ChatGPT Citation to HubSpot Deal

This page walks one example end to end. A real-shape example: a mid-market B2B SaaS buyer asks ChatGPT a category-definition question, sees a citation to your brand, eventually converts, and closes as a $48K ARR deal four months later. The playbook traces every step through HubSpot and shows what your CFO should see on the deal record at the end.

Assumes the custom property schema from AI Search Attribution in HubSpot is already in place. If not, start there.

The scenario

Buyer: Priya, VP Marketing at a 180-person B2B SaaS (fictional composite for illustration).

Prompt: "What are the best tools for tracking AI search visibility and attribution?"

First exposure: ChatGPT answer names four tools; your brand is one of them, with a citation link to your comparison page.

Path to closed-won: 4 weeks of quiet research, a demo request, a 60-day sales cycle, $48K ARR closed.

Step 1Citation serves in ChatGPT

Your AEO monitoring system samples the prompt "best tools for AI search visibility and attribution" twice per day across ChatGPT, Perplexity, Claude, and Gemini. On day zero, ChatGPT returns an answer that names your brand in position two, with a citation link to runlantern.com/compare/best-aeo-tool-with-hubspot-integration/.

Logged to the citation database with: prompt, engine (ChatGPT), position (2), citation URL, surrounding context (the full answer text), and timestamp.

Step 2Buyer reads the answer, copies the URL

Priya reads the four-tool list. Your comparison page URL is in her clipboard. She pastes it into a new tab. Session lands on your site with:

• Referrer header: (empty)
• Landing page: /compare/best-aeo-tool-with-hubspot-integration/
• GA4 source/medium: (direct)/(none)
• HubSpot tracking cookie: set at session start, hubspotutk

This is the exact moment analytics tools call "direct means we don't know." No referrer, no UTM, just a Direct session — indistinguishable from a URL typed from memory. The signal that saves this session is what happens next.

Step 3Priya reads, leaves, researches, comes back

She reads the comparison page. Bounces. Comes back a week later after seeing your brand mentioned in a Pavilion Slack thread. Reads the pricing page. Bounces again.

Three weeks later she returns from a branded Google search ("lantern aeo") and lands on the homepage. Clicks demo. Fills out the form.

Key HubSpot session log at this point:

• Session 1 (day 0): Direct, comparison page. Cookie set.
• Session 2 (day 7): Direct, pricing page. Cookie persists.
• Session 3 (day 28): Organic Search (branded "lantern aeo"), homepage, demo form submit.

Step 4Form submission — the attribution-saving moment

On the demo form, the required field is email. The optional free-text field, placed as the last field on the form:

This is the attribution-saving moment. Without this field, the contact record would show Original Source = Organic Search, Original Source Drill-Down 1 = google, Original Source Drill-Down 2 = branded. The ChatGPT touchpoint would be invisible.

With this field, the contact record gets aeo_self_reported_source = "ChatGPT a few weeks ago when I was researching AEO tools." A HubSpot workflow parses the text and sets aeo_first_touch_engine = chatgpt and aeo_first_touch_timestamp = current form submission time.

Step 5Prompt-level enrichment (optional, via Lantern)

If you're on Lantern, the integration does a probabilistic match: the contact's first Direct session (day 0, comparison page) lands within the 24-hour window of a known ChatGPT citation serve for the prompt "best tools for AI search visibility and attribution" on that exact URL. Lantern writes:

• aeo_first_touch_prompt = "best tools for AI search visibility and attribution"
• aeo_touchpoint_count = 1

Without Lantern, prompt-level attribution is lost — you know ChatGPT was the engine, not which prompt. This is the reason prompt-level monitoring matters: it enables prompt-level ROI at the deal level later.

Step 6Contact becomes MQL, then SQL, then Opportunity

Priya's demo is qualified by an AE. She becomes an MQL on day 28 (day of form fill), SQL on day 31 (AE booked), and an opportunity on day 42 (technical discovery call completed, deal stage set to Qualified to Buy).

Opportunity creation triggers Workflow 2 (from the HubSpot implementation guide): the deal record is stamped with:

• aeo_influenced = yes
• aeo_first_touch_engine = chatgpt
• aeo_first_touch_prompt = "best tools for AI search visibility and attribution"
• aeo_first_touch_timestamp = day-0 form-fill equivalent (from the contact record)

Step 7Stage-weighted pipeline reporting

At the end of the month, the CMO pulls a HubSpot list: all deals created this month where aeo_influenced = yes. Sum amount × stage_probability. Priya's $48K deal in Qualified to Buy (probability 10%) contributes $4,800 to the month's stage-weighted AEO-influenced pipeline.

As the deal progresses: Decision Maker Bought In (40%) → $19,200. Contract Sent (60%) → $28,800. Commit (80%) → $38,400. Closed Won (100%) → $48,000.

Each month's report shows both current-month AEO-influenced pipeline (forward-looking) and AEO-influenced closed-won (realized).

Step 8Closed-won credit and the CFO memo line

On day 118 (4 months after first ChatGPT citation, 90 days after the demo), the deal closes won at $48K ARR. Workflow 3 fires:

• aeo_attributed_pipeline finalizes at $48,000
• Timeline event logged: "AEO-influenced deal closed-won: $48K ARR, first-touch ChatGPT, prompt: best tools for AI search visibility and attribution"

In the quarterly memo to the CFO, Priya's deal appears in the "Evidence" section (from the memo template in the CFO's Guide to AEO Budget Defense):

The deal is evidence. It's auditable — a Marketing Ops query can pull the exact same numbers the CMO is reporting. That's what makes it CFO-safe.

Common failure modes (and how to fix them)

The self-reported field is blank

~25% of inbound forms will have a blank self-report field. If the contact has a UTM-tagged landing session or a citation-proximity session, the AEO attribution still fires. If none of those fire, the contact is marked aeo_influenced = no — known limitation, acknowledge in the methodology disclosure.

Buyer typed "AI" but meant something else

Occasionally someone will write "AI" meaning "I use AI in my job" rather than "I found you through AI." Parse with explicit engine names first (ChatGPT, Perplexity, Claude, Gemini); only fall back to "AI" when no specific engine name is present, and log the weaker confidence.

Prompt attribution is wrong

If the citation-proximity match picks the wrong prompt (multiple prompts served the same URL in the same window), the reported prompt is the highest-confidence match. Quarterly prompt-audit: cross-check the reported prompts against what sales heard on discovery calls. 80–90% accuracy is the realistic target.

Last-touch overrides first-touch in the wrong report

HubSpot's native Original Source is last-touch-biased for any session that closes via paid search or direct. This is structural and cannot be fixed. The fix is to report on the AEO custom properties in your CFO memo, not on the native Original Source fields.

What this playbook proves

End-to-end traceability of a single AEO-influenced deal — from the ChatGPT citation serve through self-reported form submission, property propagation, stage-weighted pipeline reporting, and closed-won credit. The CMO can answer, for any specific deal, "here is the ChatGPT prompt that introduced this buyer, here is the date, here is the landing page, here is the self-reported confirmation, here is the closed-won ARR."

That's the level of attribution detail that changes a renewal conversation from "share of voice is up 12%" to "here are three deals ChatGPT introduced, $127K closed-won." Different conversation, different outcome.

Where Lantern fits

The playbook above can be run manually by any RevOps team with the HubSpot property schema in place — minus the prompt-level enrichment, which requires a real-time citation monitoring system that also writes to HubSpot. Lantern is that system: continuous citation monitoring across ChatGPT, Perplexity, Claude, Gemini, and AI Overviews, plus HubSpot-native contact-and-deal property writes, plus the monthly CFO PDF.

$99/mo for the self-serve setup. Enterprise for co-built memo templates and bespoke prompt-audit cadence. Two tiers, no per-prompt math.