What changes qualify as breaking vs non-breaking. Use this to help guide your app development
When developing an integration on the HubSpot platform, you’ll need to be aware of the types of changes that HubSpot might make to its platform or data within. Below, learn about what HubSpot considers a breaking or non-breaking change, and use these lists to inform decisions you’ll make when working with HubSpot’s APIs.If a change is listed as a breaking change, we aim to provide you with 90 days notice if we do decide to make that change.
Please note:This notice period applies to developer products released to general availability. For beta and developer preview products, we’ll continue to update those regularly as we learn more about them. In certain situations, such as those related to security or privacy, we may shorten that 90 day timeframe.
As an example of correctly handling non-breaking changes, you can safely write (in pseudo code here) if err.code === 500 then retry but you should not write if err.message === "Uh oh, something went wrong" then retry. By asking yourself the question “if HubSpot makes a change listed as non-breaking, will this negatively impact the user experience?”, you can write code that will be easier to maintain and create a more reliable app.
Below, learn about what changes are breaking for incoming and outgoing RESTful APIs (e.g., contacts API and webhooks).Breaking changes
Removing an endpoint
Removing a field from the response (a “field” in this case is a JSON key/value pair)
Removing an option from a field in the request - If a request field could previously accept two String values, and we remove the ability to accept one of those, that would be a breaking change
Adding a new required field, header, or parameter in the request
Renaming a field
Changing the data type of a field
Changing field lengths past industry standard data limits (int, long, etc.)
Changing error response code or category
Changing auth types
Restricting scope
Reduce API limits
Making validation more stringent
Changing the default sort order
Significantly increase the size of a response (10x)
Non-breaking changes
Changing the error message
Adding a new option to a response (e.g., adding a new possible return value to a string field)
Adding a new optional input parameter
Property labels changing - Labels can be changed by HubSpot, application providers, and even customers. Integrations should not rely on property labels for their functionality.
Occasionally, there will be changes to the schema that describes the data share. Any change that makes querying a share fail due to changing an existing schema object is considered a breaking change.For example, renaming a table is a breaking change for the Snowflake Data Share integration because queries referencing the former name of the table will fail after the rename. In this case, HubSpot will add a new table with the same data to the share so that you can begin using the new name. The old table will continue to exist until the end of the 90 day notice period.Below, learn about what changes are considered breaking for the Snowflake Data Share.Breaking changes
Removing or renaming the database schema
Removing or renaming a table or view
Removing or renaming a column
Changing a column’s datatype
Column nullability
Non-breaking changesIf a change is listed as a non-breaking change, it will be added to the schema with no notice given. For example, adding a new column to the schema.This means that queries accessing shares should be as specific as possible when referencing schema objects. For example, you should always qualify tables by specifying the database schema they belong to. It also means that you shouldn’t use the SELECT * clause in your queries, because that might break when a new column is added.The following are examples of non-breaking changes:
Adding to the database schema
Adding a table or view
Adding a column
Changes to objects and their properties (e.g., the data within the objects and object_properties tables and views)
Last modified on February 10, 2026
Was this page helpful?
Assistant
Responses are generated using AI and may contain mistakes.