Reference customer match: happy customers in the prospect's industry
HubSpot + G2 + Intercom
Who it's for
AEs who need a reference customer that looks like their prospect, and want a shortlist backed by G2 reviews and support history before they ask the account owner. HubSpot can be swapped for Salesforce, Attio or Pipedrive. Intercom EU works the same way for EU-hosted workspaces.
What you get
Reference shortlist for a logistics prospect. Sample output: every company and person in it is fictional.
Prospect: Tidewell Robotics (fictional), industry Logistics, about 400 employees. HubSpot customers in Logistics: 7. With a G2 review we could match by company: 3. Checked in Intercom: all 7. | Customer | Owner | Customer since | G2 | Intercom, last 6 months | Fit | |---|---|---|---|---|---| | Larkspur Freightworks (fictional) | Priya Raman | Mar 2024 | 5 stars, Jun 2026: "Cut our month-end close from 6 days to 2." | 4 conversations, all closed, no complaints | Strong | | Mossgate Labs (fictional) | Sam Whitlock | Nov 2024 | 4 stars, Feb 2026: "Setup took longer than planned, but support was quick." | 11 conversations, 1 open billing issue since Sep 28 | Ask owner first | | Brindle Harbor Co. (fictional) | Dana Okafor | Jan 2025 | No review found | 2 conversations, both closed | Possible, no public proof | Left out: Quarry Lane Analytics (fictional). Two escalations about data sync in August and an open renewal deal in HubSpot. Not matched: 2 G2 reviews from Logistics reviewers with no company name. Not counted. Next step for you (text only): "Hi Priya, I'm working a logistics deal with Tidewell Robotics. Would Larkspur Freightworks be open to a 20-minute reference call in the next few weeks? Happy to work around their schedule." No one was contacted and nothing was changed in HubSpot, G2 or Intercom.
Time and cost
Estimate About 15–25 minutes, most of it reading the support history.
Assumptions: One prospect, with about 5–10 customers in the same industry checked in each tool.
- HubSpot: Reads only: one company search by industry, plus deals and owner for each match. Credit use not confirmed by HubSpot.
- G2: Reads only: your product's reviews, filtered to the matching companies. Credit use not confirmed by G2.
- Intercom: Reads only: one company lookup and its recent conversations for each match. Credit use not confirmed by Intercom.
Exact credit costs depend on your plan with each vendor. Check the tool pages before a large run.
MCP servers
Every server is confirmed in: Claude
-
ClaudeChatGPTRead + writeCredits not confirmedNeeds admin or IT
https://mcp.hubspot.com -
ClaudeChatGPTRead + writeCredits not confirmed
https://mcp.g2.com/mcp -
ClaudeClaude CodeCursorRead + writeCredits not confirmed
https://mcp.intercom.com/mcp
Set up in a chat app
Every server in this recipe is confirmed for Claude. Connect each one, then paste the prompts below into one conversation.
- Connect HubSpot without code (Claude, ChatGPT)
- Connect G2 without code (Claude, ChatGPT)
- Connect Intercom without code (Claude)
Developer setup
Auth setup and mcp.json for Claude Code and Cursor
- HubSpot: Sign in with OAuth in your client (run
/mcpin Claude Code). Do it once interactively, and the client stores and refreshes the token. - G2: Sign in with OAuth in your client (run
/mcpin Claude Code). Do it once interactively, and the client stores and refreshes the token. - Intercom: Sign in with OAuth, or pass an API key as a Bearer token for headless use.
Put this in .mcp.json (Claude Code, project root) or ~/.cursor/mcp.json (Cursor). Swap the example keys for your own, and don't commit real ones. Remove the // comment lines if your client requires strict JSON.
{
"mcpServers": {
"hubspot": {
"url": "https://mcp.hubspot.com"
},
"g2": {
"url": "https://mcp.g2.com/mcp"
},
"intercom": {
"url": "https://mcp.intercom.com/mcp"
// headless alternative: "headers": { "Authorization": "Bearer YOUR_INTERCOM_API_KEY" }
}
}
}
Step-by-step prompts
- Find {prospect} in HubSpot and tell me its industry and size. Then find our current customers in the same industry: company, owner, customer since, deal amount, and any open deals. Show me a table. Don't change any records.
- Find G2 reviews of {our_product}. Match them to the customers in the table by company name only, and quote the matching reviews with the date, and the star rating if G2 shows one. List reviews from this industry that don't name a company separately, and don't guess who wrote them.
- For each customer in the table, look up the company in Intercom and summarize its conversations from the last {months} months: how many, how many are still open, and any complaint or escalation, with one short quote. Don't add notes or change anything in Intercom.
- Rank the customers as Strong, Ask owner first, or Leave out, with one line of evidence each from HubSpot, G2 and Intercom. Then write a short note I can send to the account owner of the top pick, asking if their customer would be open to a reference call. Text only. Don't contact anyone.
Run it on a schedule (optional)
- Run it by hand when a deal reaches the stage where the buyer asks for a reference. A shortlist from last quarter may be out of date.
- If you run it on a schedule, for example monthly for your top industries, keep it read-only and treat it as a starting list. Check support history again before each ask.
- Non-admin HubSpot users need a Super Admin to approve the connector first. Do one manual run before scheduling anything.
Where this goes wrong
- An anonymous review matched to the wrong customer. Many reviews don't name the company. Count a G2 review only when its company name matches a HubSpot customer, and list unmatched reviews on their own.
- A quiet support history read as a happy customer. Few Intercom conversations can mean low usage, not satisfaction. Show the conversation count and the date of the last one, and mark customers with no recent contact as 'unknown', not 'happy'.
- Suggesting a customer in a sensitive moment. A customer with an open renewal, an escalation or a churn risk is a bad reference. Check HubSpot for open deals and Intercom for open or escalated conversations, and move those to 'Ask owner first' or leave them out.
- Industry labels that don't match. HubSpot's industry field may be empty or set differently from the prospect's. Show the industry value used for each match, and list customers with an empty industry field so you can check them by hand.
Gotchas
- Intercom's listed tools cover conversations, contacts and companies. CSAT or conversation ratings aren't among them, so support satisfaction here comes from reading conversations, not from a score.
- Intercom's directory connector in Claude covers US-hosted workspaces. EU-hosted workspaces use the Intercom EU server as a custom connector, AU-hosted workspaces aren't supported, and on Claude Team or Enterprise an org Owner adds the connector. ChatGPT support isn't confirmed.
- G2's MCP server needs a G2 subscription with the relevant entitlements, and Market Intelligence review data needs an Enterprise package. Sign in to Claude with the email used for your G2 account.
- HubSpot needs admin setup: non-admin users need a Super Admin to approve the connector, and admins can limit what it can access. If the customer list looks short, check the connected account's access.
- Using Salesforce instead? Customers by industry come from a SOQL query, and each tool call counts against your org's daily API quota. Salesforce needs Enterprise Edition or above and an admin-created External Client App, and ChatGPT needs developer mode. Attio and Pipedrive connect without an admin step.
Download
Get the prompts, guardrails and expected output as one Markdown file. Paste it into a Claude Project or ChatGPT project's instructions, or save it as CLAUDE.md for Claude Code.
New recipe and breakage notes every week
Vendor sites: HubSpot. If we join an affiliate program, these links will be labeled as affiliate links.