Revenue analyst agent for a small education and training company on HubSpot that joins enrollments, billing and learner activity by term

What The Term Analyst does

A 30-person education and training company sells programs three ways: seat contracts to schools and districts, enrollments paid by parents, and courses bought directly by adult learners. Deals and contacts live in HubSpot, invoices and payment plans live in the billing system, and enrollments, logins and course completions live in the learning platform's product database. Nobody joins them, so simple questions take days: how many seats did a district buy versus activate, which payment plans are behind going into the new term, and is spring enrollment ahead of last spring at the same point in the cycle. The Term Analyst works for the RevOps lead and the leadership team. It owns a governed join model keyed on institution, account and term (not on individual students), answers plain-English questions with the source tables and definitions shown, publishes a Monday digest aligned to the academic calendar, and raises data quality issues it finds in the joins. Because the team chose build and maintain, an outside operator keeps the connectors, definitions and term calendar current as systems and terms change.

Built on HubSpot

HubSpot is the system of record for institutions, parents and learners, so Breeze is useful for in-CRM summaries and the agent's notes belong on HubSpot company records. But the core problem is a three-way join across billing and a product database HubSpot does not hold, plus strict control over what learner data is exposed. That calls for a custom AI build: MCP connectors that return only governed, aggregated fields, a metric dictionary and term calendar the agent must follow, n8n for scheduled reconciliation and digest delivery, and Databox for the dashboards leadership drills into. This keeps student data in its source system while still giving the team one answer per question.

What starts it

• Monday 6am digest run, with the comparison window set by the current term and its equivalent point in the prior year's cycle
• A plain-English question asked in the team's chat channel or the HubSpot sidebar
• An institutional contract entering 60 days before its term start or renewal date
• A daily reconciliation run finding a closed-won HubSpot deal with no matching invoice, or an invoice with no matching enrollment
• Enrollment deadline or add/drop date on the term calendar, which triggers a same-day enrollment snapshot

What it reads

• HubSpot deals, companies and contacts (institution, parent and adult learner records, deal stage, term property, contract value)
• Billing system invoices, payment plans, refunds and credits, read through a read-only API connector
• Learning platform product database (enrollments, seat activations, logins, completions), exposed only as aggregated views by institution, program and term
• A maintained term calendar table with start dates, enrollment deadlines, add/drop windows and renewal dates per institution
• A metric dictionary defining terms like booked seat, activated seat, enrolled learner, and term revenue

When a person takes over

• Any question that would require student-level data (for example, which specific students at a school have not logged in) is refused and routed to the person authorized to access education records under the institution's agreement, typically the customer success lead.
• Identity matches the agent cannot confirm, such as a parent paying for two children under different email addresses or a district billed under a different legal name, go to the RevOps lead for a manual match before they enter the model.
• When the digest shows a term-over-term change large enough to change a hiring or marketing decision, the RevOps lead reviews the underlying numbers and the term alignment before the digest goes to leadership.
• Changes to metric definitions (for example, whether a refunded add/drop enrollment counts as term revenue) are decided by finance and RevOps together, and the agent only uses the new definition after it is updated in the dictionary.

Rules it follows

• FERPA: education records and personally identifiable student information never pass through the agent. The product database is exposed only as aggregated counts by institution, program and term, and any count under a set minimum cell size is suppressed so small classes cannot be re-identified.
• Audiences stay separate: institutional, parent-paid and adult learner revenue are never blended into one number without being labeled, and the agent never drafts communication to parents or students.
• Every answer cites its source systems, the definition used, and the term window. If systems disagree, it shows both figures and the reason rather than picking one.
• Read-only access everywhere. The agent can write summary notes to HubSpot company records, but cannot change deal stages, invoices, enrollments or contact properties.
• Time comparisons use the term calendar, so a slow week before enrollment opens is never reported as a decline against a peak week from the prior cycle.

How you know it works

• Time from a leadership question to a sourced answer, compared with the manual baseline the team measures in the pilot
• Share of closed-won deals with a matched invoice and matched enrollment record, tracked weekly
• Number of institutional renewals where low seat activation was flagged at least 60 days before the renewal date
• Weekly digest delivered on time with zero manual edits to figures by the RevOps lead
• Count of student-level data requests correctly refused and routed, reviewed monthly as a compliance check

From pilot to live

1. Pilot (one term, institutional contracts only)Build the identity map and MCP views for institutional accounts, run the daily reconciliation, and send a draft digest to the RevOps lead only. Validate every figure against a manual pull for the current term and fix definition gaps before anyone else sees it.
2. Expand to parent-paid and adult learner revenueAdd payment plans, refunds and add/drop handling for direct payers, open question answering to leadership in the team channel, and start the renewal risk flags. Run the FERPA review of every exposed field with whoever owns institutional data agreements.
3. Full operation with ongoing maintenanceThe digest goes to the full leadership team weekly. The maintaining operator updates the term calendar each cycle, adjusts connectors when billing or the learning platform changes its schema, reviews refused questions monthly, and adds new governed views as leadership asks new kinds of questions.