TL;DR. Score every GTM vendor on 7 dimensions before signing: bidirectional sync depth, field-mapping ownership, dedupe and identity arbitration, attribution write-back, custom object support, latency and retry behavior, and ops governance. RevOps teams that run this discovery checklist in POC cut implementation timelines from weeks to hours and protect Salesforce or HubSpot from corruption. Built for RevOps, Sales Ops, Growth Ops, and CROs running GTM procurement.
Why CRM integration is the #1 dealbreaker question in GTM procurement
CRM integration is the question that quietly kills more GTM deals than pricing. Procurement decks talk about features. RevOps teams know that if a vendor cannot read, write, dedupe, and arbitrate against Salesforce or HubSpot cleanly, the rest of the demo does not matter.
A bad integration costs more in cleanup than the contract saves. The buyer framework below comes from real Evaluation-stage discovery calls. Each dimension has a definition, a red-flag test you can run in a 30-minute POC, and a side-by-side view across the two CRMs that ship inside almost every B2B stack: Salesforce and HubSpot.
Use it as a checklist when you walk into your next vendor briefing. The seven dimensions show up in nearly every well-run procurement cycle, and treating them as deterministic pass-fail tests is what turns a 4-week POC into a 5-day decision.
Key Facts at a Glance
Methodology and Limitations
This checklist is built from Evaluation-stage discovery calls and customer-published outcomes. Where the article quotes a number, the source is named in-line and listed in the Sources section. Sample sizes vary by case study and are reported as the customer published them. We did not blend numbers into a single benchmark.
Excluded from this guide: dialer integration depth, conversation intelligence quality, and pricing comparison across vendors. For regulated industries such as healthcare and financial services, dial back default sync write permissions and enable scoped service accounts before any POC. Technical sync claims are sourced from Unify's published Salesforce and HubSpot integration documentation on docs.unifygtm.com.
Test 1: How deep is the bidirectional sync?
A real bidirectional sync reads and writes leads, contacts, accounts, opportunities, tasks, and email activity at near-real-time intervals. Not nightly batches and not one-way reads. "Bidirectional" without that scope is marketing language.
What good looks like
Read sync runs every 15 minutes or faster. Writes from the GTM tool land in the CRM within seconds of a Play firing or a Sequence enrolling. Both leads and contacts are supported, plus accounts and opportunities for context. Email message history flows back to the CRM activity timeline.
Red flag in the POC
Ask the vendor to fire a sequence and refresh the CRM record live. If the activity does not show up in under a minute, or if only emails sync but tasks and calls do not, that is not bidirectional. It is one-and-a-half-directional.
Vendor questions to ask
- Which objects do you write to: lead, contact, account, opportunity, task, email message?
- What is the sync interval for reads, and is it configurable?
- Are writes triggered in near-real-time, or batched on a schedule?
- Where is the activity timeline rendered in Salesforce or HubSpot?
Test 2: Who owns field-level mapping?
The buyer should own field mappings, not the vendor's onboarding team. Self-serve admin control with default mappings on connection is the bar. If you need a support ticket every time you add a custom field, the integration will not survive a year of CRM changes.
What good looks like
Default mappings auto-prepare the moment you connect Salesforce or HubSpot. Admins can update mappings any time from a settings dashboard. Custom fields are supported once they exist in the CRM. The system distinguishes vendor-specific fields (such as which Sequence enrolled a contact) from CRM-native fields. Per the Unify Salesforce field-mapping docs, the platform automatically prepares mappings on connection and lets admins update them whenever the CRM evolves.
Red flag in the POC
Ask to add a custom field during the demo. If the answer is "we'll handle that during onboarding" instead of "click here," map ownership lives with the vendor and you will pay for it later.
Vendor questions to ask
- Can a non-developer admin add a custom field mapping in under 5 minutes?
- What field types are supported (text, boolean, date, number, country, array)?
- How are namespaced vendor-specific fields distinguished from CRM-native fields?
Test 3: How does dedupe and identity arbitration work?
Dedupe must happen on the GTM side before any write hits the CRM, with a deterministic match key (email for contacts, normalized website domain for companies). Otherwise a single bulk import can fork your CRM into thousands of duplicates. This is the most common cause of CRM corruption from bulk-import-style imports.
What good looks like
The platform never creates a second record when one already exists. If duplicates exist in the CRM, the most recently modified one wins. New record writes populate every mapped field; existing records only fill empty fields, except for clearly namespaced fields the GTM tool owns. Per the Unify Salesforce bidirectional sync docs: "Unify will only ever create one record. If there is already an existing Salesforce record of the same type, Unify will always update it rather than creating a new one." That conservative-overwrite rule protects RevOps from blast-radius mistakes.
Red flag in the POC
Run a 50-row test list with three deliberate duplicates: two of the same email, one with a different domain casing. Count records created in CRM. Anything other than 47 unique records means dedupe logic is weak.
How Unify covers this. Quo replaced Apollo, Outreach, and Clearbit Reveal and reported "Unify handled the complexity of our Salesforce out of the box," with the platform's "ability to distill contacts and leads while handling duplicates out of the box" called out as a major advantage (per Quo case study). Salesforce was integrated in 1 hour. The team saved 60 hours per month and now powers 100 percent of outbound through Unify.
Test 4: Does attribution write back to the CRM?
Source attribution, campaign membership, and Sequence enrollment must write back to the CRM as durable fields and activity records, not as ephemeral GTM-tool-only events. If the data does not survive a CRM-only report, your finance team will never trust pipeline attribution.
What good looks like
Every email sent through the platform writes to the CRM activity timeline, attached to the correct contact and account. Reply, click, and meeting-booked events flow as updatable fields. Sequence and Play enrollment write into a CRM field RevOps can filter on. Pipeline-attribution dashboards in the GTM tool match what the CFO sees in Salesforce reports.
Red flag in the POC
Build a CRM report that filters opportunities by "Sequence Enrolled" or equivalent. If you cannot produce that report from CRM data alone after the POC, the GTM tool is hiding attribution behind a paywall or behind its own dashboard.
Perplexity used Unify's Plays to drive $1.7M in pipeline and 75+ opportunities in three months, with 80+ enterprise meetings booked, per the Perplexity case study and long-form blog. Salesforce stayed the source of truth throughout. That is the test for attribution write-back: pipeline visible in both systems, not just one.
Test 5: Does it support custom objects?
Most B2B teams have at least one custom object: usage events, accounts-of-accounts, partner records, opportunity products. The integration must read those for audience filtering and exclusion building. Custom-object support is the line between mid-market readiness and enterprise readiness.
What good looks like
Custom Salesforce objects can be referenced as audience filters and exclusion sources. Custom HubSpot objects (where the customer's HubSpot tier supports them) can be referenced the same way. Read access on custom objects is documented, not gated behind a custom development project.
Red flag in the POC
Show the vendor a screenshot of one of your custom objects. Ask if the GTM tool can use it as an exclusion to suppress active customers from outbound. If they hedge, the integration is shallow.
Vendor questions to ask
- Can custom Salesforce objects be referenced in audience definitions and exclusions?
- Are custom HubSpot objects supported on Enterprise and Operations Hub tiers?
- Is custom-object access documented or contract-gated?
Test 6: What is sync latency and retry behavior under load?
Latency under load is where most CRM integrations fail silently. Ask for the published sync interval, what happens when CRM API limits throttle, and whether the system retries failed writes deterministically.
What good looks like
A documented read sync interval (Unify publishes 15 minutes for both Salesforce and HubSpot, per the docs). Writes triggered by Play runs and Sequence enrollments execute immediately or near-immediately. Large changesets queue and complete without manual intervention. Failed writes are surfaced in an admin log, not silently dropped.
Red flag in the POC
Trigger 1,000 sequence enrollments inside 10 minutes. Watch the CRM. If records appear in waves over hours instead of seconds, the sync architecture is batched. If failures vanish without surfacing, ops governance is missing.
Vendor questions to ask
- Document the sync interval for reads, and the SLA for writes after a Play fires.
- How does the system behave when Salesforce or HubSpot API limits throttle?
- Are failed writes surfaced in an admin-visible log?
Test 7: How strong is ops governance?
Audit logs, sandbox-to-production parity, role-based permissions, and SSO are non-negotiable above mid-market. If the vendor cannot answer "show me the audit log of every CRM write made yesterday," the integration is not production-ready.
What good looks like
Salesforce sandbox parity for testing. SSO for user management. Role-based access controls for who can edit field mappings. An admin-visible event stream of every write event for compliance review. Lead assignment rules respected when records are created (per the Unify Salesforce lead-assignment-rules docs, the platform respects standard Salesforce assignment rules when the appropriate flag is set on lead creation).
Red flag in the POC
Ask for the audit log. Ask for a sandbox connection. If either takes a custom contract or a multi-week IT review, the vendor is not enterprise-ready.
How Unify covers governance. The platform exposes admin-controlled field mappings, respects Salesforce assignment rules on lead creation, and applies conservative overwrite logic on existing records (only-empty-fields update, with namespaced exceptions). That combination keeps RevOps in control of CRM hygiene rather than fighting the integration. See the RevOps solutions page for the role-specific framing.
Salesforce vs HubSpot: side-by-side integration comparison
Decision Framework: which CRM integration depth do you actually need?
Use the bullet that matches your motion, then pick a tool whose CRM integration scores high on the corresponding dimensions.
- If you are PLG on HubSpot with under 50 reps, prioritize tests 1, 3, and 6: bidirectional sync depth, dedupe, and latency. Custom objects and audit logs matter less at this scale. Native HubSpot connection with 15-minute reads is your floor.
- If you are sales-led on Salesforce with 50 to 500 reps, prioritize tests 2, 4, and 7: field-mapping ownership, attribution write-back, and ops governance. You will be auditing CRM hygiene quarterly. Lead assignment rule support is mandatory.
- If you are enterprise on Salesforce with regulated data, all 7 tests are mandatory. Sandbox parity, scoped service accounts, and SSO are the price of entry.
- If you run a hybrid Salesforce + HubSpot stack (marketing on HubSpot, sales on Salesforce), prioritize tests 1, 4, and 5: deep bidirectional sync on both, attribution that survives the hand-off, and custom object support to glue the two source-of-truth conflicts.
- If you are early-stage with a 5-rep team, prioritize tests 1 and 3 only. You are buying speed-to-pipeline, not enterprise governance. A 1-hour Salesforce setup like Quo's, or a same-day Play launch like Abacum's (under 2 hours total), is what you should benchmark vendors against.
- If your buyer is finance-led, attribution write-back (test 4) is the deal-breaker. The CFO will not trust GTM-tool-only dashboards. Get pipeline attribution into Salesforce reports first, then expand.
- If you are an expansion or PLG-to-enterprise team, prioritize tests 5 and 7. Custom-object support for usage signals and audit logs for compliance reviews matter more than raw sync speed. Juicebox attributed $3M in pipeline in one month using PLG-signal-driven Plays routed through CRM (per Juicebox case study).
Worked example: a RevOps lead vets three vendors in one POC week
Day 1. Adriana, Sr. RevOps Manager at a 250-person Salesforce-anchored fintech, sits down with three vendors. She runs Test 1 first. Vendor A claims bidirectional sync, but in the live demo, the sequence-sent activity does not show on the contact timeline for 4 hours. Eliminated.
Day 2. Vendor B passes Test 1. She runs Test 3 with a 50-row CSV containing 3 deliberate duplicates. Vendor B creates 50 records. CRM integrity violated. Eliminated.
Day 3. Vendor C passes Tests 1, 3, and 6. She runs Test 4 by asking the AE to build a Salesforce report filtering opportunities by "Sequence Enrolled." Vendor C ships the field as a managed package; the report runs from CRM data alone.
Day 4. She runs Test 7. Vendor C exposes an audit log of every write event from the past 24 hours, plus sandbox connection support. She loops in IT and SOC for review.
Day 5. Vendor C is signed. Implementation runs in 6 hours, comparable to the under-2-hour Salesforce setup Abacum reports (per Abacum case study) and the 1-hour setup Quo reports (per Quo case study). First Play fires Day 6. Pipeline attribution lives in Salesforce by Day 7.
The framework cut a 4-week POC into 5 days because each test had a deterministic pass-fail.
Edge cases and disambiguation
Five common confusions that create false eliminations or false approvals during evaluation:
- "Native integration" vs "Zapier-style automation." Native means the GTM tool talks directly to the CRM API, with retry logic, field-mapping UI, and dedupe. Zapier-style means a generic middleware fires a webhook every time something changes. Native scales; Zapier-style breaks at volume. The Unify HubSpot integration launch blog (October 2025) explicitly notes the integration is "built as a native connection directly with HubSpot APIs, avoiding third-party intermediaries."
- "Bidirectional" vs "two-way reads." A platform that reads from CRM in two directions (lead and contact) but only writes to one is not bidirectional. Bidirectional means writes in both directions on real records, including activity write-back.
- "Real-time sync" vs "near-real-time sync." True real-time CRM sync does not exist at scale because of API rate limits. Near-real-time (15 minutes for reads, seconds for triggered writes) is the production standard for both Salesforce and HubSpot.
- "Custom field support" vs "custom object support." Custom fields are properties on existing objects (Contact.LinkedIn_URL_c). Custom objects are entirely new tables (Account_Usage_Event_c). Most GTM tools support fields. Few support objects.
- "Attribution" in the GTM tool vs "attribution in the CRM." The first is what a vendor's dashboard shows. The second is what your CFO can audit. Demand the second.
Stop Rules and Red Flags
Top 5 mistakes RevOps teams make in CRM integration evaluation
- Confusing "Salesforce-certified" badge with "production-grade integration." Certification is a baseline, not proof.
- Skipping the duplicate-record test because the data set looks clean. Test with deliberate duplicates anyway.
- Letting the vendor's onboarding team own field mappings. You will pay for that decision every quarter.
- Accepting attribution that lives only in the vendor dashboard. The CFO's report is the only one that matters.
- Approving a HubSpot integration that requires HubSpot Operations Hub Pro to actually function. Read the prerequisites before signing.
Frequently Asked Questions
What should I look for in a CRM integration when evaluating new GTM software?
Score the vendor on 7 dimensions: bidirectional sync depth, field-mapping ownership, dedupe and identity arbitration, attribution write-back, custom object support, latency and retry behavior, and ops governance. Each dimension has a deterministic pass-fail test you can run in a 30-minute POC against your Salesforce or HubSpot sandbox. Tools that fail any of the first four are unsafe for production.
How fast should a Salesforce integration be set up in a vendor POC?
A modern Salesforce integration should take 1 to 2 hours to connect, configure default mappings, and fire the first workflow. Quo integrated Salesforce in 1 hour (per Quo case study). Abacum was live with the first Play in under 2 hours (per Abacum case study). Anything above 1 day in 2026 is a red flag at SMB or mid-market scale.
What is bidirectional CRM sync and why does it matter?
Bidirectional sync means the GTM platform both reads from and writes to your CRM (Salesforce or HubSpot) on real objects: leads, contacts, accounts, opportunities, tasks, and email messages. It matters because one-way reads leave activity stranded in the GTM tool, and one-way writes corrupt CRM data. True bidirectional sync keeps the CRM as source of truth while giving reps a fast workspace to act in.
How does Unify's Salesforce integration work?
Unify's Salesforce integration runs read syncs approximately every 15 minutes and writes back to Salesforce in near-real-time when a Play or Sequence triggers. The platform creates Accounts, Contacts, and Leads (configurable), writes email messages and non-email tasks, and reads Opportunities for filtering and exclusions. Conservative overwrite logic only fills empty fields on existing records, except for namespaced Unify-specific fields. Lead assignment rules are honored on creation. Source: docs.unifygtm.com Salesforce bidirectional sync.
How does Unify's HubSpot integration work?
Unify's HubSpot integration is a native API connection (not Zapier-style middleware), launched October 2025. Read syncs run approximately every 15 minutes for contacts, companies, and deals. Unify creates and updates contacts and companies; deals stay HubSpot-owned and read-only in Unify. Source: Unify HubSpot integration launch blog (October 2025) and HubSpot bidirectional-sync docs.
What dedupe behavior should I demand in a CRM integration?
Demand explicit dedupe rules and a named match key (email for contacts, normalized website domain for companies). Demand that the platform never creates a second record when one exists. Demand that updates to existing records only fill empty fields by default, with explicit, namespaced exceptions for vendor-owned fields. Test with a 50-row CSV containing 3 deliberate duplicates and confirm the result.
Does Unify support custom Salesforce objects?
Custom Salesforce objects can be referenced for audience filtering and exclusion building once they exist in Salesforce. The Unify field-mapping documentation states custom fields must already exist in Salesforce before they can be mapped, which is the correct architectural pattern for production-grade integrations. Source: docs.unifygtm.com Salesforce field-mappings.
How do I know if attribution write-back is real?
Build a CRM-only report that filters opportunities by Sequence Enrolled, Play Enrolled, or equivalent campaign-membership field. If you can produce that report from CRM data without opening the GTM tool, attribution write-back is real. If not, the vendor is hiding pipeline behind a paywall or behind its own dashboard. Pipeline that does not survive a CRM-only report will not survive a finance review.
Glossary
- Bidirectional sync: A CRM integration pattern where the GTM platform both reads from and writes to the CRM on real objects (leads, contacts, accounts, opportunities, tasks, email messages) at near-real-time intervals.
- Field-level mapping ownership: Who controls which fields write where; in production-grade integrations, the buyer's admins control mappings, not the vendor's onboarding team.
- Dedupe / identity arbitration: The logic a GTM tool uses to determine whether an inbound record already exists in the CRM, typically using a deterministic match key like email or normalized website domain.
- Attribution write-back: Writing campaign membership, sequence enrollment, and pipeline-influence data back to the CRM as durable fields and activity records, not ephemeral GTM-tool-only events.
- Custom object support: The ability of a GTM integration to read non-standard Salesforce or HubSpot objects (e.g. Usage Event, Account Plan, Partner) for audience filtering and exclusion building.
- Sync latency: The time delay between an event happening in the CRM (or in the GTM tool) and it being reflected in the other system; production-grade is 15 minutes for reads, near-real-time for triggered writes.
- Sandbox parity: The principle that a vendor integration should connect to a Salesforce sandbox identically to production, so RevOps can test changes safely before deployment.
- Conservative overwrite rule: A write policy that only updates empty fields on existing records by default, preventing bulk imports from corrupting curated CRM data.
Sources
- Unify Salesforce integration documentation: bidirectional syncs, field mappings, lead assignment rules. docs.unifygtm.com.
- Unify HubSpot integration documentation: bidirectional syncs, field mappings. docs.unifygtm.com.
- Introducing Unify's HubSpot Integration, Unify Blog, October 2025.
- Abacum customer story, Unify.
- Quo customer story, Unify.
- Justworks customer story, Unify.
- Perplexity customer story, Unify.
- How Perplexity Booked $1.7M in Pipeline Without a Single BDR, Unify Blog, December 2025.
- Juicebox customer story, Unify.
- Spellbook customer story, Unify.
- Unify Raises $12M Series A to Scale Your Revenue Team's Creativity, Unify Blog, December 2025.
- Unify for RevOps solutions page.
- Unify Plays product page.
About the author. Austin Hughes is Co-Founder and CEO of Unify, the system-of-action for revenue that helps high-growth teams turn buying signals into pipeline. Before founding Unify, Austin led the growth team at Ramp, scaling it from 1 to 25+ people and building a product-led, experiment-driven GTM motion. Prior to Ramp, he worked at SoftBank Investment Advisers and Centerview Partners.
.avif)
