The Academic Year Calendar.

Two patterns drive the chart. The first is the September spike. Move-in week at most colleges is the third or fourth week of August. By the first week of September, every campus is back in full session. For two weeks, the metro absorbs the equivalent of a mid-sized American city in incoming population. Allston-Brighton alone takes on roughly 30,000 students; Cambridge another 18,000; the Fenway and Mission Hill 12,000 between them. Order volume at restaurants within a one-mile radius of any campus runs 1.5 to 2x baseline for at least four weeks. The defaults customers form in the first seven days, including which app they open first when they want pad see ew at 10:30 PM, stick for the rest of the academic year.

The second is the May cliff. Six universities graduate inside a twenty-one-day window. Harvard, MIT, BU, BC, Northeastern, Tufts. Then dozens of smaller schools. Then the families leave. The compression is unusual for the United States and globally distinctive: New York spreads its student calendar across more dispersed neighborhoods; Los Angeles is too big to feel a single move-in week; London skews older and less dense; Tokyo runs a different academic year. Boston has no peer for what happens to a restaurant's P&L between mid-May and mid-September.

The intermediate months tell the same story in lower amplitude. November is steady. December finals push some customers off and bring corporate holiday catering on. January reopens. February holds. March has St Patrick's Day, which in South Boston runs to a 4x baseline, plus a spring break dip in the second week. April is the Marathon (Chapter III), then Red Sox home opener at Fenway. Then May. Then the cliff.

For an Allston-Brighton operator, the question isn't whether the calendar exists. The question is whether the platform handles the calendar. Can you flip your delivery zone from 5 miles to 1.5 miles in fifteen seconds when a student crush concentrates around the Harvard Ave T? Can your phone line handle Spanish for Eastie callers and English for the dorm calls? Can you push a graduation menu live by mid-April so it indexes in time? Can you flip back to a corporate-catering-only summer menu the second Sunday of June without rebuilding the site? Most restaurants can do parts of this. Almost none can do it on a marketplace.

The platform decision is the calendar decision. If your direct ordering system can absorb a September spike of 1.5 to 2x baseline and a July trough of 40 to 60 percent below baseline, the year works. If it cannot, you spend the calendar reacting to a 25 to 30 percent commission ceiling while the calendar itself moves the ground underneath you.

The next two chapters cover the two big detours off the calendar: Kendall Square (where the calendar matters less because biotech does not graduate) and Marathon Monday (the one day each year where the calendar bows to a single civic event).

Kendall Square is an exception to the academic calendar argument. Moderna, Biogen, Vertex, Novartis, Pfizer Cambridge, Takeda, and Sanofi do not graduate. Their workforces hold steady through July. Their catering orders run on a different rhythm than retail dinner: weekly, recurring, $20 to $40 per head, with 24 to 48-hour lead times and procurement-grade invoicing. The Cambridge sub-market between Main Street and the Charles is a year-round economy with the lowest seasonal amplitude of any Boston neighborhood.

That is the opportunity. It is also the operational challenge. Biotech procurement at every one of these companies runs the same playbook: department heads request catering through internal portals; admins place orders 48 hours in advance; corporate AP processes invoices on net-30 terms; finance reconciles against quarterly budgets. Each step requires data the marketplaces do not produce. PDF receipts with no invoice number, no department code, no PO reference. Charge backs that arrive 30 days after the meal. Commission lines that procurement cannot explain to the CFO. The result is that even though biotech employees order from DoorDash for personal lunch, the catering channel runs through whoever has a real catering portal with net-30 invoicing.

Whoever has the catering portal answers the Tuesday-afternoon email from Moderna's admin asking for forty boxed lunches by Thursday at 11:30 AM. Whoever does not, loses the line item for the year. The line item is meaningful: a single recurring Vertex weekly order of $2,000 across 50 working weeks is $100,000 in annual revenue at zero marketplace commission. For a fast-casual concept with $1.2M in topline revenue, that is 8 to 9 percent of annual sales from one customer, on a payment structure that pays on the fifth of every month at net-30 invoicing.

The geographic note is short: the corridor we are calling Kendall Square fits inside 02139 and 02142 with a small extension into 02141. Walking it takes about twenty minutes. Within that mile and a half are roughly half of America's top thirty publicly-traded biotech firms, the largest concentration of biotech venture capital, MIT's research campus, and approximately 75,000 daily workers per the Kendall Square Association. There is no second cluster in the United States of comparable density. The corollary for restaurant operators is that there is also no second comparable lunch and catering market.

For an MIT-adjacent operator, the platform decision is whether your direct ordering channel supports a catering portal with department-level budget caps, group order links shareable in Slack and Teams, scheduled pickup windows, net-30 invoicing, and itemized receipts. If yes, you compete for the recurring weekly orders. If no, you compete for the one-off marketplace tickets where DoorDash takes 30 percent.