What is a watch?
A watch is a computed figure, a scope and a threshold. Watch days of cover on this variant and tell me when it drops below its lead time. Watch realised discount on this code and tell me when it exceeds the headline rate. Watch Saturday conversion at this store and tell me when it falls below the chain. The analyst turns the sentence into a stored query and a rule.
From then on the analytics engine evaluates the watch after every sync. When the rule fires, the movement is recorded with the figure's history, so the owner sees not just that it crossed but how it got there.
How do you set a watch?
From the chat analyst, ask it to watch the figure you just discussed. From any KPI card, choose watch this and set the threshold. From the inventory analyst's list, watch a line. The analyst confirms the exact figure, the scope and the rule back to you before saving, because a watch on the wrong scope would fire on the wrong thing.
Watches are listed in one place with their current value, their threshold and their last movement, so an owner with a dozen of them can see at a glance which are quiet and which are close.
- Days of cover on a variant, per location, against its lead time.
- Realised discount on a code against the headline rate printed on it.
- Footfall conversion at a store on a given weekday against the chain's figure.
- Return rate on a style after a page fix, to see whether the fix held.
- Unfulfilled backlog age at the warehouse against the promise made at checkout.
Where does a movement show up?
On the Morning Brief, in the section for what the platform is watching, with the figure, the threshold and a link to the surface it comes from. In the chat analyst, where asking what moved lists the watches that fired since you last looked. And, where an owner has asked for it, in the manager email for the location the figure belongs to, through Review & Send.
A movement is information, not an action. The watch never reorders, never pauses a code and never sends anything. It tells the owner, and the owner decides, which is the third law applied to alerts.
Why do watches beat dashboards for figures you already track?
A dashboard shows everything, every day, whether it moved or not. A watch shows one thing, only when it moved. For the handful of figures an owner already knows to care about, a watch is the honest interface: it is quiet until the number changes, and it carries the history when it speaks.
Watches also survive the owner's attention. A concern raised in a conversation in January is still being evaluated in April, on every sync, whether or not anyone remembered to check the chart.
What a watch cannot do
It cannot watch a figure the engine cannot compute. If a store's footfall is not connected, a conversion watch for that store says so instead of watching nothing. It cannot predict; it evaluates the current value against the threshold. Forecasts live in Foresight, where they are scored daily against reality.
And it cannot act. Every watch is a sentence that ends in tell me, never in do this. That boundary is what lets an owner set as many as they like without worrying about what might happen overnight.
What is the Watching section on ftes.ai?
On ftes.ai each e-commerce Digital FTE keeps a Watching section on its morning page: the figures it is tracking for its own function, such as the Cash Officer FTE watching lines approaching dead stock or the Pricing Officer FTE watching a code's realised discount. The engine's watchlist is the same mechanism with the owner as the author.
Start by asking the analyst to watch the one figure you check most often. When it moves, you will hear about it on the brief, with its history, and you can stop checking.