How a company that models its customer data in Snowflake (health scores, product usage, revenue and lifetime value) can put those numbers on the HubSpot records that reps, marketers and workflows act on, and keep HubSpot data in Snowflake for analysis. HubSpot's own Snowflake tools are the starting point; a custom build on Snowflake's APIs covers the rest.
Snowflake is where many companies calculate the numbers sales and marketing need: product usage, account health, lifetime value, churn risk. Those numbers only change outcomes when they reach the HubSpot record a rep opens or a workflow reads, and when HubSpot activity flows back into Snowflake for analysis.
Public documentation. The client's own admin issues credentials, with no vendor approval needed.
Open API. The SQL API is documented publicly. The client's Snowflake admin creates the service user and role.
Based on the vendor's public documentation. Access terms are confirmed with the vendor and the client in discovery.
Snowflake exposes its data through a documented REST SQL API and built-in change tracking, so modeled warehouse data can drive HubSpot records, lists and workflows.
Campaign Creators assessment
HubSpot's own tools are the starting point. With Data Hub Enterprise, HubSpot's Snowflake data sync reads a table or view on a schedule and creates or updates contacts, companies, deals and custom objects, and the Snowflake Data Share makes HubSpot objects, associations, events and property history queryable in Snowflake. The custom case is narrower. It covers what those tools don't: portals without Data Hub Enterprise, per-field write rules (fill blanks only, never overwrite a rep's edit, write only on change), custom events on the HubSpot timeline, near-real-time pushes driven by Snowflake streams, labeled associations, more than one Snowflake account or database feeding one portal, and governance over which columns and rows leave the warehouse. Compute is a planning item: every query runs on a Snowflake warehouse that bills credits while it runs.
Ratings are CC judgments from Snowflake's public documentation and HubSpot's knowledge base. Warehouse size, schedule and credit budget are set with the client's Snowflake admin during discovery.
Snowflake streams record inserts, updates and deletes on a table, and triggered tasks run only when a stream has new data (every 30 seconds by default, down to 10). Only changed rows go to HubSpot.
A warehouse value written every hour will overwrite a rep's manual edit unless rules say otherwise. Each synced property gets one owner and one write rule, written into the SOW.
Decide which flows HubSpot's data sync covers under Data Hub Enterprise and which need the custom build, and have the client's Snowflake admin approve the service user, role, warehouse and network policy.
Seven parts of Snowflake and HubSpot matter here. HubSpot's two Snowflake tools cover standard flows; the custom build uses Snowflake's APIs and in-database features for the rest.
Syncs a Snowflake table or view one way into HubSpot on a schedule you choose. Rules can be create and update, create only, or update only.
Makes HubSpot data queryable with SQL inside the client's Snowflake account, so models can join CRM data with product and finance data.
Snowflake's REST API for running SQL. The integration layer uses it to read approved views and to record HubSpot IDs and sync status back in Snowflake.
A stream tracks row changes on a model table. A triggered task runs when the stream has data and writes only the changes to an outbound queue table.
Scores, rollups and payload shaping run where the data lives, so the integration layer receives HubSpot-ready rows instead of raw tables.
Lets a Python procedure in Snowflake call the HubSpot API directly over an allowed network rule, with the token held in a Snowflake secret. Useful when the client wants no middle tier.
Every query runs on a warehouse that bills per second, with a 60-second minimum each time it starts. Schedule, warehouse size and auto-suspend are agreed before build.
Models stay in Snowflake. A CC-built integration layer reads only approved views and changed rows, applies governance and write rules, matches each row to the right HubSpot record, and writes in batches. HubSpot data returns to Snowflake through HubSpot's Data Share.
Scores and rollups are calculated once in the warehouse. HubSpot stores the result on the record where people and workflows use it.
Streams and change-only reads keep HubSpot API calls and Snowflake credits low, and make every sync run small enough to replay.
The integration reads only views the client approves. Restricted columns and rows never leave Snowflake.
Each phase ends on a gate the client signs off before the next begins. The first data set goes live while the next is being built.
List the decisions sales, marketing and service should make from warehouse data, and the Snowflake tables or models that hold each number.
For each data set, decide whether HubSpot's Snowflake data sync covers it under Data Hub Enterprise, or whether it needs write rules, events, associations or timing that call for the custom build.
The client's Snowflake admin creates a service user with key-pair authentication, a read-only role, a dedicated warehouse and the approved views. Compliance signs off the column list.
Create the properties, custom objects, association labels and custom event definitions in a HubSpot sandbox.
Build streams, triggered tasks and the integration layer for the first data set, with identity matching, write rules, idempotent upserts and a replay queue.
Load current values for all matched records once, in batches sized to HubSpot's API limits and the agreed warehouse budget.
Build lists and workflows with sends paused, run in parallel for two weeks, review credit use, then switch on.
Warehouse column names are defined by each client's models, so the source column describes the data plainly. Exact table, view and column names are confirmed during discovery.
| Snowflake field | HubSpot property | Object | Direction | Notes |
|---|---|---|---|---|
| Identity & keys | ||||
| HubSpot record ID (stored in Snowflake) | hs_object_id | Contact / Company | ↔ Both | Best match key once known. Written back to Snowflake after first match. |
| Email address | Contact | SNOW → HS | Match key for contacts without a stored HubSpot ID. | |
| Company domain | domain | Company | SNOW → HS | Match key for companies without a stored HubSpot ID. |
| Customer account ID | snow_account_id | Company | SNOW → HS | Client's own account key. Unique property; used for batch upserts. |
| User ID | snow_user_id | Contact | SNOW → HS | Product user key. Unique property. |
| Sync status and last synced | Status table in Snowflake | — | HS → SNOW | Written back by the integration layer for audit. |
| Account metrics · companies | ||||
| Annual recurring revenue | snow_arr | Company | SNOW → HS | From the revenue model. Snowflake owns this field. |
| Plan or tier | snow_plan_tier | Company | SNOW → HS | Dropdown values agreed in discovery. |
| Lifetime value | snow_ltv | Company | SNOW → HS | Model output. Snowflake owns this field. |
| Health score | snow_health_score | Company | SNOW → HS | Model output, 0 to 100. |
| Churn risk band | snow_churn_risk | Company | SNOW → HS | Low, medium, high. Can trigger a CSM workflow. |
| Seats purchased vs used | snow_seats_used | Company | SNOW → HS | Rollup from product data. |
| Product usage · contacts | ||||
| Last active date | snow_last_active | Contact | SNOW → HS | Write only when changed. |
| Active days, last 30 | snow_active_days_30 | Contact | SNOW → HS | Rollup from product events. |
| Key feature adopted | snow_features_adopted | Contact | SNOW → HS | Multiple checkboxes. |
| Product-qualified flag | snow_pql | Contact | SNOW → HS | Can move lifecycle stage by workflow, not by direct write. |
| Milestone reached | Custom event: Product milestone | Contact | SNOW → HS | Timeline entry that workflows can trigger on. |
| Phone, job title | phone, jobtitle | Contact | SNOW → HS | Fill blanks only. HubSpot edits win. |
| Custom objects & associations | ||||
| Subscription or license | Subscription (custom object) | Custom object | SNOW → HS | Plan, term, renewal date. Object design confirmed in discovery. |
| Workspace or product account | Workspace (custom object) | Custom object | SNOW → HS | For multi-workspace customers. |
| Account-to-user link | Association (labeled) | Company ↔ Contact | SNOW → HS | Labels such as Admin or Billing contact. |
| Subscription-to-company link | Association | Custom object ↔ Company | SNOW → HS | Set from the customer account ID. |
| Renewal opportunity | Deal in renewal pipeline | Deal | SNOW → HS | Created when renewal date enters the agreed window. |
| HubSpot → Snowflake | ||||
| Contacts, companies, deals | HubSpot Data Share views | All objects | HS → SNOW | Queried in Snowflake for attribution and models. |
| Email and web events | HubSpot Data Share events | Events | HS → SNOW | Joins marketing activity to product usage. |
| Property history | HubSpot Data Share history | All objects | HS → SNOW | Most recent 45 values per contact, 20 for other objects. |
Snowflake access is self-serve for the client's admin. Values Snowflake does not publish are marked for confirmation.
| Topic | Snowflake | HubSpot |
|---|---|---|
| Access model | A service user and dedicated role created by the client's Snowflake admin, with access to approved views only. | Private app token for the custom build. Data Hub Enterprise for HubSpot's data sync and Data Share. |
| Interfaces | SQL API (REST) to submit statements, check status and cancel; Snowpark and stored procedures in the warehouse; external access integrations for outbound calls. | REST APIs: CRM objects, batch upserts, associations, custom events, imports. |
| Authentication | Key-pair JWT (valid at most one hour), OAuth, or workload identity federation. | Private app access token or OAuth 2.0, scoped per object. |
| Permissions | Usage on one warehouse, database and schema; select on approved views. Nothing broader. | Scopes for the objects, associations and events in scope. |
| Freshness | Triggered tasks run when a stream has data (30 seconds by default, 10 at minimum). Data Share live schema refreshes up to every 15 minutes. | Writes are immediate; batch endpoints handle volume. |
| Cost & limits | Compute billed per second while a warehouse runs, 60-second minimum per start. SQL API rate limits: not published. Confirm with Snowflake. | Per-10-second burst and daily limits by subscription tier. |
| Network & sensitive data | Network policies may restrict inbound access; allowlist the integration layer's addresses and HubSpot's ranges if the data sync is used. | Sensitive data properties available on Enterprise; field-level permissions by team. |
Severity reflects impact on the SOW, not technical difficulty alone.
Each read resumes a warehouse and bills at least a minute. Change-only reads, a small dedicated warehouse, auto-suspend and an agreed schedule keep the credit budget predictable.
If both systems write the same property, the last write wins. Every synced field gets an owner and a rule (overwrite, fill blanks, or never) before the build starts.
HubSpot's data sync needs Data Hub Enterprise and runs on a schedule. Agreeing which flows it handles avoids building twice and keeps the custom scope narrow.
A full table pushed every hour can exhaust daily limits. The integration sends changed rows only and uses batch endpoints.
Only approved columns and rows leave Snowflake. The views enforce this, and compliance signs off the column list.
A stream that is not read before its retention window ends becomes stale. Triggered tasks and monitoring keep streams current.
Health, fit and churn risk sit on the company record, so reps and CSMs see them in views, sequences and tasks.
A usage milestone becomes a HubSpot custom event that enrolls the contact in the right workflow within minutes.
Subscriptions as custom objects create renewal deals on time, with seat and usage data attached.
HubSpot's Data Share puts marketing and sales activity next to revenue in Snowflake, so attribution models use real outcomes.
Revenue and health are calculated once in Snowflake, so HubSpot reports and warehouse dashboards agree.
Scheduled CSV imports of warehouse exports are replaced by an audited, replayable sync.
HubSpot's Snowflake data sync and Data Share handle standard, scheduled flows under Data Hub Enterprise. The custom build is scoped only to what those tools don't cover, so the client pays for the gap, not for a rebuild.
All custom properties use the snow_ prefix in a Snowflake property group. Values are calculated in Snowflake so every report agrees.
| Internal name | Label | Object | Type | Source / logic |
|---|---|---|---|---|
| snow_account_id | Snowflake Account ID | Company | Single-line text | Client's customer account key; unique |
| snow_user_id | Product User ID | Contact | Single-line text | Product user key; unique |
| snow_arr | ARR | Company | Number (currency) | Revenue model output |
| snow_plan_tier | Plan Tier | Company | Dropdown | Current plan from billing model |
| snow_ltv | Lifetime Value | Company | Number (currency) | LTV model output |
| snow_health_score | Health Score | Company | Number | Model output, 0 to 100 |
| snow_churn_risk | Churn Risk | Company | Dropdown | Low, medium, high |
| snow_seats_used | Seats Used | Company | Number | Active seats from product data |
| snow_last_active | Last Active Date | Contact | Date picker | Latest product activity |
| snow_active_days_30 | Active Days (30) | Contact | Number | Distinct active days, last 30 |
| snow_features_adopted | Features Adopted | Contact | Multiple checkboxes | Key features used at least once |
| snow_pql | Product Qualified | Contact | Single checkbox | PQL rule in Snowflake model |
Choose the data sets and fields for this engagement and the direction each one syncs. Required fields are the keys the integration cannot match without. Copy the summary into the SOW or a reply to the client.
Based on public documentation from Snowflake and HubSpot. Credit budget, network policy and table names are confirmed with the client's Snowflake admin during discovery.
Yes. With Data Hub Enterprise, HubSpot's Snowflake data sync reads a table or view into HubSpot on a schedule, and the Snowflake Data Share makes HubSpot data queryable in Snowflake. Campaign Creators adds a custom build on Snowflake's SQL API for what those tools don't cover, such as per-field write rules, custom events, near-real-time pushes and column-level governance.
Modeled Snowflake data such as product usage, health scores, revenue and lifetime value is written to HubSpot contacts, companies, deals and custom objects, and can create custom events that trigger workflows. HubSpot objects, associations, events and property history flow back to Snowflake through the Data Share, along with HubSpot IDs and sync status.
Scores and rollups are calculated once in Snowflake, and HubSpot stores the result where people and workflows use it. Each synced property gets one owner and one write rule in the SOW, so a warehouse value never overwrites a rep's edit unless agreed. Restricted columns and rows never leave Snowflake.
Seven phases, each closed by a client sign-off: a use-case inventory, the native-or-custom decision for each data set, access and governance (a service user with key-pair authentication, a read-only role and a dedicated warehouse), the HubSpot data model, the build, a backfill, and a two-week parallel run before workflows are switched on.
Timeline and cost depend on the data sets in scope, how many need the custom build rather than HubSpot's native sync, and the agreed Snowflake compute budget. Campaign Creators confirms these in discovery and provides a fixed scope before the build starts.