When a request comes in through the website, the voice agent calls the lead back, runs a structured qualification flow, scores the result, and forwards only the leads worth a manager's time into HubSpot.
After submitting a request on the website, a potential customer receives an automated call from the voice agent. The system asks a structured set of qualification questions — rental dates, equipment type, location, and budget limits — captures responses as structured data, generates a short conversation summary, and scores the lead for quality. HubSpot is then updated with the qualification result, so managers instantly see who has already been called by the agent, who passed qualification, and which leads are worth a live follow-up.
Futura Trailers rents out trailers and specialized equipment across the Benelux region. Inbound requests come in through the website and land directly in HubSpot. Content marketing was generating more and more inquiries — roughly 400–600 new submissions a month — but a small sales team of four to six managers couldn't keep up with demand. Ruben, the company's Content Marketing Executive, championed a system that would filter out irrelevant requests so managers could focus on the leads that actually move revenue — without disrupting the HubSpot funnel the team already relied on.
Inconsistent qualification workflow
Managers were qualifying leads on intuition. There was no shared list of quality criteria, and institutional knowledge lived in individual managers’ heads.
HubSpot integration without breaking the existing pipeline
The system had to receive new leads and update fields and statuses in HubSpot without disrupting the existing pipeline structure and automations the team already depended on.
State synchronization with the voice AI provider
The website form, HubSpot, and the voice-calling service had to stay in sync under real conditions — handling errors, timeouts, and retries without losing or duplicating leads.
Turning an open conversation into a scorable lead
A call is an open dialogue, but sales needs simple, sharp indicators: whether the lead is a fit, what priority level it should get, what the main parameters of the request are, and — if disqualified — why.
Call-time restrictions across markets
Automated calls couldn't go out at night or during quiet hours, and allowed windows to differ between markets. The system needed flexible launch and retry rules per region.
Team transparency, so managers actually trust the AI
For managers to rely on the voice agent's assessment, the system had to be transparent — clear logs, full transcripts, and short summaries on every call.
We built a dedicated lead-qualification service sitting between the website and HubSpot. The system monitors new submissions, triggers calls, stores transcripts, and turns responses into structured fields used for scoring. Results flow automatically into HubSpot, where managers see a call status and a ready-made conversation summary inside the existing lead card.
Form and HubSpot integration
We connected the service to new-lead creation events in HubSpot. Form data is normalized, mapped into a single internal schema, and stored alongside technical call status data. Records in HubSpot are updated through the official API, respecting the client's existing properties and automations.
Voice-agent call orchestration
The backend builds requests to the selected voice AI provider (Vapi) with the contact details and conversation script. The system tracks call attempts and states — "no answer," "call back later," "successful conversation" — and stores them in its database. Each lead carries a full call history so the team can avoid endless retries. Retry policy: up to three attempts, spaced 30 minutes / 2 hours / 24 hours apart, all respecting per-market calling windows.
Question flow and answer capture
We defined the dialogue as a structured business flow — questions about trailer type, rental duration, site of use, budget, and any special requirements. Answers are converted into structured fields (dates, categories, numeric values) and enriched with tags such as "long-term rental," "non-standard request," or "outside service area."
Lead scoring model
Scoring is calculated from the answers: baseline service-area fit, availability of the requested equipment type, budget relevance, and readiness to commit to specific dates. Threshold values live in configuration so marketing and sales can adjust the criteria themselves.
Logs, transcripts, and summaries
For every call, the system stores a short written summary, the key answers, and a technical log (start time, duration, completion reason). HubSpot carries a link to the extended call view, while the lead card itself shows a short summary that lets a manager grasp the context in 10–15 seconds.
Monitoring and settings panel
On top of Directus, we built an administrative interface in React with call lists, status filters, individual conversation views, adjustable scoring thresholds, and configurable calling windows. This gives the operations team control over the agent's behavior and visibility into its quality without involving developers.
With the client's sales team, we broke the existing process into clear steps — from the form submission, through the manager's first call, to follow-up inside HubSpot. We separately listed the questions managers always ask, the ones that come up only sometimes, and what actually drives the decision to send a commercial offer. That gave us the mandatory data points the voice agent had to capture and the minimum set of statuses the funnel could work with.
+1 Resource
After the voice agent went live, Futura Trailers had a predictable first contact on every new lead and a consistent qualification standard across the team. Managers could focus on leads with real deal potential.
handled by the sales team, thanks to pre-qualification
reaching a conversation with a manager — contact happens without delay and the manager already has the context
answers, call status, and a short request summary
across the whole team
thanks to configurable questions, calling windows, and scoring rules
Or send us a message and we'll get back to you within 15 minutes during business hours