Skip to content

Setup Check

Setup Check analyzes your relationship configuration — link groups, visibility groups, rules — and reports everything that would behave unexpectedly in the published viewer. Run it before publishing; each finding jumps to the exact item that needs fixing.

You’ll find it as Check setup in the matching editor, with the current finding count next to it.

Step 1/5

Clicking around a live preview only exercises the states you happen to visit. A product with dozens of linked parts has far more states than that.

Setup Check — catch it before shoppers do
Step 1 of 5
  1. Clicking around a live preview only exercises the states you happen to visit. A product with dozens of linked parts has far more states than that.
  2. Relationships are authored in one place — links, show/hide groups and rules all end up in the matching editor.
  3. Open the matching editor — the check runs against everything authored inside it.
  4. The badge already counts what is wrong. This is static analysis of the authored configuration — not a simulation — so it reports the states you never visited.
  5. Findings name the exact item: empty mapping cells, groups with no effect, dead axis values, rules that contradict a link, variations no shopper can reach, and missing SKUs that would drop out of the cart.

Clicking around the live preview only exercises the states you happen to visit. Setup Check analyses the authored configuration itself, so it surfaces the states you didn’t visit.

Link groups

  • Groups with no driven parts yet — configured but having no effect.
  • Empty mapping cells — driver values that silently leave a driven part unchanged.

Visibility groups

  • Dead coupling values — axis values that reference nothing, which would make a switch position show an empty scene.

Rules

  • Contradictions with links — e.g. “Link ‘Finish’ sets Cushion to ‘Walnut’, but rule ‘Body = Leather’ blocks it — the rule will immediately undo the link.” Named with the exact group and rule so the fix is obvious.
  • Unreachable variations — options no shopper can ever select.
  • Conditions a link keeps apart — a rule with several conditions that can never hold together because a link moves those parts as one, e.g. “Rule ‘Body = Walnut and Shelf = Black’ never applies in the picker: link ‘Finish’ moves Body and Shelf together.”
  • Duplicated rules.

Assembly floor

  • A part reaching below the floor plane — in the published viewer the product appears to float above the ground. The finding names the part; fix its origin in the source file and re-upload. If the part genuinely should reach lower than the others’ floor, ignore the finding.

SKUs

  • No SKUs at all — a reminder that add-to-cart from an embedded store needs one SKU per variation. Harmless if you’re publishing a showcase rather than selling.
  • Mixed coverage — some variations have SKUs while others don’t, so the empty ones silently drop out of the add-to-cart payload.
  • Duplicate SKUs — deep links and cart mapping resolve to only the first; the rest become unreachable. The finding lists the exact variations.

Findings are warnings (will misbehave publicly — fix before publishing) or info (harmless but probably unintended, like an inert group).

Where a gap has an unambiguous fix — like unfilled mapping cells — Auto-fill completes it for you. Review what it chose, then publish.