Net is gross less refunds and it is the settlement base — the figure a percentage agreement is worked out on. It is food only: fees and tips were never the event's money.

These are distinct PEOPLE across the whole event, not the sum of the per-truck figures — somebody who ate at three trucks is one person. The anonymous orders are stated separately and must never be added to the figure above: a sponsorship priced on attendees that turn out to include thousands of blanks is a deal that gets pulled.

Every hour is labelled on the UTCclock — yours, not the server's. Hours in which nothing sold are not drawn.

One row per approved truck, including the ones that took nothing. A truck missing from its own event's ranking would be indistinguishable from a read that dropped it.

The number on the left of each row is how many of that truck's customers had never bought at one of your events before. The number on the right is how many people bought there and left a contact at all.

THESE ARE TWO COUNTS AND NEITHER IS A REASON. A truck at the top of this list sold to people who were new to your events — it does not prove the truck is why they came. A queue that nobody could get past also puts a truck at the top of "bought at no other truck here", and a market that simply got busier makes everybody on site look like a draw. Read it as a question to ask, not an answer.

"New" means they had not bought at any of your last 2 events. Somebody who came to a market of yours before that, or who paid cash and left no contact, counts as new here.

Counts people who BOUGHT something and left a contact — not everybody who walked the site, and not the cash sales. The orders with no contact are stated above and must never be added to the headcount: they may be the very same people.

What to do with it: if a truck is high on both columns, ask them back and ask what they did — a post, a regular crowd, a queue people came for. If a truck brought no new faces and sold well anyway, that is still a good booking; it is just a different one.

The median, deliberately, and there is no average anywhere on this line. At an event where two trucks take a third of the money, the average sits above almost every truck on the site — price a booth off it and you charge most of your roster more than they made. The middle half is the range a fee actually has to survive.

Trucks that took nothing are counted in, because a booth fee was charged to them too.

This is YOUR money — what the trucks paid you to stand on your site. The takings at the top of this report are what the trucks took through their own windows, and the two must never be added together: unless your agreement is a percentage, none of that money was ever yours.

What the booths paid you, less what the site cost. It counts money COLLECTED, not raised — an invoice nobody paid is not income.

What you owe each truck for the discounts they honoured. Only trucks that took one are listed — a row of $0.00 against every truck on the roster is a line you would have to read and dismiss forty times.

We do not tell you that passport holders spent more than everybody else, and we are not going to. They almost certainly did — and it would prove nothing, because the people who buy a festival passport are the keenest people coming. They would have spent more without one. What is above is what was sold, what was spent and what it cost you: three counts, and no claim about anybody's behaviour.

Every price here is the one that buyer actually paid, from the terms frozen when they bought. Changing a tier this morning does not rewrite what last month sold for.

The percentage is how much of the event had already run when their first item went — the only version of this that compares between a four-hour market and a twelve-hour festival. 1 of 2 trucks whose log answered ran out of something.

1 item was taken off a board by hand. Counted beside the sell-outs and never added to them: running out is the day happening to a truck, an 86 is a decision they made.

AN ESTIMATE, and it is a statement about an observed rate rather than about people. Nobody records the customer who read "sold out" and walked on, so this is what those items were selling at before they went, extended across the hours they stayed gone. It is not a count of customers and must not be quoted as one.

• Takings, against the best truck here50%
• Reliability — showed up, and checked in at the gate30%
• How much of the event they kept serving20%

Published weighting before anything was dropped: takings 50%, reliability 30%, serving 20%. Reliability splits 60% for turning up and 40% for being checked in at the gate — a truck that withdrew after being approved scores zero on the whole of it, because that is a hole in your site plan that no takings repair.

This is a starting point, not a verdict. It cannot see that a generator failed through nobody's fault, or that the truck at the bottom is the only vegan pitch on the site and the site needs one.

• Net takings$1,158.00 · median of your last 2: $829.50 · +40%
• Orders70 · median 51 · +37%
• People who bought41 · median 34 · +21%

The baseline is the MEDIAN of your own past events, not the average — one wet Saturday or one blowout year would otherwise become the bar every future event is judged against. At least 2 previous events that traded are needed before any of this is said at all.

This compares your events to your events. CraveZee does not tell you where you sit against other organizers, at any level of aggregation — with a handful of festivals in any size band, a percentile is somebody else's takings with a rosette on it.