.3 to 0.6 FTE of platform-engineer time in data freshness long-term, or $70k to $140k a year of loaded cost. This is the line that organisations most often underestimate when budgeting for a catalogue. The catalogue UI is the visible part; the data freshness is the invisible work that makes the UI useful.

What an under-maintained catalogue costs

The most common failure mode is a catalogue that loses data freshness within six months of launch. The pattern:

  • Catalogue launches with great initial data quality (lots of attention during the launch).
  • Engineers start using it, find some stale data, lose a bit of trust.
  • Platform team is busy with other things, data freshness work slips.
  • More stale data accumulates as services are created and reassigned without the catalogue being updated.
  • Trust erodes further, engineers start asking "who owns this" in Slack again.
  • Within twelve to eighteen months, the catalogue is largely invisible. Engineers know it exists but do not use it.

The cost of this failure is significant but invisible. At a 100-engineer organisation, engineers asking "who owns this" several times a week each consume an aggregate 1 to 3 hours per engineer per week. At a $100 loaded hourly rate, that is $260k to $780k a year of product-engineering time spent on questions the catalogue should have answered.

Avoiding this failure mode is the single most important discipline in service-catalogue work. The investment in data freshness automation pays back several times over.

The ownership-review process

Ownership review is the human process that pairs with the automated data freshness work. The pattern that works:

  • Quarterly review with each engineering manager: confirm services owned by their team, confirm primary on-call rotation, confirm runbook links are current.
  • Anomaly-driven review: when an alert fires for a service with no owner, the platform team raises it to engineering leadership for resolution within two weeks.
  • Service-creation gate: a new service cannot be created without complete catalogue metadata. Enforced via the scaffolder (the new repository template includes the catalogue manifest with required fields).
  • Service-deprecation discipline: when a service is deprecated, the catalogue is updated to mark it deprecated and link to its replacement. Without this, deprecated services accumulate in the catalogue and dilute its signal.

The platform-team time investment in this review process is typically 1 to 2 engineer-days per quarter (the formal reviews) plus marginal time per week (anomaly response). Total ongoing cost: about 0.1 FTE of platform-engineer time, or $20k to $30k a year of loaded cost.

The data-source integrations cost

Beyond the catalogue's own data, value comes from integrating with other systems that have relevant per-service information:

  • CI/CD provider (latest deployment status, recent deploys).
  • Observability stack (SLO status, recent incidents, error budgets).
  • Incident management (active incidents, recent incident history, on-call schedule).
  • Secrets manager (secrets owned per service, rotation status).
  • Cloud provider (resources per service, cost per service).

Each integration costs 1 to 3 engineer-weeks to build (or comes stock with hosted Backstage and commercial IDPs for common vendors). Maintenance is typically low (refresh-token rotations, occasional vendor API changes) but non-zero. A typical mid-sized organisation has 6 to 10 catalogue integrations by year two, with build cost spread across the first 12 to 18 months and maintenance cost of about 0.05 FTE long-term.

Year-1 cost by approach at 100 engineers

For a 100-engineer organisation with about 100 to 200 services, year-1 catalogue cost by approach:

  • Spreadsheet plus discipline (only viable under 50 services): $10k to $30k.
  • Self-hosted Backstage catalogue: $200k to $400k (mostly platform-engineer time).
  • Hosted Backstage catalogue: $30k to $90k subscription plus $80k to $150k platform-engineer time. Total: $110k to $240k.
  • Commercial catalogue (Port, Cortex, OpsLevel, Compass): $20k to $100k subscription plus $80k to $150k platform-engineer time. Total: $100k to $250k.

The platform-engineer time line is roughly constant across the buy options ($80k to $150k for initial setup, integration, adoption). The subscription line is the variable. Self-hosted Backstage subscription is zero but headcount is roughly two to three times higher, reflecting the operations cost that hosted and commercial vendors absorb.

Cost crossover with the developer portal

The catalogue is one component of a developer portal. In practice it is impossible to discuss catalogue cost cleanly without discussing the wider portal cost, because most teams deliver the catalogue as part of a portal product rather than as a separate catalogue product.

For the wider developer-portal cost framework see /developer-portal-cost. For per-vendor cost framings see /backstage-cost, /backstage-hosted-cost, /port-cost, /cortex-cost, /opslevel-cost, /compass-cost.

Cost bands triangulated from Backstage project documentation, vendor public marketing pages, CNCF Platforms Working Group material, and IDP buyer case studies. Verified 2026-05-11.

Frequently asked questions

How much does a service catalogue cost?
Headline range $0 to $200k+ a year depending on approach. Spreadsheet (genuinely free, viable under 30 services): effectively zero. Open-source Backstage catalogue: $0 licence plus 0.3 to 0.5 FTE of platform-engineer time on data freshness, or $30k to $100k a year all-in. Hosted Backstage catalogue: $30k to $120k a year. Commercial IDP catalogue (Port, Cortex, OpsLevel, Compass): $20k to $200k a year subscription. The headline cost is usually the smaller line; the data-freshness automation cost is often larger and persists across all approaches.
What is the cost of data freshness?
Data freshness is the work to keep the service catalogue's information current as services are created, deprecated, and reassigned to new owners. Common patterns: automated entity sync from Git (services discovered from repo manifests), automated ownership lookup from a CMDB or identity provider, automated dependency mapping from observability data. Each of these costs 1 to 4 engineer-weeks to build and ongoing 0.05 to 0.2 FTE to maintain. A typical mid-sized organisation invests 0.3 to 0.6 FTE of platform-engineer time in data freshness long-term, or $70k to $140k a year of loaded cost.
What is the cost of an under-maintained service catalogue?
Genuinely high. The most common failure mode is a catalogue that loses data freshness within six months of launch. Engineers stop trusting it, fall back to Slack-based 'who owns this' questions, and the catalogue becomes invisible. The platform-team time spent building the catalogue becomes a sunk cost; the organisation has to repeat the work to rebuild trust later. Indirect cost of untrustworthy catalogue at a 100-engineer organisation: easily $100k to $300k a year in product-engineering time spent on questions the catalogue should answer.
Do I need a catalogue at all?
Below 30 services, no. A spreadsheet listing services with owners and runbook links, updated quarterly, works fine. Discovery is genuinely manageable at that scale through Slack and tribal knowledge. Above 50 services, the cost of not having a catalogue starts to exceed the cost of building one. Above 80 services, a catalogue with data-freshness discipline is the only practical option. Most organisations underestimate when they cross from 'no catalogue works' to 'catalogue is needed'; the symptom is the rise in 'who owns this' Slack messages, which typically becomes notable around 40 to 60 services.
What is the role of scorecards in catalogue cost?
Scorecards (automated rubrics that grade services on reliability, security, ownership, documentation, on-call) are sometimes counted as part of catalogue cost and sometimes counted separately. Scorecards add roughly 30 to 50 percent to total catalogue cost in the year they are introduced and 10 to 20 percent ongoing. The cost is mostly platform-team time on rubric authoring and tuning, not subscription. For organisations that treat scorecards as the primary value-driver of the catalogue, see /cortex-cost; for those that treat scorecards as one feature among many, the cost is usually rolled into the broader catalogue cost.
What is the typical year-1 catalogue deployment cost at 100 engineers?
For a 100-engineer organisation choosing hosted Backstage catalogue or a commercial IDP catalogue: $40k to $100k subscription plus $80k to $150k of platform-engineer time on initial catalogue import, ownership data cleanup, integration setup, and adoption work. Total year-one: $120k to $250k. For self-hosted Backstage catalogue: $0 subscription plus $200k to $400k of platform-engineer time. Total year-one: $200k to $400k. For 'spreadsheet plus discipline' upgraded with automation: $10k to $30k of platform-engineer time. Total: $10k to $30k, but only viable for organisations under 50 services.
Developer portal cost
Parent component.
Golden paths cost
Adjacent component.
Self-hosted Backstage
Cost framework.
Cortex IDP cost
Rubric-first commercial catalogue.
OpsLevel cost
Catalogue plus maturity alternative.
Run your numbers
Plug in your specific inputs.

Updated 2026-05-11