Workato credits and usage metrics are the units in which Workato meters consumption of its platform under the usage-based billing model. The documentation defines the Workato credit as “a single, common billing unit used across every capability”. Customers buy a balance of credits, and every capability draws from that same balance.[1] Each capability measures its own activity in a capability-specific metric, and that usage “is converted into credits”.[1]
The model applies to customers on usage-based billing. Several metric pages note that some plans measure usage differently, for example in Business actions instead of Tasks, and refer readers to their contract.[2][3] Workato does not publish the number of credits each Task, API Call, Row or Genie action consumes. A software asset manager therefore needs the Order Form to turn metric counts into credit consumption.
Editions
Usage is measured independently of the platform edition. The edition determines which capabilities are available, and the credit balance pays for using them.[1] Several metrics only arise if the edition includes the capability. Per the pricing page, the API platform appears from Business MCP, while Data Pipelines, Event Streams and IDP appear from Enterprise MCP.[19] Editions are described in Workato platform editions, workspaces and add-ons. Add-ons such as additional concurrency and on-premises agents are billed separately from credit usage.[1]
Metrics
| Asset | Metric | Basis |
|---|---|---|
| Workflow recipes | Task | Each executed action step |
| Managed API | API Call | Each successful request; long responses count as more than one |
| API Proxy | API Proxy Call | Each successful request, regardless of response time |
| Data pipelines | Row | Rows extracted from successfully synced objects |
| Genies | Genie action | Each prompt sent to a Genie |
| Skills | Task | Billed when invoked over MCP; included in the Genie action when invoked by a Genie |
| MCP | Varies | The metric of the asset the MCP server invokes |
| MCP proxy | MCP Proxy Call | Each successful proxied request |
| IDP | Page | Each page processed, added on top |
| Event streams | Event | Each message published, added on top |
| Workato GO; Workflow apps | Seat | Each provisioned end user |
Source for the table: Understand usage.[2]
Tasks
A Task “is a single action step that a recipe executes”. Triggers and control-flow steps are not Tasks.[3] The rule applies to every connector type, including platform, community, SDK custom and Workato utility connectors. IF conditions, Repeat loops, Stop job and Handle errors steps are free. Steps nested inside them count each time they execute, so a loop over five rows with two actions inside produces ten Tasks plus any steps outside the loop.[3] A bulk or batch action counts as one Task however many records it processes. A failed action that succeeds on retry counts once. Mocked actions in Test Automation count.[3] AI by Workato action steps “count as more than one Task each”, at a rate the documentation leaves to the contract.[3]
Recipe functions follow the calling asset. When a workflow recipe calls a recipe function synchronously, the call counts as one Task and the function’s own action steps also count.[3] For Managed APIs the rule differs: a synchronous call is part of the API Call, while an asynchronous call runs as an independent workflow measured in Tasks.[4]
API platform
For Managed APIs, the number and type of steps inside the API recipe “don’t contribute to usage”. What counts is the request and its response time: a quick response is one API Call, and longer-running responses count as more.[4] Cached responses count the same as uncached ones. A request that ends with an expected 4xx or 5xx error defined in the recipe logic counts as successful.[4] API Proxy requests count one API Proxy Call each, whatever the response time. They consume fewer credits per call than Managed API requests.[5]
Data pipelines
Rows are counted when they are extracted from the source, across objects that sync successfully. Transformation and load steps add nothing.[6] Rows from a failed object are not counted until a later sync of that object succeeds. ETL or ELT built inside a workflow recipe is measured as Tasks, not Rows.[6]
Genies, skills and MCP
Each prompt sent to a Genie is one Genie action, whether it comes from a user, an App event or another Genie. Reasoning, knowledge retrieval and handoffs to other Genies are included.[7] Skills that a Genie invokes are included in the Genie action. IDP pages, Event streams messages and recipe functions called from a skill remain billable in their own metrics.[7]
When an external agent or client calls a Workato MCP server, usage is attributed “to the underlying asset, not the MCP server”. A skill invoked over MCP is billed as Tasks, and a Managed API invoked over MCP as API Calls.[10] A Workato proxy MCP server, which fronts a third-party MCP server, is counted in MCP Proxy Calls. Each tool invoked in a session counts separately.[11]
Pages, Events and seats
IDP counts each page of a standalone or multi-page document as one Page, on top of the Tasks or API Calls of the asset running it.[8] Event streams counts each published message, including each message in a batch publish, as one Event. Consumed messages do not count.[9] On some plans, publish steps are counted as Tasks and Event streams API usage is non-billable.[9]
Workato GO and Workflow apps are the user-based exceptions. Both count seats “by provisioned assignment, not by active use”, and reassigning a seat does not change the total.[12][13] Workflow app builders are not counted as seats. Recipes invoked by either product are billed separately in their own metric.[13]
Counting / floors
Three general rules apply across metrics.[2]
- Only successful operations count. Skipped steps, failed operations and cancelled jobs are excluded. Steps that completed before a failure, reruns, and operations that return an expected error are counted.
- Some usage is additive. IDP Pages and Event streams Events are added to the hosting asset’s usage instead of replacing it.
- No operation is counted twice. An IDP step is counted as Pages and “is not also counted as a workflow step”.
The billing FAQ confirms that rerun jobs add usage in the same way as the original run. In a parent-child recipe chain, the parent job does not show the children’s usage, because each child job reports its own.[18]
Workato’s worked examples show how a single business process consumes several metrics. A supplier-onboarding flow uses 1 API Call, 3 Pages and 1 Event. A procure-to-pay invoice flow uses 6 Tasks, 6 Pages and 1 Event. An AI-assisted recall uses 2 Genie actions, 5 Tasks and 12 Events.[14] To forecast usage, the documentation recommends identifying each asset in a use case, finding its metric, adding Pages or Events where used, and summing the result.[14]
Floors and caps are commercial. Credit volumes are bought per Order Form. Self-service plans come with a one-time credit grant (Workato Free) or a monthly credit limit (Workato Pro), covered in Workato platform editions, workspaces and add-ons. Under Capped Recipes, once a recipe reaches its predefined cap “its cost becomes fixed for that period”. The customer does not have to designate which recipes are high volume.[17] The same page states that Genies are priced on a fixed-cost basis, so variation in task and token usage inside a Genie interaction does not change the price.[17]
Monitoring usage
The Billing and Usage dashboard shows credits used against entitlement, the platform plan, the billing period and renewal date, and usage by type. It can be grouped by environment where the plan has several environments.[15] Automation HQ workspaces add a per-workspace view. Embedded admin workspaces show aggregate and per-customer usage, the latter against each end customer’s own billing period.[15] Access requires the Environment admin role, the legacy Admin role, or a custom role with the Usage dashboard privilege.[15]
On task-based plans, the task count is also reported per job, per recipe (Operations hub dashboard, Recipes API, RecipeOps connector) and per account in the Subscription tab. The Subscription tab shows a calendar-monthly quota.[16] Workato documents a pattern for capping individual recipes: the RecipeOps connector or Recipes API sets a quota on lifetime task count, and the recipe stops automatically when it exceeds that quota.[16]
Virtualization & partitioning
Usage metrics are independent of infrastructure. No metric refers to processors, hosts or virtual machines.[2] Partitioning is by workspace and environment. The dashboard can split credit usage by environment, by managed workspace in Automation HQ, and by customer workspace in Embedded.[15]
Cloud / BYOL
The metered capabilities run in Workato’s cloud. On-premises agents, which connect recipes to systems behind a firewall, are an add-on billed separately from credit usage.[1] Requests that a Workato proxy MCP server routes to a third-party MCP server are counted as MCP Proxy Calls, one per successful proxied request.[11]
Programs
- Capped Recipes: a ceiling on the task cost of an individual recipe. The model was introduced in July 2025 and was formerly called High-Volume Recipes.[17]
- Automation HQ pooled quota: one usage allowance shared and allocated across managed workspaces. See Workato licensing.
Out of scope
This article does not cover the plans of customers who joined before February 2024, which follow a separate model set by each customer’s contract.[20] It also does not cover the commercial conversion of metrics into credits, which Workato does not publish.[2] Platform limits such as concurrency and data retention are covered in Workato platform editions, workspaces and add-ons.