HubSpot
- OAuth scopes control access. Confirm the write scopes for your target objects.
- Custom objects require Enterprise for the relevant HubSpot scopes.
Compare Clay with HubSpot or Salesforce by access, record IDs and write rules.
A connector checklist for founders, RevOps and GTM leads who need to update an existing CRM safely.
A&A | Checked 2026-11-01 | Documentation comparison; no live integration tests
Both pairs support imports and writes. Pilot only when access and update keys are clear. Ask about unknowns; hold when a required condition is missing.
| Connection | Existing-record updates | Check first |
|---|---|---|
| Clay × HubSpot | Object ID; associations use a separate action | Object type, write scopes and blank values |
| Clay × Salesforce | Record ID; upsert requires an external ID | API entitlement, user permissions and update key |
Keep your CRM fixed and compare the connector against your workflow.
Source: Clay HubSpot and Salesforce docs. Checked 2026-11-01. Decision method: A&A's proposal.
Clay Growth includes CRM auto-sync and enrichment. Check the CRM's own access requirements separately.
Source: Clay plans, HubSpot connector, Salesforce connector, Salesforce access. Checked 2026-11-01. Confirm regional terms and data residence with vendors.
Keep object type and original ID from import to write-back. A&A's proposal: do not identify a person from a company name or domain alone.
Source: HubSpot identifiers, Clay Salesforce. Checked 2026-11-01. Matching procedure: A&A's proposal; image: generated analogy.
Keep separate maps for imported fields and permitted writes. A display label alone does not define a valid update.
| Check | Clay × HubSpot | Clay × Salesforce |
|---|---|---|
| Read vs. write | Imports can include calculated fields; update actions show writable fields | Access follows the connected user's field permissions |
| Blank handling | Blank values are ignored by default during updates | Blank-ignore applies to text; blank numbers are dropped, not cleared |
| Example mapping | Research note → a dedicated text property, research_note | Industry candidate → exact, case-sensitive picklist API value |
Source: HubSpot connector, Salesforce connector, permissions FAQ. Checked 2026-11-01. Mappings are fictional examples proposed by A&A.
Design CRM → Clay → CRM as distinct steps. An import schedule alone does not establish safe write-back or conflict handling.
Select the records in scope. Scheduled sources have a separate option to refresh existing rows.
Retain IDs. A&A's proposal: use dedicated enrichment fields to protect values maintained by people.
Configure update actions and conditions. Test concurrent edits and deletion behavior; neither is verified here.
Source: Scheduled sources, HubSpot actions, Audiences. Checked 2026-11-01. Audiences is a separate feature: direct HubSpot write-back is unavailable there; Salesforce write-back is supported.
Scheduled-source docs list daily, weekly and monthly runs; hourly is Enterprise-only. Confirm your feature's cadence: Clay's pricing table lists different sync intervals.
| Constraint | Clay × HubSpot | Clay × Salesforce |
|---|---|---|
| Platform allowance | Public OAuth apps: 110 requests per 10 seconds per account; CRM Search excluded | Org-wide allowance varies; read remaining capacity through Limits |
| Import scope | Choose all objects or a list; reconcile expected and imported counts | Reports and non-SOQL-compatible list views are capped at 2,000 records |
| Measure in your pilot | Record count, API use and last completed update | Record count, API use and last completed update |
Source: Schedules, pricing, HubSpot limits, Salesforce monitoring, Clay imports. Checked 2026-11-01. Platform quotas are not measured Clay throughput.
A completed batch is not proof that every row was saved. Read uncertain results back before running them again.
| Check | Clay × HubSpot | Clay × Salesforce |
|---|---|---|
| Documented behavior | Connector-specific partial-failure handling remains unverified | Batch records are independent; some can succeed while others fail |
| Ask the vendor | Which errors retry, at what interval, and with what retry limit? | Which errors retry, at what interval, and with what retry limit? |
| Proposed test | Mix in an invalid field value; reconcile saved IDs with failed rows | Mix in an invalid restricted-picklist value; verify successful rows are not recreated |
Source: Clay Salesforce batch behavior. Checked 2026-11-01. HubSpot's connector overview does not establish a detailed retry policy. Tests are A&A's proposal and have not been run.
Job: refresh a dedicated company-research field each day. Decide what can enter a pilot, not what is production-ready.
| Assumed setup | Proposed decision | Evidence still needed |
|---|---|---|
| HubSpot Company + Clay Growth; IDs retained and writes allowed | Pilot updates to existing records in a dedicated text field | Blank protection, concurrent edits and read-back after failure |
| Salesforce Professional without API access | Hold connection setup until API entitlement is available | Written access confirmation, user permissions and a test environment |
Source: Clay plans, Salesforce API access. Checked 2026-11-01. Cases and decisions: A&A's proposal. Regional terms remain unverified for both. No customer or performance claim.
Connection: ________ Workflow: ________ Reviewer / date: ________
| Your requirement | Vendor answer / source | Test result / reviewer / date |
|---|---|---|
| Authentication, plan, data residence | __________________ | __________________ |
| Object types, IDs, writable fields | __________________ | __________________ |
| Direction, acceptable delay, API budget | __________________ | __________________ |
| Partial failure, retries, recovery | __________________ | __________________ |
Decision: □ Pilot □ Clarify □ Hold Required evidence still missing: ________
Source: A&A's proposal, 2026-11-01. Hold if record identity or write outcomes cannot be verified. Real-time guarantees, full migrations, deletion sync and outbound sending need separate designs and tests.
Sources & editorial note: AI researched and drafted this deck; a separate AI reviewer and machine checks reviewed it. No human reviewed the content.
Documentation supports a specification claim; it does not prove your setup works. Follow the source links on each slide before implementation.
| Question | English primary sources |
|---|---|
| Actions, authentication and fields | Clay HubSpot, Salesforce, permissions FAQ |
| Plans, schedules and feature scope | Clay pricing, Scheduled sources, Audiences |
| HubSpot IDs and API allowances | HubSpot Properties, usage guidelines |
| Salesforce access and capacity | Salesforce API access, API monitoring |
Source: Vendor documentation above. Recheck changed specifications and contract terms before rollout. The decision sheet, examples and test procedure are A&A's proposal. Illustration is conceptual.
Compare access, fields, direction and recovery against the same workflow. Resolve required unknowns before treating the connection as ready for production.
https://www.automateaugment.com/en/decks/ai-gtm-stack-compatibility
Related: Existing tools or custom development? · Build a comparable AI demo evaluation set
Explore A&A's AI-native GTM support or contact A&A.