From one executive to the whole buying group: Demandbase MCP + LLMs

What you’ll learn

By the end of this playbook, you will know how to:

• Connect Demandbase MCP to an LLM and verify the connection.
• Build a research workflow that combines account intelligence, buying-group signals, and web research.
• Move from one executive contact to a coordinated buying-group motion.
• Tailor outreach for Above the Line (ATL) and Below the Line (BTL) stakeholders.
• Test, govern, and measure an AI-assisted prospecting workflow.

Why this matters

B2B buying decisions rarely belong to one person. A typical decision includes 13 internal stakeholders and nine external influencers, according to Forrester’s The State of Business Buying, 2026. Yet outbound still reaches one or two people: the person who filled out a form and perhaps their manager.

The result is single-threaded pipeline that can stall when one contact goes quiet. The answer is not more volume from the same email. It is relevant outreach to the executives who own the budget and the practitioners who validate the decision.

The best prospectors already know the motion: find an account with strong engagement or a high predictive score, identify signals, research the company’s priorities, compare executive language over time, identify the metric under pressure, match a relevant proof point, and write a message that could only have gone to that company. Demandbase MCP can make that discipline repeatable.

What you need to get started

Before you begin, assemble:

• Access to Demandbase MCP and a supported LLM. The documented setup and screenshots in this playbook use Claude Team or Enterprise.
• A new-business account with at least one intent signal, news event, or identifiable buying-group contact.
• The Research Assistant instructions and approved enablement files.
• Executive metrics, customer stories, persona guidance, and proof points.
• Three to five accounts whose real situation and buying group you already understand.
• A measurement baseline for reply rate, meetings, buying-group engagement, and pipeline.

Part 1: From one executive to the whole buying group

A single executive email can open the door. The rest of the buying group is what keeps a deal from stalling. The goal is to create one coherent company point of view, then adapt the framing for each stakeholder.

Step 1. Start with the company initiative

Research company context before writing. Pull the account’s intent, buying group, engagement, news, and relevant financial information from Demandbase. Then use web research to understand executive commentary, earnings calls, filings, interviews, podcasts, partnerships, launches, and hiring trends.

Use these five research prompts adapted from DemandJen to connect research to messaging:

• What are the company’s current growth objectives?
• What is difficult about the contact’s role right now?
• How does the contact appear to address the problem today?
• What is the cost of leaving the problem unchanged?
• How could Demandbase address that cost of inaction?

Treat every answer as a hypothesis to validate. If the research does not support a conclusion, report the gap instead of filling it with an assumption.

Step 2. Identify the metric under pressure

Map the company’s situation to the metric that matters most.  For Demandbase they are:

• Pipeline and revenue: ARR growth, conversion rate, or average selling price.
• Sales efficiency: CAC payback, quota attainment, or rep productivity.
• Marketing efficiency: LTV:CAC, wasted ad spend, or marketing-sourced pipeline.
• Retention and expansion: NRR, churn signals, or competitive intent.
• Strategic planning: Rule of 40, EBITDA margin, or forecast accuracy.

Match each proof point to the mechanism it proves. A buying-group handoff should use win-rate or pipeline evidence, while an ad-efficiency problem should use an ad-waste proof point.

Step 3. Map the full buying group

Start with the people already showing meaningful engagement. Then identify adjacent contacts the champion will need to convince. Label each person ATL or BTL so the rep knows how the message should change.

The company story should stay consistent across the group:

• Executive or ATL: company initiative, executive commentary, the metric under pressure, and a proof point in boardroom language.
• Engaged practitioner or BTL: engagement minutes, intent topics, day-to-day workflow, and a practical “how do I?” angle.
• Adjacent stakeholder: the same company initiative reframed for that person’s specific metric.
• Recently hired contact: tenure under six months and an assessing-mode angle rather than an urgent fix-it pitch.

ATL creates energy and budget. BTL creates technical validation. Winning the deal takes both.

Part 2: Build the AI research workflow

The advantage is not access to AI. It is what the assistant has to work with and what you tell it to care about. A well-built Project gives every rep the same research discipline as a top performer, without asking them to start from a blank page.

Step 1. Connect Demandbase MCP

The setup documented here is Claude-specific. Open Customize, select Connectors, select the plus button, and choose Browse connectors. Search for Demandbase, select Connect, and complete your Demandbase sign-in. On Claude Team or Enterprise, an organization administrator may need to make the connector available first, and Demandbase MCP must be enabled for the organization.

Step 2. Verify the connection

Start a new chat, open the plus menu in the composer, select Connectors, and turn on Demandbase. Send a read-only test prompt such as: “List the top 10 VP contacts at [account].”

For an isolated connection test, temporarily turn off Web search if Demandbase recommends it. Turn Web search back on for the Project because executive commentary, earnings calls, and filings live on the open web.

Step 3. Create the Project instructions

Create a Project with a clear role, process, messaging rules, and output format. Keep these non-negotiables explicit:

• New-business outreach only. If the account is an existing customer, stop and route to the appropriate variant.
• Research qualification: require at least one intent signal, news event, or buying-group contact.
• Research budget: one account brief, up to three web searches, and one buying-group lookup before drafting.
• Research sequence: Demandbase MCP data, web research, then the five DemandJen prompts.
• Messaging sequence: company initiative first, persona second, product third.
• ATL and BTL contacts receive different messages.
• Inferences are labeled as hypotheses, research gaps are reported, and facts or quotes are never fabricated.
• Emails follow the Problem Prompter structure, stay under 75 words when possible, and avoid meeting asks, em dashes, and AI writing tells.

The required output should include a research summary, recommended contacts labeled ATL or BTL, two or three email variants, five subject lines per email, and talking points.

Step 4. Load the knowledge files

The instructions tell Claude how to think. The knowledge files give it something to think with. Load the approved Executive Metrics & Customer Stories, Marketing, Sales & Data Personas Cheat Sheet, Selling Above and Below the Line workshop, Elite SDR playbook, Case Study Library, and Prompts by Persona guide.

If you are looking for the specifics, here is a link to my instructions for an executive messaging plug-in.

Part 3: Test and operationalize the workflow

Do not move directly from setup to rep-wide adoption. Test the workflow against accounts whose real story you know.

Step 1. Test known accounts

Run three to five accounts and compare the output with what the account team knows. Check whether the assistant:

• Finds the correct company pressure.
• Identifies the expected buying-group members.
• Distinguishes ATL from BTL framing.
• Matches the proof point to the mechanism.
• Separates verified findings from hypotheses.
• Reports missing information rather than inventing it.

If the assistant misses an obvious stakeholder or creates a fabricated quote, tighten the instructions before any rep uses the Project.

Step 2. Build focused variants

Do not force every motion into one giant prompt. Keep a base Project, then create focused variants for new business, advertising buyers, stalled opportunities, and other materially different use cases. The base instructions should stop and route an existing customer to the appropriate variant.

Step 3. Pilot with ownership

Choose one segment, name the owner, and compare the new motion with a segment still prospecting the old way. Track reply rate, meetings, buying-group members engaged per account, opportunity creation, and pipeline. Review the results with Sales, Marketing, and RevOps before expanding.

Part 4: What this looks like in practice

A rep adds a Tier 1 new-business account and a VP of Marketing contact to the Project. The assistant combines account intent, buying-group engagement, recent news, and public executive commentary. It identifies a likely shift from growth language toward efficiency language, maps that hypothesis to the relevant executive metric, and recommends a proof point that supports the mechanism.

The assistant then widens the lens. It recommends a BTL thread for the most engaged practitioner, identifies an adjacent RevOps stakeholder, and changes the framing based on that person’s role and tenure. The output is a research summary, recommended contacts, role-specific email variants, subject lines, and talking points.

This example is illustrative. Replace it with a real, dated Project run before publication, or label each account-specific detail as hypothetical. In particular, confirm the account facts, the leadership-change signal, the CEO language shift, and the Unanet example before treating them as verified.

The results

Modern prospecting should not require every rep to recreate a top performer’s research process from scratch. With Demandbase MCP connected to a well-built Project, the rep’s job shifts from assembling research to judging it: which angle is strongest, which gap matters, and whether the message is worth sending.

The operating model is simple:

1. Research the company initiative.
2. Identify the metric under pressure.
3. Map the full buying group.
4. Adapt the message by ATL or BTL role.
5. Validate the research before sending.
6. Measure whether engagement expands across the group.

The goal is not to send more AI-written emails. It is to reach the executive with something worth reading, bring the rest of the buying group into the same company story, and give the account team a clearer basis for action.

Sources and open items

• Forrester, The State of Business Buying, 2026 press release: 13 internal stakeholders and nine external influencers per typical buying decision.
• Demandbase Developer Guide: Claude, current connector setup steps, updated September 2026.
• Labs by Demandbase, Introducing the B2B GTM Benchmarks Report