Equivalents

The Match rules page, in the Configuration section of the sidebar, holds the criteria the Comparable Engine uses to select Equivalent Matches. Rules form a hierarchy that mirrors your product taxonomy. Set broad defaults at the global level, then tighten them at the Area, Section, or Product Group level as needed. The Equivalents page itself will show coverage and review status for those rules.

Tabs

The Equivalents page has two tabs. Both are placeholders until later releases populate them:

Tab	Status
Coverage Overview	Placeholder. Coverage metrics will move here. For now, see Market Benchmarks.
Needs Review	Placeholder. Once rule-driven evaluation runs, this tab will list candidates that match the required attributes but fall below the confidence threshold.

The Match rules page

The Match rules page opens to a tree. The tree lists every product category that contains products, nested as Area, then Section, then Product Group. Each row shows the confidence threshold in effect and whether its settings are custom or inherited. Categories with no products anywhere beneath them do not appear in the tree.

At the top of the Match rules page, the Global settings banner shows the current global confidence and price-closeness values. Click it to open the Global settings editor. The editor sets the fallback for any category that has no custom settings of its own and no ancestor with custom settings.

Filtering the tree

Use the toolbar above the tree to narrow the list:

Search. Filters by node name.
Level. Narrows to Areas, Sections, or Product Groups, which are the leaves. Parent rows stay visible above each match so you can see where it sits in the taxonomy.
Source. Narrows to nodes with Custom settings, or to nodes that are Inherited. Ancestor rows of a match stay visible for context.
Overrides only. Narrows to nodes that have custom selection settings or a custom attribute list. Ancestor rows of a match stay visible for context.

Without active filters, rows expand and collapse when you click the row or the chevron. While a filter is active, the rows above a match stay open.

Tree columns

Column	What it shows
Product Area / Group	The hierarchy node name, indented by level.
Settings source	Custom settings when the node has its own settings, or a label naming what it inherits, either an ancestor (Inherits from <name>) or Inherits Global settings.
Confidence	The effective minimum confidence threshold, either the node’s own value or the one it inherits.
Products	Total product count under this node across own and competitor catalogues.
Review	The number of matches awaiting review in this category, rolled up from everything beneath it. This counts candidates queued for review (those below the confidence threshold), plus any engine-generated matches marked stale by a regeneration run. The value is a link that jumps to those candidates in Market Benchmarks, pre-filtered to the category, or a dash when there are none.
Updated	When the node’s rule was last saved.

Every row’s action button reads Edit and opens the node rule editor, whether the node has custom settings or still inherits. An inheriting node opens in inherited mode; click Override settings inside the editor to switch it to custom values.

Editing a node rule

Click Edit on any row to open the node rule editor. Use the breadcrumb at the top to move up to any ancestor node without returning to the full list.

Selection settings

Every node, including the Global scope, can carry selection settings. These are a group of four controls that determine how the engine scores and routes candidates for this category.

Setting	What it does
Minimum confidence score	The threshold, from 0 to 100 %, that separates auto-selected candidates from those sent to the Needs Review queue. The engine auto-selects candidates that pass the required attribute gate and score at or above this value. It queues those that score below for review.
Price closeness	How strongly price proximity influences the confidence score for candidates that already pass the required attribute gate. Low keeps the engine’s baseline price influence. Medium strengthens it by 1.5×. High strengthens it by 2×. Price closeness never affects the required attribute gate itself. But it scales part of the confidence score, so raising it can change whether a candidate lands above or below the confidence threshold, as well as how candidates rank.
Allow out-of-stock candidates	Whether out-of-stock products are eligible as comparable candidates.
Use next best if selected equivalent is out of stock	Automatically falls back to the next-best candidate that passes the rule when the chosen equivalent goes out of stock.

Inheriting vs overriding

By default a node inherits selection settings from the nearest ancestor that has custom settings, then the Global rule, then built-in engine defaults. The subtitle above the settings controls, under the Selection settings heading, tells you where the current values come from. For example, it reads “Inherited from <ancestor> settings” or “Custom settings”.

Click Override settings to activate custom values for this node and for any descendants that do not have their own override.
Click Reset to inherited to remove the node’s custom settings and revert to the inherited values.

Attributes to match (Product Groups only)

On leaf-level nodes, the Product Groups, the editor also shows an Attributes to match section. These are the product attributes that a comparable candidate must share with your product to qualify. The engine excludes candidates that fail even one required attribute. It does not send them to Needs Review. It does not select them.

The available attributes are the attribute definitions that apply to this category, Aperture’s built-in definitions plus any you have configured in Attribution, that are marked for matching. Attributes with matching turned off do not appear in the picker. Each chip shows a live percentage. Of the potential matches your other selected attributes would still allow, it shows how many also carry a value for this one. The percentage recalculates every time you add or remove an attribute, and it drops off entirely, rather than reading 0 %, when your current selection leaves no potential matches at all.

Well-covered attributes appear first, with no heading. Attributes present on fewer than 25 % of those potential matches appear below, under a Low coverage heading. If you require a low-coverage attribute, few or no comparable candidates will qualify. Hover the info icon next to the Low coverage label for an explanation.

Unattributed outliers

A required attribute can exclude competitor products in this category because they carry no value for it. When this happens, a banner appears below the picker. It shows how many competitor products are excluded and what share of the category they make up. Click Show unattributed outliers to see the excluded products and the attribute each one is missing. Or use Export CSV to export the full list and follow up through your usual catalogue enrichment process.

Coverage warnings and save confirms

When you save a Product Group with custom settings or at least one required attribute selected, two situations prompt a confirmation dialog before saving:

No own products in this category. The Product Group contains only competitor products. A rule cannot generate any comparisons until you have your own products here. A warning badge also appears in the editor header.
A selected attribute has zero anchor coverage. None of your own products in this group carry a value for one or more of the attributes you have selected as required. The required gate will therefore fail for every candidate, and this category produces no Equivalent Matches until the data is populated.

In both cases the dialog lets you save anyway, for example when you author rules ahead of a catalogue enrichment.

Attributes are not inherited

Unlike selection settings, attributes are set only on the leaf node where you define them. They do not cascade to sibling or parent nodes, and ancestor nodes cannot contribute attributes to a leaf’s rule.

Viewing and changing settings from chat

The AI Assistant (Chat) can read and save selection settings for the Global scope or for a category you name. A view reports the effective settings, whether they are custom or inherited and from where, the engine defaults, and the category’s required attributes. A save needs all four selection settings and takes effect without an approval step, like saving on the Match rules page. The assistant reports the previous values so you can ask it to restore them. A save from chat changes only the four selection settings and keeps the category’s saved required attributes.

For a Product Group, chat can also list its unattributed outliers and report attribute coverage, either for the saved required attributes or for a set of attributes you ask about.

How the engine uses your rules

When the Comparable Engine evaluates a candidate pair:

Required attribute gate. All selected attributes must match between your product and the candidate. If any attribute fails, the engine does not select the candidate. It does not persist it and does not queue it.
Confidence threshold. If the required gate passes:
- If the score is at or above the minimum confidence, the engine auto-selects the candidate. It counts as an included Equivalent Match alongside manually accepted ones.
- If the score is below the minimum confidence, the candidate goes to Needs Review. It is queued as a suggestion for human review.

Price closeness scales the price component of the score compared in step 2. It can therefore change whether a gate-passing candidate is auto-selected or sent to Needs Review, as well as how candidates rank relative to each other.

The engine resolves the effective rule for a product by walking up the taxonomy from its Product Group. The nearest node with custom selection settings wins, whether that is the group itself or an ancestor. If no node in the lineage has custom settings, the Global rule applies. If no Global rule is set, the built-in engine defaults apply: 70% minimum confidence and medium price closeness. Attributes come only from the product’s own Product Group. They are never inherited from a parent node.

For a step-by-step walkthrough of setting up rules and reviewing the results, see Setting Up Equivalent-Product Benchmarking End-to-End.