Fair knowledge has a half-life of about a week. The maker who writes down four numbers and three observations on Sunday evening builds a season-planning machine; the maker who doesn't gets to rediscover next October that the Town Hall fair is all browsers and the tiny church one is all buyers. Twenty minutes, while the van's still warm.
Takings (counted properly: cash minus float, plus the card reader's total). All-in cost (pitch fee + travel + food + card fees). Profit — the subtraction nobody does while tired. And items sold, roughly by product line. That last one converts fairs from "felt busy" to "sold 24 small things and 2 big ones", which is the difference between a mood and a plan.
What did people pick up and put down? That's demand with a pricing or presentation problem — next fair's cheapest experiment. Who was the crowd? Grandparents buying for grandchildren shop differently from twenty-somethings buying for flatmates; match future stock to the actual room. What did neighbouring stalls charge? Not to copy — to place yourself deliberately next time.
Every fair gets one of three words in its notes. Rebook: profitable, pleasant, apply early next year. Tweak: promising but something specific needs changing — different stock mix, better position, Saturday instead of Sunday. Drop: the maths said no twice; believe it, however nice the organiser. Two seasons of verdicts and your applications list writes itself — all bankers, chosen experiments, and no third year at the fair that's never once covered its fee.
The review is worthless in your head and priceless in the same place as next season's applications — a note against the fair's row ("took £340, rebook!", "all browsers, drop") turns last year's data into this year's decisions automatically.
General guidance, August 2026. CAAC.