We make all year and sell in ten days. Can we pay for something like this out of one October?
That is exactly how we would price it. The figure is fixed and written down beforehand, and we size a first build so it comes out of a single festival rather than out of savings. We would also want it finished and tested by late summer — anything we cannot land with a month in hand we will tell you to postpone until November rather than deliver into your only trading window.
Our stall is at two villages. Does that actually change the software?
Completely, and it is why generic market-stall tools disappoint people here. One pitch is a till. Two pitches ten miles apart, for ten days, with stock you cannot easily reallocate, is an inventory problem — you need to know by Tuesday lunchtime that Mansfield is running out of something Rosedale has a box of, while somebody can still drive it over.
Two million visitors sounds like a made-up number. Do you believe it?
We treat it as a tourism estimate, which is what it is, and we would not build anything that depended on it being precise. It does not need to be. A county of sixteen thousand hosting ten days of this is extraordinary whether the true figure is two million or a quarter of that, and the operational consequences — parking, permits, sell-through, beds — are identical either way.
We are the county office, not a business. Does any of this apply on our side of the counter?
It applies most in October. The ordinary work does not pause because the festival started, and the usual failure is that record requests and permits fall behind for three weeks and take until Thanksgiving to recover. Putting the routine requests online and separating them from general enquiries is the cheapest fix we know for that, and it also means fewer people at the window asking where to park.
Everything we know about what sells is in one person’s head. What happens when it goes into a system?
It stays yours, permanently and exportably, and that matters more here than most places — a decade of what sold at which village on which day is genuinely valuable commercial information about a market almost nobody else has data on. It is not shared, not sold, not used to train anything, and you can take the whole record out in a standard format whenever you want.
We already use a card reader and a spreadsheet. What would you actually add?
Possibly nothing, and we will look for that answer first. If you trade at one site and carry your stock in a van, keep what you have. The point where it stops working is when you are running several locations, or holding deposits a year out, or reconciling three sets of takings against one production run — that is where the spreadsheet starts costing more than it saves.
Where is the artificial intelligence in a craft festival, honestly?
In two narrow places. Reading documents — a vendor’s insurance certificate or food permit arriving as a phone photograph, with the dates pulled off it rather than typed. And pattern work on your own past sales to propose a stock split you then overrule. Everything else, the schedules and the queues and the deposits, is ordinary programming and we price it as such.
Ten days is a long time to be on your feet. Will anyone be reachable during the festival?
Yes, and we plan for it. Anything we deliver before October gets an agreed contact and a response time for the festival window itself, because a problem on the second Saturday cannot wait until Monday. We would rather set that expectation in writing in August than have you find out how we behave in a crisis during your busiest week.