From spreadsheets to a sales platform — built through five rebuilds.

In 2024, the sales process was built around the founder as the sole salesperson. Bookings went through her Calendly, she manually reached out before calls to nurture leads, and she tracked everything herself. It worked — until we added salespeople.

Every expansion broke the system. Adding a second salesperson meant switching Calendly to round-robin, which broke every downstream automation. Tracking leads in Google Sheets worked for dozens of contacts but collapsed at hundreds. We moved to Asana with prepopulated stages, but it was built around specific people, not the role — so it broke when the team changed. We rebuilt again in ActiveCampaign deals to keep all the data in one place, but contacts ended up with multiple deals across multiple pipelines, scattering their data.

Then we tried to automate pre-call nurturing. The sales team needed automated booking confirmation texts via JustCall — a welcome message with event times, reschedule links, and the understanding that while texts were automated, the numbers were tied to real people. We built it on Zapier. It worked until someone needed to cancel or reschedule — reminder texts would still go out for cancelled events, and any change to the Zapier flow would break it for everyone currently inside it.

Each version solved one problem and created two more. The fundamental issue was the same every time: the system wasn't designed to evolve.

Booking Automation Engine

When a prospect books a discovery call through Calendly, the system takes over. A webhook hits the booking engine on Railway, which:

• Checks ActiveCampaign for any existing deals across all pipelines
• Removes scattered deals so the contact has one clean record
• Creates a fresh deal on the booking pipeline — flagging rebooks, cancellations, and history so the sales rep sees context at a glance
• Schedules SMS confirmations through JustCall — timed based on how far out the call is
• Monitors event status — if the prospect cancels or reschedules, scheduled texts are suppressed
• After the call outcome, automatically enrolls the contact into tailored email sequences

The entire system is built as a state machine — every booking moves through defined states (BOOKED → CONFIRM_REQUESTED → CONFIRMED → COMPLETED), with every transition logged for debugging. The sales team's Zapier automations can be modified without affecting the booking flow, because the booking engine is a hard boundary that other processes can't break.

Analytics Dashboard

With the booking automation handling deal management and SMS, the next problem was visibility. The sales team needed to see the full picture — not just their own pipeline, but the entire funnel from first contact to close.

The hardest part wasn't building the UI — it was getting the data into one place. Prospect data lives across three separate systems, each with its own structure:

The export pipeline pulls from all three sources, builds a unified timeline per contact, and outputs a single dataset the dashboard can render. Rate limiting, pagination, retry logic, and deduplication handle the messiness of pulling from multiple APIs simultaneously. The booking automation engine writes directly to the same PostgreSQL database the dashboard reads from — so deal state changes, SMS events, and booking outcomes appear in the dashboard timeline in real time.

The booking engine runs on Railway with a PostgreSQL database. The dashboard is a single-page vanilla JS app — no framework overhead, instant load times. Both systems read from the same data layer.

Both systems are in production. The booking engine processes every discovery call automatically — deal cleanup, SMS confirmations, and post-call email sequences run without manual intervention. The dashboard is used weekly by the sales team for pipeline reviews and lead prioritization. An architectural rewrite of the dashboard is in progress to replace the static data export with a real-time database layer (Supabase with incremental sync), which will eliminate the manual export step and enable live updates as prospect data changes.