B2B buyers cite inconsistent information across teams as the top reason for switching suppliers, according to McKinsey's 2026 B2B Pulse survey. That single data point anchors a new Harvard Business Review analysis arguing that agentic AI is forcing — and enabling — marketing and sales to operate as one function rather than two adjacent departments.

The authors, drawing on more than 100 years of combined experience modernizing sales and marketing organizations, contend that most companies are still deploying AI function by function: embedded separately in marketing and in sales, rarely across the system that connects them. The customer, they note, does not draw that distinction. Every touchpoint registers as part of a single relationship, regardless of internal ownership.

The behavioral backdrop makes fragmentation costly. McKinsey research cited in the piece finds that more than 80% of consumers use multiple channels to research or purchase products, and B2B buyers now engage across ten interaction channels during a typical journey — double the number from less than a decade ago. Buying journeys loop backward and pause; they no longer run linearly. When marketing and sales work from separate data sets, a personalized campaign can be followed by an AI-driven sales touch built on different data, different assumptions, and different timing. Each function may hit its own metrics while jointly delivering a fragmented experience.

The differentiation question has shifted. Omnichannel execution is now table stakes; integration is the advantage. McKinsey's research finds market leaders are four times more likely to deploy true one-to-one personalization and are further ahead embedding AI into integrated commercial workflows. McKinsey's The State of AI report places the most consistent revenue gains from AI in marketing and sales specifically: among organizations using AI in these functions, 7% report revenue increases exceeding 10%.

What the measured results look like

Two company examples show the workflow-level impact, and both come from the authors' own consulting work rather than independent measurement — a distinction worth holding onto when evaluating the claims.

A Fortune 500 tech company wanted to expand into a largely untouched segment of tens of thousands of commercial customers with a small, capacity-limited sales team. Diagnostics found sellers spending too much time on low-probability leads. An end-to-end AI solution enriched leads with signals such as M&A activity, prioritized by conversion likelihood, surfaced account intelligence, and ran agentic outreach. Within eight weeks, AI-augmented sellers were sending five times the number of lead messages while holding pre-AI open rates, response rates, and meeting scheduling rates. Prep time for initial calls fell from 1–2 hours to 10–15 minutes.

A global job search platform offers the sales-side case. The company expects its AI-enabled sales model to generate $30–60 million in incremental annual revenue, part of a projected $400–700 million uplift from agent-enabled use cases. The core system is an AI sales development representative that identifies prospects, generates personalized outreach, nurtures early conversations, and hands qualified leads to human sellers with full context and next actions. The authors attribute the result not to the agent alone but to shared data and coordinated handoffs — a claim about architecture, not just automation.

Roles and metrics are reorganizing

These changes are reshaping team structure. The piece identifies an emerging role, the go-to-market engineer, responsible for designing and managing agentic workflows that span functions and ensure continuity across interactions. Accountability is shifting too: marketing and sales increasingly share ownership of pipeline quality, conversion, and customer value, and the traditional MQL-versus-SQL distinction gives way to measures spanning the full journey, including customer lifetime value.

The risk concentrates where AI agents face customers directly. When marketing, sales, and service each deploy customer-facing agents independently, each can optimize locally while fragmenting the broader customer narrative. Without shared standards and clear ownership, the authors write, AI scales inconsistency as easily as it scales efficiency.

What leaders should audit now

The authors close with a pressure-test for commercial leaders across three areas. First: is the AI transformation a shared effort or two parallel programs that coexist, measured against a common definition of growth? Second: do marketing and sales operate from a shared, real-time customer view with full context of every prior interaction, or from parallel data sets reflecting internal boundaries? Third: is the organization building the skills the agentic model requires — marketers directing AI systems that extend into sales territory, sellers engaging customers already shaped by AI-driven interactions?

The authors' framing is direct: the technology to build a closed-loop go-to-market system is no longer the constraint. The advantage will belong to organizations whose leaders redesign the work itself — where signals emerge, where decisions occur, where coordination breaks down — rather than layering AI onto existing functional silos. The open question for most companies is whether they are building toward that model now or waiting until fragmentation becomes a measurable competitive disadvantage.