Bilby / Checks / Change

What's different on my store since last night?

Bilby's Change check writes a normalised snapshot after every visit — the set of scripts, the theme id and name, prices and stock labels on the visited product, title/meta/schema, a perceptual hash of the homepage, robots.txt, llms.txt and the top pages' redirects — and diffs it against the previous night. Most changes are informational; when a new script appears the same night a pixel stopped or the buy path broke, that coincidence is a P2 pointing at the likely cause.

What the Change check catches

Scripts added or removed since last night
Theme published (id or name changed)
Price change on the visited product, stock label change (a hero product going sold out)
Title, meta description, canonical or structured-data types changed
Homepage visual distance (dHash) above threshold
robots.txt or llms.txt changed; new redirects on top pages
Homepage got slower than last night by more than 20% (and which script changed)

Severity — the rule, not a feeling

P2A change that coincides with a Watch or Money failure the same night — "the new script appeared the same night the pixel stopped".

P3Nothing else is a finding; changes are facts in the report and the Monday brief.

Honest limit. The first visit has nothing to compare; the diff begins on night two. Snapshots compare the same visited product; if the store's navigation led Bilby to a different product, price and stock changes are not compared that night.

· 2 scripts added · 0 removed since last visit

· Price changed on the visited product: $28 → $32

· Homepage looks the same (visual distance 6/64)

✕ Homepage got slower: LCP 1.9 s → 3.4 s

✓ Same scripts as last visit

// LINES AS THEY APPEAR IN THE FEED · FROM A REAL VISIT

// CITABLE

Bilby diffs a normalised nightly snapshot — scripts, theme, prices, stock labels, metadata, a homepage perceptual hash, robots.txt and llms.txt — so a store owner learns what changed the night it changed, and a new script that appears the same night a pixel stops is named as the likely cause.

Evidence per finding: the before and after values side by side (script host + path, price strings, theme ids, hash distance out of 64).

Tier: Bilby, Bilby Pro, Agency (nightly) — not on the free scan. Methodology →

Questions people ask

Will I get an email for every change?

No. Changes are facts in the nightly report and the Monday brief. The only Change email is the P2 coincidence — a change on the same night something broke.

How is the visual diff computed?

A difference hash of the homepage frame (9×8 greyscale, 64 bits). A distance of 12 or more out of 64 is reported as "looks different tonight"; the frames are shown side by side in the report.

// SEE IT ON YOUR STORE

The free scan runs the Change check tonight — the whole report, nothing locked.

Scan my store free