> ## Documentation Index > Fetch the complete documentation index at: https://docs.dovetail.com/llms.txt > Use this file to discover all available pages before exploring further. # How credits and usage-based pricing work > Learn how Dovetail pricing combines a platform fee, users, and credits, which actions use credits, and how to estimate the credits your workspace needs. ## Overview Dovetail pricing has three parts: * **Platform:** the Dovetail platform itself, including home, search, access control, user management, user groups, data security, and compliance certifications. * **Users:** the people who can access the workspace. Users can create projects, set up channels, write docs, query data with search, Chat, and MCP connectors, and build agents and Digital Twins. * **Credits:** a shared usage balance for the workspace. Data processing and AI features, such as Chat, agents, and Digital Twins, use credits from this balance. Platform, users, and credits are billed together on one invoice, and annual contracts are paid upfront. For pricing, [contact Sales](https://dovetail.com/contact-sales/). *** ## Actions that use credits Credits measure data processing and AI usage. Some actions use a fixed number of credits. Others vary with the amount of work involved. | Action | Credits used | What counts | | :- | :- | :- | | Data processed | 1 credit per data point | A data point a [source](/help/sources) imports, a row uploaded from a CSV to a channel, or a file imported into a project as data, such as a video recording | | Data redacted | 1 more credit per data point | Redaction applied to a data point. A redacted data point uses 2 credits in total. Suggested redactions that aren’t applied don’t use credits. | | Chat response | Varies | The processing needed to generate the response | | Digital Twin response | Varies | The processing needed to generate the response | | Agent run | Varies | The processing needed to complete the run | ### Data processing Each data point uses 1 credit when it’s processed. That covers everything Dovetail does to it, including: * Transcoding, transcription, and speaker identification for audio and video * Text extraction from documents * Translation * Summarization, classification, and feedback extraction * Grouping feedback into ideas in Channels * Enrichment with details from a CRM such as Salesforce or HubSpot * Indexing for semantic search A source uses credits for every data point it imports, whether or not a channel uses its data. Adding a source’s data to a channel doesn’t use more credits. Historical data uses credits in the same way as new data, so a source’s [import period](/help/sources/create-a-source#import-period) affects how many credits it uses when you create it. Redaction uses 1 more credit for each data point it’s applied to. For a source with auto-redaction on, that’s every data point the source imports. See [Redact data from sources](/help/blur-and-redact#redact-data-from-sources). ### Chat, agents, and Digital Twins Chat responses, Digital Twin responses, and agent runs use credits based on the work needed to complete the task. A short question about a small amount of data uses fewer credits than a detailed analysis across thousands of customer conversations. Typical ranges are: | Action | Typical range | | :- | :- | | Chat response | 1–10 credits | | Chat response in Deep research mode | 5–30 credits | | Chat writing a doc | 1–10 credits | | Digital Twin or agent response | 3–25 credits | | Scheduled agent run, such as a stakeholder digest | 5–50 credits | | Event-triggered agent run | 1–15 credits | These ranges are indicative, and complex requests can use more. The number of credits depends on: * **The amount of data:** one call transcript uses fewer credits than a whole project of calls. * **Tool calls:** for example, when Chat or an agent creates or edits content in Dovetail. * **MCP connections:** for example, when Chat or an agent calls Gmail, Salesforce, or Figma. * **The model and reasoning level:** Deep research uses more credits than Fast. *** ## How the credit balance works Credits are a single balance shared by the whole workspace. Every user draws from the same balance, and admins can’t set credit budgets per user, team, or feature. * Credits are valid for the length of your contract, usually one year. * Unused credits expire at the end of the contract and don’t roll over at renewal. * Credits you buy are available immediately. * To buy more credits, contact your Account Executive or Customer Success Manager. *** ## Estimate how many credits you need Start with data volume. List each source of customer data and estimate how many data points it produces in a year. For example: | Data | Comes from | Per month | Per year | | :- | :- | -: | -: | | Customer interviews | Zoom | 60 | 720 | | Usability testing sessions | Zoom | 60 | 720 | | Survey responses | Qualtrics | 300 | 3,600 | | Support tickets | Intercom | 600 | 7,200 | | Sales calls | Gong | 200 | 2,400 | | **Total** | | **1,220** | **14,640** | Each data point uses 1 credit. If you plan to redact some of that data, add 1 credit for each data point you’ll redact. Remember to include the history you plan to import when you set up each source. Then add credits for Chat, agents, and Digital Twins, based on how your team plans to use them. The typical ranges in [Chat, agents, and Digital Twins](#chat-agents-and-digital-twins) can help. Your Account Executive can help you estimate your needs. *** ## Users All users in a workspace have the same abilities and can create content anywhere in the workspace, including at the top level. Access to folders and content is controlled in the **Share** dialog, where you can give access to individual users or to [user groups](/help/user-groups). Admins also manage workspace fields, tags, and templates. Admins count toward your number of users. You can add users when you sign your contract, or partway through it by contacting your Customer Success Manager. Users can join the workspace through an invite, SCIM provisioning, or SSO. *** ## Track your usage Workspace admins can review their renewal date, users, and credits in **Settings → Billing**. *** ## Existing customers on legacy paid plans Workspaces on legacy paid plans move to credits over time, and your Customer Success Manager will let you know before your workspace changes. When it does: * Your workspace receives an initial credit allowance based on your contract, historical usage, and renewal date. The allowance is designed to cover your usage until renewal. * Your workspace’s features don’t change. * For Channels, the monthly data point limit is removed, and Channels usage draws from your credit balance instead. * If you get close to using your whole allowance before renewal, your Customer Success Manager will contact you to discuss options. At renewal, your Customer Success Manager will work with you to move to a current plan with the right number of users and credits. You don’t need to do anything before then. *** ## FAQs No. Search, including semantic search, doesn’t use credits. When processing is complete, not when it starts or when you view the result. No. Failed, canceled, and retried processing jobs don’t use credits. No. Reprocessing data that’s already in Dovetail doesn’t use more credits. Importing the same data again as new data does. Jobs that are already queued or running complete without using credits. Yes. A Chat response uses credits for the processing it performed, even if it returns no results. Regenerated responses also use credits. No. Every user can use these features, and all usage draws from the shared workspace balance. Yes. Chat responses in [Slack and Teams](/help/chat/chat-in-slack-and-teams) use credits from your workspace’s balance, including answers to questions from people who aren’t Dovetail users. No. You can add users or credits during your contract, but you can’t reduce them until renewal. Prepaid credits aren’t refunded if you cancel or downgrade. Yes. Integrations, exports, webhooks, and the API work as before. Contact [support@dovetail.com](mailto:support@dovetail.com). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.