Watch part of a page
By default the monitor scans the whole page. The area selector PRO narrows the scan to one section — the price box, the stock line, the appointments table — so nothing else can trigger a false alert.
Why limit the scan area?
Real pages are noisy. Menus, footers, rotating ads, "customers also bought" rails and cookie banners all contain text that changes. If you watch the whole page:
- a keyword like
in stockcan match a recommended product instead of yours; - Any change mode fires on every rotating banner, making it useless.
Point the monitor at just the product box and both problems disappear. This works together with everything on the keywords page — all your keywords simply look only inside the chosen area.
The CSS selector box
In the Monitor tab you'll find an area selector field. It takes a CSS selector —
a short "address" for a part of the page, like #price (an element with id "price") or
.product-info (elements with the class "product-info"). Leave it blank
to scan the whole page, which is the default.
If you know a little HTML you can type a selector by hand. If not — don't worry, you never have to. That's what the element picker is for.
The element picker
The Pick element button PRO lets you just click the part of the page you want to watch. Refresh Pilot writes the selector for you.
- In the Monitor tab, click Pick element. The popup closes and the page takes over.
- Your mouse cursor becomes a crosshair. As you move it, a dashed box outlines whatever is under the pointer, with a tooltip reading "Click to select · Esc to cancel."
- Hover until the dashed box wraps exactly the section you care about, then click.
- A small "Selector captured" toast confirms it worked.
- Reopen the popup — the area selector box is now filled in with the generated selector.
The picker in action: the crosshair moves to the price box and one click captures its selector.
Why the captured selector keeps working
A selector is only useful if it still finds the same element after every reload. The picker builds the most stable address it can, in this order of preference:
- the element's own
#id— the gold standard, unique by definition; - stable attributes such as
data-testid, other data attributes,nameor aria attributes; - a filtered combination of dependable class names, scoped under the nearest element that does have a unique id;
- as a last resort, a positional path (
nth-child) down the page structure.
Every candidate is checked for uniqueness before it's accepted, so the captured selector points at exactly one element — and keeps pointing at it, reload after reload.