IBM Cloud Docs
Managing your plan

Documentation for the classic Watson Assistant experience has moved. For the most up-to-date version, see Managing your plan.

Managing your plan

This topic provides:

Plan information

Billing for the use of Watson Assistant is managed through your IBM Cloud® account.

The metrics that are used for billing purposes differ based on your plan type. You can be billed based on the number of API calls made to a service instance or on the number of active users who interact with the instance.

For answers to common questions about subscriptions, see How you're charged.

Explore the Watson Assistant service plan options.

Paid plan features

The following features are available only to users of a Plus or Enterprise plan. Plus or higher plans only

The following features are available only to users of Enterprise plans. Enterprise plan only

The plan type of the service instance you are currently using is displayed in the page header. You can upgrade from one plan type to another. For more information, see Upgrading.

User-based plans explained

Unlike API-based plans, which measure usage by the number of API calls made during a month, the Plus and Enterprise plans measure usage by the number of monthly active users.

A monthly active user (MAU) is any unique user who has at least one meaningful interaction with your assistant or custom application over the calendar billing month. A meaningful interaction is an exchange in which a user sends a request to your service and your service responds. Welcome messages that are displayed at the start of a conversation are not charged.

A unique user is recognized by the user ID that is associated with the person that interacts with your assistant. The web chat and other built-in integrations set this property for you automatically.

Specifying the user ID with the REST API

If you are using a custom client with the Watson Assistant API, you must set the user_id property in the message payload your client sends to the message method. The user_id property is specified at the root of the request body, as in this example:

{
  "input": {
    "message_type": "text",
    "text": "I want to cancel my order"
  },
  "user_id": "my_user_id"
}

In some older SDK versions, the user_id property is not supported as a top-level method parameter. As an alternative, you can specify user_id within the nested context.global.system object.

For more information about the user_id property, see the API reference documentation:

If the user ID is not specified

If you are using a custom client application and do not set a user_id value, the service automatically sets it to one of the following values:

  • session_id (v2 API only): A property defined in the v2 API that identifies a single conversation between a user and the assistant. A session ID is provided in /message API calls that are generated by the built-in integrations. The session ends when a user closes the chat window or after the inactivity time limit is reached.

    If you use the stateless v2 message API, you must specify the session_id with each message in an ongoing conversation (in context.global.session_id).

  • conversation_id (v1 API only): A property defined in the v1 API that is stored in the context object of a /message API call. This property can be used to identify multiple /message API calls that are associated with a single conversational exchange with one user. However, the same ID is only used if you explicitly retain the ID and pass it back with each request that is made as part of the same conversation. Otherwise, a new ID is generated for each new /message API call.

For example, if the same person chats with your assistant on three separate occasions over the same billing period, how you represent that user in the API call impacts how the interactions are billed. If you identify the user interaction with a user_id, it counts as one use. If you identify the user interaction with a session_id, then it counts as three uses (because there is a separate session that is created for each interaction).

Design any custom applications to capture a unique user_id or session_id and pass the information to Watson Assistant. Choose a non-human-identifiable ID that doesn't change throughout the customer lifecycle. For example, don't use a person's email address as the user ID. In fact, the user_id syntax must meet the requirements for header fields as defined in RFC 7230.

The built-in integrations derive the user ID in the following ways:

  • For Facebook integrations, the user_id property is set to the sender ID that Facebook provides in its payload.
  • For Slack integrations, the user_id property is a concatenation of the team ID, such as T09LVDR7Y, and the member ID of the user, such has W4F8K9JNF. For example: T09LVDR7YW4F8K9JNF.
  • For web chat, you can set the value of the user_id property.

Billing is managed per monthly active user per service instance. If a single user interacts with assistants that are hosted by different service instances that belong to the same plan, each interaction is treated as a separate use. You are billed for the user's interaction with each service instance separately.

Test activity charges

Test messages that you send from the Preview button are charged. For the preview, a random user_id is generated and stored in a cookie. The multiple interactions that a single tester has with the assistant embedded in the preview are recognized as coming from a single user and are charged accordingly. If you are doing your own test, running a scripted regression test for example, use a single user_id for all of the calls within your regression test. Other uses are flagged as abuse.

Handling anonymous users

If your custom application or assistant interacts with users who are anonymous, you can generate a randomized universally unique ID to represent each anonymous user. For more information about UUIDs, see RFC 4122.

  • For web chat, if you do not pass an identifier for the user when the session begins, the web chat creates one for you. It creates a first-party cookie with a generated anonymous ID. The cookie remains active for 45 days. If the same user returns to your site later in the month and chats with your assistant again, the web chat integration recognizes the user. And you are charged only once when the same anonymous user interacts with your assistant multiple times in a single month.

If an anonymous user logs in and later is identified as being the same person who submitted a request with a known ID, you are charged twice. Each message with a unique user ID is charged as an independent active user. To avoid this situation, you can prompt users to log in before you initiate a chat or you can use the anonymous user ID to represent the user consistently.

Data centers

IBM Cloud has a network of global data centers that provide performance benefits to its cloud services. See IBM Cloud global data centers for more details.

You can create Watson Assistant service instances that are hosted in the following data center locations:

Data center locations
Location Location code API location
Dallas us-south N/A
Frankfurt eu-de fra
Seoul kr-seo seo
Sydney au-syd syd
Tokyo jp-tok tok
London eu-gb lon
Washington DC us-east wdc

Upgrading your plan

You can explore the Watson Assistant service plan options to decide which plan is best for you.

The page header shows the plan you are using today. To upgrade your plan, complete these steps:

  1. Do one of the following things:

    • Trial plan only: The number of days that are left in your trial is displayed in the page header. To upgrade your plan, click Upgrade from the page header before your trial period ends.
    • For all other plan types, click Manage user icon, and then choose Upgrade from the menu.
  2. From here, you can see other available plan options. For most plan types, you can step through the upgrade process yourself.

    • If you upgrade to an Enterprise with Data Isolation plan, you cannot do an in-place upgrade of your service instance. An Enterprise with Data Isolation plan instance must be provisioned for you first.

    • When you upgrade from a legacy Standard plan, you change the metrics that are used for billing purposes. Instead of basing billing on API usage, the Plus plan bases billing on the number of monthly active users. If you built a custom app to deploy your assistant, you might need to update the app. Ensure that the API calls from the app include user ID information. For more information, see User-based plans explained.

    • You cannot change from a Trial plan to a Lite plan.

For answers to common questions about subscriptions, see the How you're charged.