Outreach

What it is

Outreach is the enterprise sales engagement platform: sequences, an integrated dialer (Outreach Voice), conversation intelligence under the Kaia name, and deal and pipeline management over Salesforce or HubSpot. It relaunched on 27 April 2026 as an “agentic AI platform for revenue teams” and moved from outreach.io to outreach.ai. The agent layer is now the product surface: Omni, a conversational agent in Slack and mobile; Agent Studio, where RevOps builds and schedules custom agents; and named agents for research, personalization, meeting prep, deals, and revenue follow-up. It remains the default engagement layer for $50M+ ARR sales orgs running structured outbound across many reps, and its closest peer is Salesloft.

Why it shows up in RevOps stacks

• Scale and control. Hundreds of seats, granular permissions, deep Salesforce coupling, audit trails, and an admin model that survives reorgs. Vendr has recorded 914 Outreach purchases; this is the incumbent most enterprise outbound teams already run.
• Engagement data an agent can reach. Outreach’s MCP server went generally available on 24 February 2026, and the 2 June 2026 release added an MCP client, making Outreach the first revenue platform to ship both. External agents in Claude, ChatGPT, Microsoft Copilot, Salesforce Agentforce, and Codex can search prospects, sequences, opportunities, and Kaia meetings, and can write as well as read — create prospects and accounts, add prospects to sequences. In Claude it installs from the connector directory with an Outreach sign-in. The client runs the other direction, letting Outreach agents pull from Glean, Google Drive, SharePoint, Snowflake, ZoomInfo, and Seismic. Every Amplify tier lists “Outreach MCP (Server & Client)”, so agent access is no longer an upsell.
• The agents are being used. On 10 September 2026 Outreach reported 12x growth in AI-credit consumption and 480% year-over-year AI ARR growth in its fiscal Q2. Those are vendor figures and they measure usage, not outcomes — but they are the opposite of shelfware.

Pricing

• Four Amplify tiers, no published rates. outreach.ai/pricing lists Essentials (10,000 AI credits), Core (25,000), Plus (50,000, marked most popular), and Pro (100,000), under the line “Per user pricing. No platform fees. Support included.” The old Engage-plus-modules menu (Engage, Meet, Deal, Forecast) is gone as the front-door offer.
• Essentials is not the cheap sequencer. It carries conversation intelligence, deal management, and the Omni, Research, Deal, and Meeting Prep agents, but no sequences. Sequencing starts at Core. Plus combines Core’s engagement with Essentials’ conversation intelligence and deal management; Pro adds forecasting (Scenario Planner, rollups, territory management) and managed services. API allowances are 250,000 calls on Essentials and Core, 500,000 on Plus, and 1,000,000 on Pro.
• What teams pay. Third-party pricing research puts Core at ~$100-120 per user/month, Plus at ~$130, and Pro at $160+, on annual contracts only; Essentials is reported below Core, inconsistently. Vendr’s median annual contract is $45,600 across 914 purchases, in an $8,560-$213,832 range, at 12.44% average savings. Negotiated discounts of 15-35% are reported, with volume breaks at 20-25, 50, 100, and 250 seats.
• The 50-seat math. 50 seats on Plus at ~$130 is ~$78,000/year in subscription, plus $5,000-25,000+ of implementation in year one. Budget a credit buffer on top: Outreach sells additional credit packs but does not publish per-action credit costs or pack prices.

Best for

• RevOps leaders at $50M+ ARR B2B companies with 50+ outbound reps, where sequencing, calling, and pipeline inspection have to run in one permissioned system
• Teams that want their own agents in Claude or ChatGPT to act on engagement data without building an integration — create prospects, enroll them in sequences, pull meeting history
• Not for teams under ~20 reps: volume discounts don’t start below 20-25 seats, and Apollo covers the same motion at published prices

Alternatives — and when to pick them instead

• Salesloft — the peer incumbent, now part of Clari + Salesloft. Pick it if forecasting sits at the center of the decision, since Clari’s forecast layer is native there, or if you run 10-40 reps and want a vendor with a lower seat floor.
• Apollo — pick it under ~$25M ARR. It bundles a B2B contact database with sequencing at published prices (Basic $49 per user/month on annual terms), and its MCP connector works on any plan, including free.
• Gong — the comparison for anyone quoted Essentials. That tier is conversation intelligence plus deal management with no sequencing, which is Gong’s job. Pick Gong if calls and deal inspection are the purchase and engagement already lives elsewhere.
• 11x or Artisan — the fastest-growing challengers, selling autonomous AI SDRs that run outbound themselves. Different job: pick one only if you want fewer reps rather than better-equipped ones.

Watch-outs

• Credits are the line that moves. Seat count is fixed at signing; credit consumption is not, and Outreach does not publish the per-action rate. Guard: before signing, get the credit cost of the three agent actions you will run most written into the order form, and set a usage alert at 70% of the annual allotment.
• Third-party quotes still carry the old fees. Several 2026 pricing guides still list a $2,000-5,000/year platform fee and $30-50/seat module add-ons. The vendor page now says no platform fees and folds conversation intelligence, deal, and forecast into the tiers. Guard: if a quote contains a platform-fee line or a separately priced Meet or Deal module, ask which tier it maps to and have it removed.
• Choosing the tier by credit count. Essentials looks like the entry point and has no sequences; buying it for an outbound team means an upgrade before the first campaign. Guard: map the features you need to tiers on the pricing page first, then compare credit counts.
• Agent writes are real and irreversible. Outreach’s own MCP usage guide says the server “does not throttle or undo”, and its prospect tools include delete. Independent 2026 testing found create and delete exposed but not field-level edits, so an agent told to fix a record can delete and re-create it, losing its sequence history. Guard: switch off the delete tools in the connector configuration for everyone outside RevOps admins, and test the exact edit you need in a sandbox org before relying on it.
• Migration lock-in. Annual contracts, auto-renewal with a 90-day notice window, and sequences, reporting, and integrations that all need rebuilding. Guard: diary the notice date at signing and negotiate seat flexibility into the contract, not at renewal.