Steps
- Go to Sort Orders and select a sort order.
- Use Preview to see how a collection will look.
- Review the sort effect annotations on each product to understand why it appears where it does.
- Adjust settings if something doesn’t look right.
- Save changes when you’re happy.
Preview session and customer segments
When your sort order uses a metric segmented by Session Type or Customer Type, use the matching picker in Preview Results. These pickers also appear when a segment is used by a feature inside a weighted group.
- Session Type simulates whether this is a new or returning session. Choose Any to leave the session type unspecified, New for a new session, or Returning for a returning session.
- Customer Type simulates a signed-in customer with no past orders or at least one past order. Choose Any to leave the customer type unspecified, New for no past orders, or Returning for at least one past order.
Preview a saved audience
- Create an audience if your store does not have one yet.
- In Preview Results, open Preview as and choose the audience you want to simulate.
- Review the products and annotations. Audience-specific conditions and matching browse request transforms now use that audience for the preview.
- Choose All shoppers to clear the simulated audience.
The picker appears when your store has saved audiences. It does not sign you in as a customer or change anyone’s audience membership. Collaborators in the same editor see the same audience selection.
If products disappear after choosing an audience, check your request transforms for matching product filters. The preview applies those filters while keeping the sort order you are editing, even if a transform would normally select a different sort.
Sort effect annotations
The preview panel shows annotations on each product so you can see exactly which sorting rules affected its position. Each annotation corresponds to an expression in your sort order configuration.
What you’ll see
- Promoted (green badge) — a priority rule promoted this product toward the top of results. Hover the badge to see which expression did it, for example Promoted by Priority: Vendor.
- Demoted (red badge) — a priority rule demoted this product toward the bottom of results, or a soft boost’s spacing seated it below its natural rank (see below). Hover the badge to see which priority rule did it, for example Demoted by Priority: Tags.
- Boosted badge — appears when a soft boost matched. Hover the badge to see the concrete effect:
Boosted +N positions when the product moved up in the ranking, Boosted +<lift> when the boost changed the score without changing rank, or Matched, no change when the match had no measurable effect. When more matches exist than Min positions between can fit, the extra matches read Boosted · over spacing capacity and the detail card shows Boosted cohort exceeded spacing capacity. Those products keep their incoming rank instead of dropping to the bottom. If a soft boost’s spacing layout places a matched product below its natural rank, the product shows a Demoted badge instead of a Boosted badge, and its scoring card labels the movement Moved down by spacing. A single soft boost never shows both badges.
- Sequence badge — appears when the product is placed by a product sequence. The tooltip reads Grouped by product sequence, which confirms the position came from a curated sequence group rather than a metric or rule.
- Diversity ↑ / Diversity ↓ badges — Diversity ↑ (tooltip: Surfaced for coverage) appears when a product was surfaced to broaden diversity coverage. Diversity ↓ (tooltip: Capped by family limit) appears when a product was pushed down because its product family already filled the diversity cap.
- Metric values — when your sort order includes a metric or numeric attribute, the actual value used for sorting appears next to each product. Date attributes (like
published_at) are rendered as human-readable dates instead of raw epoch timestamps. When the sort uses segmented metrics, values are tagged Segmented (…) so you can see which segment the number came from. Hover over the value to see which metric it belongs to.
All of these badges appear on the product image. Hover a badge to see the expression that caused it, so you can trace each annotation back to your sort order configuration. If a badge other than Tied would appear on every product in the loaded results, the preview hides it, because it doesn’t tell products apart.
Scoring details on hover
Open View ranking details on any product in the preview to see a scoring card that breaks down how its final score was assembled. For each matched soft boost the card shows:
- Base value — the score before the boost applied, formatted as a date when the underlying attribute is a date.
- Position change — the product’s rank before and after the boost, e.g.
#42 → #11 (+31). Movement is signed: a match that spacing seated lower is labeled Moved down by spacing and reads like #11 → #36 (−25) · 25 positions apart. A match that didn’t move reads #N (no change).
- Lift — the magnitude the boost added to the score. A negative lift shows with a minus sign, and a matched boost that didn’t change the score reads No score lift.
- Boost mode — appears when Layers switched a multiplicative boost to additive because the base metric’s values were too large for the decaying multiplier to lift anything. The row reads Switched to additive · base value too large for multiplicative decay.
- Score — the final score after the boost.
Use the scoring card to tell apart boosts that only shifted the score from boosts that actually moved the product up in the ranking. It also confirms which base value (global or segmented) the boost applied to.
When the sort order includes a weighted group, the scoring card also shows a per-feature blend breakdown. It renders a stacked contribution bar plus one row per feature with the raw value, normalized score, weight share, and contribution (normalized × share = contribution), followed by a Weighted total. Row colors match each feature’s slider tint in the editor. Layers scopes normalization to the current result set, so use the breakdown to compare contributions within one preview rather than absolute values across sort orders.
How to use annotations
- Verify priority rules: Confirm that the correct products show a “Promoted” or “Demoted” badge for each rule.
- Tune soft boosts: Check that soft boost badges appear on the expected products and compare the reported position change or lift across products to gauge the effect. Open View ranking details to see the base value, position change, and final score. If products you expect to boost show no position change or zero lift, raise Boost Strength in Amplify strong performers mode, or Percentile Target in Lift matches into view mode.
- Watch the spacing warning: If more products match a boost than Min positions between can fit in the loaded preview window, the editor shows a warning with the approximate number of matches that can be boosted at the current spacing. Lower the spacing if you want more of the cohort lifted.
- Check sequences: Confirm that products in a curated sequence show the Sequence badge and appear together in the expected order.
- Confirm diversity: Look for Diversity ↑ badges on products surfaced to widen coverage and Diversity ↓ badges on ones capped by their family limit, so you can see the diversity rule is working before you publish.
- Debug unexpected ordering: If a product appears higher or lower than expected, its annotations reveal which expressions influenced its position.
Hover over metric values and badges to see a tooltip with the full label of the sorting expression. Tooltips are especially helpful when your sort order has multiple metrics or rules.
