A win, when there is one
On a morning with nothing to fix, a GM may hear what went right. Most mornings they won't.
On a quiet morning, with nothing at a store worth fixing, AI COO may tell that store about one thing going right. Most mornings it won't, and that is deliberate.
Why wins exist
A brief that only ever reports problems teaches its reader that it is a complaint machine. A GM who is only ever told to fix something stops reading, and then misses the one that matters. A real win, told plainly, keeps the channel worth opening.
Why most mornings have none
The word this feature refuses is "always". A win every day is a quota, and a quota produces a manufactured win on a bad week. "Labor was only slightly over target" the morning after a rough week reads as not paying attention. So a win has to clear the same kind of bar a lever does. Three things have to be true.
The move is unusual for this store
Each store is compared with its own history of week-over-week changes, measure by measure. A store that climbs 2% every week doesn't earn a win for climbing 2% again. A change has to stand out from that store's usual range. Stores without enough history yet fall back to a fixed minimum.
The gain is the whole week, not one day
A weekly total can be carried by a single large order. One catering order on a Thursday can make a flat week look like a 10% jump. AI COO compares each week's total with its typical day, so a gain counts only when it is spread across the week. Both weeks must cover the same days, and at least five of them.
The top line didn't fall
Serving more orders while net sales fell is true, and it is also a cherry-pick. A win on any measure needs net sales not to have gone backwards, beyond a small tolerance.
A return to normal is not a win either. A store that bounces back after a soft week is described as back in its usual range after a soft week, not as up 19%. Net sales is a win only when the week is above the store's own normal range.
How it's written
A win states what happened, with its comparison and both figures, for example "up 13.7% this week vs last, $17.0k against $14.9k". It never says who did well, because nothing in the data shows who did what. Hourly labor is never a win measure: a labor percentage that fell is as often a store that ran short through a rush.
Who hears it
- The GM. On a morning with no lever for their store, the win is their whole message, kept to about two text segments, with nothing to reply to. On a morning with a lever, the win sits underneath it and is the first thing trimmed if space runs short.
- The owner, and corporate readers. Their brief names the store and carries the best win across the portfolio.
Told once
A good fortnight would otherwise qualify every morning with a slightly different percentage. So once a store has heard a win, it goes quiet on wins for a few days. Seven texts saying roughly the same thing is how a GM learns to turn notifications off.
Wins never appear in a brief that couldn't be produced properly. Good news paired with a failure teaches people to distrust both.
See how AI COO decides what to send.
Common questions
- Why does Franlever sometimes send a quiet-day win?
- A store can be running normally while something specific improved. A quiet-day win acknowledges a supported improvement without turning it into another task. Most quiet mornings do not need one.
- Does AI COO invent a positive message for every store?
- No. A win needs evidence in the store’s numbers. If no meaningful improvement is supported, the brief can simply say the store is running normally.
- Does a quiet-day win mean the store has nothing worth monitoring?
- No. It means no lever cleared the bar for that brief. AI COO continues monitoring connected records, and a later change may call for a new lever.
- Will the same store win be repeated every day?
- A win is acknowledged once rather than repeated as daily filler. This keeps the next brief focused on what changed and what deserves attention now.