The 60th Annual Edgewood Fair is the organization's flagship fundraiser, and I owned its data and operations end to end. As the real work holds protected data, the version explored here is TerraGlade, a fictional nonprofit running its 100th Fair in a similar context. Before the rebuild, the gala ran across siloed spreadsheets, one per team, with table planning split across files and guest data unstructured. Reconciliation was manual, data-driven rollups were difficult, and post-event cleanup was lengthy.
I rebuilt it as a single relational Airtable base, designed to stay lean: a live Constituents base and Gifts base synced in, so the event pulled only what it needed while the data stayed clean and reusable. Eight linked tables carried the night. I anchored the structure on a fixed table reference and modeled every seat as its own ticket record rather than a count, so the base could always say who held a seat, which gift paid for it, and whether it was open. That granularity is the spine the rest of the build resolves down to.
Attendees linked to the synced Constituents base, each record tied back to the organization's CRM, which made the base an operational layer on the existing system rather than a replacement. The table is only as good as the data flowing in, and the people closest to that data were the volunteers. Moving every team into Airtable was the clean design on paper, but the practical plan met the outside planners in the tool they already used. I built a Table Captain system as 30 linked Google Sheet templates and ran an automated, data-templated email strategy inviting captains to fill them. Twenty of thirty came back active, a 67% rate, their inputs flowing into the model without anyone learning a new tool.
The deeper decision was timing. Instead of cleaning records after the event, I collected upstream so data came in correct rather than getting repaired on the way out, with Claude Sonnet field agents drafting suggestions, surfacing data issues, and assisting cleanup inside the base. The Attendees table held better than 90% accuracy against day-of attendance through a final week of constant change.
Sponsorships were where value leaked between a pledge and fulfillment. A commitment has to clear a dozen steps before it becomes a fulfilled gift with the right recognition, and on the old sheets that lifecycle was error-prone. I built a robust sponsorships table, organized by tier and linked to both fulfillment status and the gifts that paid each one down, so a commitment could be followed from the moment it was made to the moment it cleared. Because the figures rolled up from the same synced gift records, the team could always see which were open, which directly informed outreach strategy. Unfulfilled sponsorships dropped 60% year over year, from fifteen to six. The improved tracking was fundamental.
The gift pipeline was the point of the whole build. Rather than stand up a second place to enter gifts, I kept the Gifts base I already ran as the system of record and synced it into the Fair base, so every gift, whether it came through GoFundMe Pro Live or by check, entered one coded workflow first and then flowed into the event. The Fair base never became a silo. Every gift linked down to the seat it purchased, which let the system roll totals up cleanly per attendee, table, and sponsor.
That single source of gift truth made real-time visibility possible. On top of the live totals I built dashboards and a weekly progress email to key stakeholders, so the team could see where gifts, sponsorships, and the fund-a-need stood without pulling a report. The fund-a-need proved it: pledge reconciliation that once took three business days closed in a single post-event session of about three hours.
The Fair was a milestone success: 30%+ year-on-year revenue, 330 incredible guests, 650+ gifts, and extraordinary stories. Records and process were captured cleanly by the relational engine I built in collaboration with multiple teams and stakeholders.
The screenshots, clips, and linked demo is that of TerraGlade, a fictional nonprofit hosting a fictional gala, with mock data for demonstration purposes.
