Major

Automate ZoomInfo-to-Outreach Lead Handoffs on Major

Move enriched ZoomInfo contacts into the right Outreach sequence without losing context. This brief maps the data flow, judgment points, duplicate checks, approvals, and controls an SDR team needs.

Ritvij GautamUpdated
Major and ZoomInfo lead-enrichment handoff for Outreach sequence automation

TL;DR: what you'll build. SDR managers own the handoff from ZoomInfo discovery to Outreach sequences, which often runs on CSV exports and copy-paste. This guide builds a ZoomInfo Outreach integration on Major that polls ZoomInfo on a schedule, checks eligibility and duplicates, asks an agent for one segment and sequence call, routes risky records to human review, and logs every Outreach write.

The SDR handoff that breaks between discovery and outreach

A rep's day starts in ZoomInfo, building a list of accounts that match an ideal customer profile (ICP). It should end with those contacts in an Outreach sequence, with context attached.

In most SDR orgs, someone exports a CSV from ZoomInfo, checks it against the CRM by eye, and manually creates prospects in Outreach before assigning a sequence.

Where the manual process fails

The SDR manager owns this process, and it breaks in predictable spots:

  • Duplicate prospects. Contacts get added twice because nobody checked for an existing prospect record.
  • Wrong sequence. A contact lands in the wrong sequence because nobody validated segment fit.
  • Stale enrichment. Data goes stale between export and import, so personalization fields are wrong by the time a rep opens the mailer.

None of this is a reasoning problem. It is a state and governance problem, and it belongs in a deterministic app rather than a spreadsheet.

Where this fits in the sales function

This sits inside outbound pipeline velocity, covered in Major's AI agents for sales hub. It is also a lead-generation problem at its root, covered in AI agents for lead generation.

What ZoomInfo and Outreach expose to a ZoomInfo Outreach integration

ZoomInfo knows who the contact is; Outreach knows what the rep is doing with them.

ZoomInfo (GTM.ai)

ZoomInfo holds company and contact discovery, firmographic and contact-level enrichment, and intent signals. The ZoomInfo API overview documents a Data API V1 with two groups of endpoints.

  • Search endpoints: Search Contacts, Search Companies and Search Intent, all POST endpoints.
  • Enrich endpoints: Enrich Contacts, Enrich Companies and Enrich Intent, which return full records once you have an identifier.

Auth is OAuth 2.0, either Authorization Code with PKCE for a user-facing flow or Client Credentials for a service integration. Both require a licensed ZoomInfo/GTM.ai account with API access provisioned through the Developer Portal.

Rate limits are tiered and published: Builder at 5 requests per second, Standard at 25 and Scaling at 35.

The ZoomInfo gotcha: no webhooks

Webhooks are not documented as a core feature, so you cannot subscribe to "contact enriched" events. The only reliable trigger is a scheduled poll against the search endpoints for records updated since the last run.

Outreach

Outreach holds the sequence engine and the prospect record reps work from. The Outreach API reference describes a REST API built on the JSON:API 1.0 specification, covering Prospect, Sequence, SequenceState, Mailing, Account, Opportunity, Task and custom objects.

Every call requires OAuth 2.0, with scopes limited per application and per token.

Outreach webhooks and rate limits

Unlike ZoomInfo, Outreach documents a real webhooks system. You register a webhook against a resource and action pair, for example prospect created or sequenceState advanced, and Outreach POSTs a JSON:API-styled payload to your URL.

Outreach does not publish specific rate limits in its public API overview; they are enforced per account, so check the tenant's actual plan before building a high-volume sync.

The manual path: the ReachOut Chrome extension

The ReachOut Chrome extension guide for using ZoomInfo with Outreach lets a rep inspect and export a ZoomInfo lead list by hand, getting a few contacts into Outreach for one rep on one day.

It gives an SDR manager no durable field mapping, dedupe rule, approval trail, or measurement of how many enriched contacts convert into sequence enrollments across a team. That is the gap a Major app closes.

Regulatory scope

Neither system is positioned as touching regulated personal data under GDPR, HIPAA or similar frameworks, beyond the general CAN-SPAM and TCPA obligations that govern any outbound sales tooling. The governance burden is mostly about not enrolling the wrong contact in the wrong sequence unnoticed.

Split the build: app, agent, workflow

Architecture diagram showing ZoomInfo and Outreach connectors feeding a Major app, an agent making sequence judgment calls, an approval decision gate, and an output step notifying the SDR

ZoomInfo to Outreach lead handoff architecture on Major

Handing every contact to an agent is expensive and the least auditable option, since decisions would live in model calls instead of a queryable table. Major splits the work by decision type.

The workflow: trigger and deterministic steps

The workflow owns the schedule trigger (ZoomInfo gives you no event to listen for), the ZoomInfo search and enrich calls, and the eligibility and dedupe checks against the app's tables. It calls the agent only for records that clear those gates, executes approved Outreach writes, and handles retries, backoff and error branching.

The agent: one kind of judgment call

Given an eligible, de-duplicated contact and its enrichment snapshot, the agent answers three questions:

  • Which approved segment does this contact belong to?
  • Which Outreach sequence fits that segment?
  • What does a defensible first-touch message look like, using only the fields the company has approved for personalization?

Segment fit is fuzzy at the edges, and a first-touch message needs real context, not a template fill.

The app: system of record and human interface

The app holds the enrichment snapshots, eligibility and dedupe decisions, the agent's recommendation and confidence score, approval state, SDR owner and final Outreach IDs. It is also where an SDR manager sees what is queued for review, what auto-enrolled and what failed.

Why state in the app saves tokens

On a rerun, the workflow sees the record already has a decision and skips straight to the write or the "already handled" branch. The model gets called once per contact, not once per sync.

Build the ZoomInfo Outreach integration on Major

Neither ZoomInfo nor Outreach appears in Major's existing connector catalog as a pre-built connector, so you configure both as custom API connectors.

  1. Connect both systems: for ZoomInfo, register a Client Credentials OAuth 2.0 app in the GTM.ai Developer Portal and store the client ID and secret in Major's credential proxy, scoped to the read-only search and enrich endpoints; for Outreach, register an OAuth application in an Outreach account's developer settings, store its credentials the same way, and scope it to grants like prospects.all, sequences.all, sequenceStates.all and accounts.read depending on what the tenant's admin approves, confirming exact scope names against the live Outreach OAuth guide because the public docs do not enumerate a fixed scope list. You should see both connectors listed in Major with stored credentials and a successful test call to each API.
  2. Define the data model: create a handoff_records table with one row per ZoomInfo contact considered, holding the fields listed below, plus a segment_rules table mapping each ICP segment to eligible Outreach sequence IDs, maintained by the SDR manager, not the agent. You should see two empty tables in the app, with the segment rules table editable by the SDR manager.
  3. Build the app by giving Major the app prompt shown after this list. You should see a queue view filtered to pending_review records and a history view filtered by outcome.
  4. Add the agent with the agent prompt shown after this list. You should see the agent return a JSON object with a segment, a recommended sequence ID, a rationale, a first-touch draft, a confidence score and an escalate flag when you run it on a sample contact.
  5. Set the workflow trigger as a scheduled poll, for example every four hours, against ZoomInfo's Search Contacts and Enrich Contacts endpoints, filtered to the ICP's target segments and to records enriched since the last successful run. You should see new rows appear in the handoff records table after each scheduled run.
  6. Wire the deterministic gates before the agent call, checking required-field completeness, suppression and consent status, and the dedupe key against the handoff records table and against Outreach through a Prospects lookup, sending only records that clear both checks to the agent. You should see duplicates and ineligible contacts marked in the app without any agent call attached to them.
  7. Add the human-in-the-loop step, so any record where the agent sets escalate to true, or where confidence falls under a threshold the SDR manager sets (for example 0.75), lands in the review queue instead of Outreach, where an SDR manager or the assigned rep approves it, edits the sequence choice or rejects it. You should see low-confidence and ambiguous records waiting in the review queue with the agent's rationale visible.
  8. Write and notify on approval, automatic or manual: the workflow calls Outreach's Prospects create-or-update endpoint, enrolls the prospect in the approved sequence, notifies the assigned SDR with the first-touch draft attached, and writes the resulting IDs and timestamp back into the handoff records table. You should see the Outreach prospect ID and timestamp filled in on the record and a notification in the SDR's inbox.
  9. Test end to end: force a rate-limit response, interrupt a batch partway through, replay the same contact, feed in a stale enrichment snapshot, send a contact that matches no segment and expire a token. You should see backoff and resume on rate limits, only unfinished records reprocessed, no double enrollment, a re-enrich before the agent step, an escalated review record and a loud auth error in the app's log.
  10. Roll out with SSO and role-based access so only SDR managers and assigned reps can act on review-queue records, only admins can edit the segment rules table, and every change to the mapping or agent prompt ships as a versioned deploy. You should see reps limited to their own queue items and an audit log entry for every approval and write.

The app prompt for step 3:

Build an app that tracks ZoomInfo contacts moving toward an Outreach sequence. Each record needs the enrichment snapshot, a dedupe key, a segment and sequence recommendation with confidence, an approval status, an assigned SDR, and the resulting Outreach prospect ID. Give the SDR manager a queue view filtered to pending_review and a history view filtered by outcome.

The agent prompt for step 4:

Given one contact's enrichment snapshot and the list of approved segment-to-sequence mappings, return a JSON object with segment, recommendedsequenceid, rationale, firsttouchdraft (using only the fields marked personalization-safe), confidence (0-1), and escalate (boolean, true if the contact does not clearly match an approved segment or if required fields are missing). Never invent a sequence ID that is not in the approved list, and never claim confidence above what the input data supports.

The handoff_records fields

  • zoominfocontactid and zoominfocompanyid → the ZoomInfo contact and company identifiers.
  • enrichment_snapshot and enriched_at → the enrichment data as JSON, and when it was captured.
  • icp_segment → the ICP segment assigned.
  • dedupe_key → typically a normalized email plus domain.
  • outreachprospectid → the Outreach prospect ID once written.
  • candidatesequenceid and agent_confidence → the agent's recommended sequence and confidence score.
  • approval_status → one of autoapproved, pendingreview or rejected.
  • reviewer and assigned_sdr → who reviewed the record, if anyone, and the rep who owns the contact.
  • run_status and last_error → where the record is in the workflow, and the most recent error, if any.

Governance and IT control

This workflow touches two licensed, paid platforms and writes directly into a rep's working queue, so the control surface matters as much as the logic.

Access and credentials

  • SSO and role-based access on the Major app gate who can act on the review queue and edit the segment rules.
  • Scoped credentials through Major's credential proxy keep the ZoomInfo and Outreach OAuth tokens out of any agent prompt or workflow code. The agent only sees the enrichment snapshot and approved segment list, never a raw API key.

Records, change control and approvals

  • Audit logs on every write record who or what approved a sequence enrollment, when, and with what confidence score.
  • Managed database and storage keep snapshots and decision history queryable past any run, without re-pulling from either API.
  • Versioned deploys make mapping or prompt changes reviewable updates, not silent edits to a spreadsheet formula.
  • Approval required below the confidence threshold and for ambiguous records keeps a person in the loop on risky outbound decisions.

Failure handling and edge cases

Diagram of the seven-step data flow from scheduled ZoomInfo discovery through enrichment, eligibility, dedupe, agent recommendation, approval, and logged Outreach outcome

Seven-step data flow from enrichment to logged outcome

Rate limits and partial syncs

  • 429 and rate-limit responses. ZoomInfo's tiered limits (5, 25 or 35 requests per second by plan) can be hit quickly on a large target-account list. Back off with increasing delay and resume from the last successfully processed page rather than restarting the batch.
  • Partial syncs. If a write to Outreach fails partway, the per-record run status lets the next scheduled run pick up only unfinished records.

Duplicates and stale data

  • Duplicate events. Outreach's webhook docs note that create and update payloads only include changed attributes, and any webhook system can deliver the same event more than once. Matching on the ZoomInfo contact ID plus the dedupe key before writing prevents double enrollment.
  • Stale enrichment. If the enriched-at timestamp is older than a set threshold (for example 30 days) at the agent step, re-enrich first so the agent does not draft from outdated firmographic data.

Missing segments and expired tokens

  • No eligible sequence. If a contact matches no segment in the segment rules table, the agent escalates and the record sits in review rather than defaulting to a generic sequence.
  • OAuth expiry. Both ZoomInfo and Outreach tokens expire. Fail loudly into the app's error log, since a silent skip looks identical to "nothing new to process."

What the agent must never do

  • Write directly to Outreach.
  • Invent a sequence ID outside the approved list.
  • Bypass the eligibility or dedupe checks that run in the deterministic layer before it is called.

Those stay as code so they behave the same way on contact number one and contact number ten thousand.

What the SDR manager should measure

  • Eligible-to-enrolled rate: the share of eligible, de-duplicated contacts enrolled in a sequence, showing how much enrichment converts into pipeline.
  • Duplicate and rejected count: records the dedupe or eligibility checks caught before reaching a rep, a measure of upstream list hygiene.
  • Time from enrichment to disposition: the lag between enrichment in ZoomInfo and enrollment or rejection. A long lag means stale personalization data.
  • Approval rate and override rate: how often SDRs accept the agent's recommendation versus overriding it, showing whether segment rules need tightening.
  • Exceptions by cause: low confidence, missing fields, a suppression hit or no segment match, pointing at where to fix upstream data or segment rules.

Set your own baseline

There is no universal benchmark; measure your own baseline in the first few weeks and set thresholds from there.

Adjacent builds

For the same split on the CRM side, see sales workflows that start in Salesforce and sales automation across Salesforce.

Before you go live

  • [ ] ZoomInfo and Outreach credentials are in Major's credential proxy, with Outreach scope names confirmed against the live OAuth guide.
  • [ ] The SDR manager has filled the segment rules table with only approved sequence IDs per ICP segment.
  • [ ] The poll schedule and Search Contacts filters match your ICP segments and ZoomInfo tier's rate limit.
  • [ ] Required-field, suppression and consent, and dedupe checks, including the Outreach Prospects lookup, run before any agent call.
  • [ ] The review confidence threshold (for example 0.75) is set, and escalated records land in the review queue.
  • [ ] A stale-enrichment threshold (for example 30 days) triggers a re-enrich before the agent step.
  • [ ] Auth failures write to the app's error log instead of silently skipping a run.
  • [ ] SSO, role-based access and audit logs are on, and only admins can edit the segment rules.

Start building on Major

Build a stateful ZoomInfo-to-Outreach handoff app that checks fit, stages the prospect, and keeps sequence decisions and exceptions reviewable. Let an agent handle the judgment call while deterministic steps run the repeatable work.

Related articles

About the author

Ritvij Gautam is Head of Growth at Major.

Last reviewed November 6, 2026 by Ritvij Gautam.

Frequently asked questions

How do you connect ZoomInfo to Outreach?
There is no native one-click connector. Outreach's support docs describe a manual path using the ReachOut Chrome extension to inspect and export leads, while a governed integration calls the ZoomInfo API (OAuth 2.0, licensed account required) to enrich contacts and the Outreach API (OAuth 2.0, JSON:API) to create prospects and enroll sequences. Exact scopes and access depend on each account's entitlements, so verify against current docs before building.
Can ZoomInfo automatically add contacts to an Outreach sequence?
Not by itself. ZoomInfo has no documented mechanism to push into Outreach directly. An API-driven workflow can poll ZoomInfo for enriched contacts, check eligibility and duplicates, then call the Outreach API to create a prospect and enroll it in a sequence, provided the tenant has the right API access and approved automation rules. Don't assume this is default behavior in either platform.
How should a team prevent duplicate Outreach prospects?
Match incoming ZoomInfo contacts against existing Outreach prospects using a stable identifier, typically a normalized email plus domain, before creating a new record. Check this dedupe key against your own state store and against Outreach's `Prospects` resource directly, log the outcome, and update existing records instead of creating new ones. No single key works universally, so validate the matching rule against your own data quality.
Does ZoomInfo provide webhooks for newly enriched contacts?
The saved ZoomInfo API documentation does not list webhooks as a core feature, so there's no documented way to subscribe to a "contact enriched" event. The reliable approach is a scheduled poll against ZoomInfo's search and enrich endpoints for records updated since the last run. Confirm with your account team whether a tenant-specific event capability exists before assuming otherwise.
What should trigger human review instead of automatic sequence enrollment?
Route a contact to a person when the agent's confidence in the segment or sequence match falls below your threshold, when required fields are missing, when the contact doesn't clearly match any approved segment, or when a suppression or consent flag is present. Auto-enrollment should never be the default for ambiguous matches; treat it as the exception path for clearly eligible, de-duplicated contacts.
Is this workflow subject to CAN-SPAM or TCPA rules?
Any automated outbound email or call sequence it feeds into is still subject to the FTC's CAN-SPAM Rule and, for automated calls or texts, the FCC's TCPA, including opt-out handling and consent records. The integration itself doesn't need special compliance logic, but the Outreach sequences it enrolls contacts into must follow your organization's existing compliance process and counsel guidance.