Setting Up Equivalent-Product Benchmarking End-to-End

Equivalent Matches are assigned automatically once a competing product passes the active criteria for its category. This walkthrough covers the prerequisites, rule authoring, candidate generation, candidate review, and finally where results surface across the product.

Before you begin

Rule authoring lives on the Match rules page, in the Configuration section of the sidebar. The Equivalents section in the main navigation is a separate area reserved for coverage and review metrics. See the note at the end of this guide. Both appear only if equivalent-product benchmarking is enabled for your workspace.

The attribute criteria available in each Product Group are the attributes defined for that category that are used for matching. Attributes you add through Attribution are used for matching by default. If an attribute you expect is missing from a Product Group’s list, contact us. Categories without any attributes used for matching can still use confidence-threshold and price-closeness settings; they just won’t have attribute-match criteria to select.

How Equivalence Rules work

Each product group, a taxonomy leaf, operates under one active rule. That rule specifies:

which attributes a competing product must share with yours for the pair to qualify
a minimum confidence score, the threshold above which a qualifying pair is selected automatically
how heavily price proximity contributes to a pair’s confidence score, which affects both whether it clears the confidence threshold and how qualifying pairs rank against each other

Rules inherit downward through the taxonomy. A setting applied to a Product Area flows to every Section and Product Group beneath it unless a more-specific rule overrides it. The Global settings are the fallback when no category-specific rule exists.

Step 1: Configure Global settings

The Global settings define the baseline applied to every Product Group that has no custom rule of its own.

Go to Match rules, in the Configuration section of the sidebar.
Click the Global settings banner at the top of the page.

Adjust the four controls as needed:

Setting	What it does
Minimum confidence score	Pairs scoring at or above this threshold are auto-selected as Equivalent Matches. Pairs below it go to Needs Review. Default is 70 %.
Price closeness	Low / Medium / High. Scales how much price proximity contributes to a candidate’s confidence score. Low keeps the engine’s baseline price weight, Medium, the default, adds 50%, and High doubles it. This can move a pair from Needs Review to Auto-selected, and can change its rank among other qualifying pairs.
Allow out-of-stock candidates	Off by default. Saved with the rule. The engine does not yet use this setting to filter candidates.
Use next best if selected equivalent is out of stock	On by default. Saved with the rule. The engine does not yet use this setting to promote replacement candidates.

Click Save settings. A saved rule applies the next time comparable suggestions are generated (see Step 4); saving does not re-evaluate existing candidates by itself. The editor notes that changes apply after the next overnight data refresh, but a data refresh doesn’t generate suggestions by itself; ask us to run a suggestion pass.

Step 2: Browse the rules tree

The Match rules page shows your product taxonomy as a collapsible tree, but shows only the part that currently contains products. Rows are categories with products filed under them, plus their ancestor categories. Categories with no products aren’t listed. Each row is annotated with:

Settings source. A badge showing whether the node uses its own custom settings or inherits from an ancestor. The badge names the source, for example “Inherits Global settings”.
Confidence. The effective minimum confidence score in force for that node, custom or inherited.
Products. Total products filed under the node.
Review. A badge showing how many candidates in that node’s subtree are awaiting review, suggested or stale, rolled up from every Comparison Set anchored underneath it. Click it to jump straight to Market Benchmarks filtered to that category with the Needs Review filter already applied. Nodes with nothing awaiting review show ”—”.
Updated. When the node’s own rule, its custom settings and/or its attribute selections, was last saved. Nodes with no rule of their own show ”—”.

Use the toolbar to narrow the view:

Search by name. Filters nodes whose names contain the typed text; matching ancestors are also shown to preserve context.
Level. Shows rows at the chosen level, Area, Section, or Product Group, together with their ancestor rows for context.
Source. Shows rows with the chosen settings source, Custom or Inherited, together with their ancestor rows for context.
Overrides only. Shows rows that have custom settings or custom attribute selections, together with their ancestor rows for context.

With no filters applied, click any parent row (or its chevron) to expand or collapse it. While a search or filter is active, matching rows’ ancestors are held open and cannot be collapsed until you clear the filters.

Step 3: Override a Product Group’s rule

To set rules at a specific level, click Edit at the right of any row. The editor that opens depends on the level you select.

Non-leaf nodes (Areas and Sections)

Non-leaf nodes let you override selection settings for the whole subtree without selecting attributes. Attributes are a Product Group-only feature. Click Override settings to make the node’s settings independent of its parent, adjust the four controls described in Step 1, then click Save settings.

To remove a custom override and revert to inheriting, click Reset to inherited, then click Save settings (or Save changes on a Product Group). Nothing changes until you save.

Leaf nodes (Product Groups)

Product Group editors contain two sections within the Equivalent criteria card.

Selection settings. The same four controls as Global settings, with an “Override settings / Reset to inherited” toggle.

Attributes to match. A set of chips listing the attributes defined for this category that are used for matching. Attributes not used for matching aren’t offered here. Each chip’s percentage is live and selection-aware. It is the share of your potential matches, pairs between your own products and competitor products that already satisfy your other selected attributes, that also carry this attribute, and it recalculates every time you select or deselect a chip. The chips shown by default keep a healthy share of your potential matches. A separate Low coverage section appears below them for attributes whose live share falls under 25%; selecting these as required criteria may silently eliminate most or all candidate pairs, so prefer the chips shown above.

Selecting an attribute makes it a required criterion. A candidate must share that attribute’s value with your product to qualify at all. Selected attributes are a pass/fail gate only; they don’t add to a pair’s confidence score. That score comes from category match, product-name similarity, price proximity, and size compatibility, with price closeness scaling the price part of it. The outcome logic is:

All selected attributes match and confidence ≥ threshold → Auto-selected as Equivalent Match
All selected attributes match but confidence < threshold → Needs Review, suggested
Any selected attribute does not match → Not selected, the pair is dropped and not persisted

Click Save changes. A saved rule applies the next time comparable suggestions are generated (see Step 4); saving does not re-evaluate existing candidates by itself.

Step 4: Generate comparable suggestions

Comparable Candidates are the pool of pairs the engine scores and classifies. Our team runs the suggestion generator, which is an internal-only control on the Market Benchmarks page that you don’t see in your account. Ask us to run a suggestion pass, for example after changing rules. This is what happens in a run:

Our team clicks Generate Suggestions on Market Benchmarks.
They click Dry Run first. The generator compares your products against competing products in the same category, scores each pair using the active Equivalence Rule for that product group, and skips pairs that:
- look like an Exact Match, meaning the two products share a canonical ID, GTIN, or MPN, regardless of whether that Exact Match has actually been recorded
- already have a decision the generator doesn’t own: either a manual decision, which means one that is accepted, rejected, or manually added, or else a candidate created by another source such as clustering or import
Existing generator-created suggestions, whether suggested, auto-selected, or stale, are not skipped this way. They’re re-scored every run and either refreshed, counted as Would update, or, if they no longer rank among the product’s top matches, retired, counted as Would retire.
The dry-run results show:
- Would create. New candidates the run would write, auto-selected plus needs-review.
- Auto-selected. Newly created pairs whose confidence met or exceeded the threshold; candidates that refresh an existing generated suggestion are counted under Would update instead.
- Would update. Existing suggestions that still rank among their product’s top matches and would be refreshed with a current score, status, and reason.
- Would retire. Previously generated suggestions that no longer rank in their product’s top matches and would be removed. A suggestion drops out of the top matches when a better match displaces it or when it now fails the rule; manually decided candidates are never retired.
- Skipped (required gate). Pairs dropped because a required attribute did not match.
- Total Candidates / Anchors Evaluated / Anchors with Candidates / Skipped Existing / Skipped Exact. Bookkeeping counts for the run.
- A per-category breakdown table and a preview of up to 25 candidate pairs with their confidence scores and reasons
If the preview looks correct, our team clicks Create N Suggestions. The generator writes Comparable Candidates to the database, refreshing existing generated suggestions that still rank and retiring generated suggestions that no longer do. Your own manual decisions are never touched. Auto-selected candidates enter the included pool immediately. The top-ranked one for each competing retailer becomes that retailer’s Equivalent Match, and any further auto-selected candidates for the same retailer count as Approved Market Comparables instead. Needs-review candidates appear in the Market Benchmarks Needs Review count. The Market Benchmarks table may take a moment to reflect the new totals.

Step 5: Review comparable candidates

The Market Benchmarks page lists every Comparison Set in your workspace, one per product you’re benchmarking. Each row shows the set’s selected Equivalent Match, or how many retailers have one, its count of included candidates, and how many candidates need review.

The four stat cards at the top provide a workspace-level summary:

Card	What it counts
Products with Benchmarks	Your products that have at least one Comparable Candidate in any state
Equivalent Matches	The number of selected Equivalent Matches across all Comparison Sets, up to one per competing retailer within each set. A single product with matches at three retailers contributes three to this count, not one
Approved Market Comparables	Included candidates, accepted, auto-selected, or manually added, that are not the top-ranked candidate for their retailer within a Comparison Set
Needs Review	Comparable Candidates across the workspace in the suggested or stale state, awaiting a decision

Use the search box to filter Comparison Sets by product name or brand.

Clicking anywhere on a row opens that product’s Matches tab on the Product Detail page. To review pending candidates, click the Review button with the count in the row’s Needs Review column. That is the only way to open the Review Comparable Candidates dialog. Work through the suggested pairs, accepting or rejecting each one. Switch the status filter in the dialog to see stale candidates too. Stale candidates are suggested, accepted, or auto-selected candidates that were not added manually and whose stored rule version differs from the version recorded on their Comparison Set. They become stale the next time the suggestion generator writes candidates for that product. Manually added candidates never become stale. The version is an internal marker on the Comparison Set itself, separate from the category’s Equivalence Rule version, so editing an Equivalence Rule does not by itself make existing candidates stale. Stale candidates show a Reaccept action and are included in the Needs Review count. When you close the dialog after making changes, the overview refreshes automatically.

Per-product view

Every product also has its own view of Equivalent Matches, on its Product Detail page.

Equivalents tab

The Equivalents tab lists one row per competitor configured for your workspace, showing whether that retailer already has an Exact Match, a selected Equivalent Match, or neither, along with the matched product’s name, matched attributes, current price, availability, a quality indicator based on the candidate’s confidence score, a status badge showing how the match was established (Auto-selected, Manual, or Selected), and when it was last updated. A summary line at the top counts total retailers, exact matches, Equivalents, rows still awaiting review, and rows with no equivalent yet.

Click Manage Equivalents, above the table, to open the Matches tab for the whole product. Within the table, the row-level Action column offers a narrower action per retailer: rows with a selected Equivalent Match show Manage, which also opens the Matches tab; rows with unreviewed candidates show both Review, which also opens the Matches tab, and Add; rows with neither an Exact Match, a selected Equivalent Match, nor any candidate awaiting review show Add alone; clicking Add opens the add-relationship wizard directly for that retailer. Retailers already covered by an Exact Match show no action in this column.

Matches tab

The Matches tab lists, for each competing retailer, up to three ranked Comparable Candidates with their price, confidence score, and matched attributes. From here you can:

Select a candidate as that retailer’s Equivalent Match. Replacing an already-selected match asks you to confirm.
Remove a candidate from the comparable pool entirely
Report a candidate as incorrect

If a retailer has no Comparable Candidate, for example because the automatic generator hasn’t produced one yet, use the add-relationship wizard from the Equivalents tab to search for the right product and create the relationship manually.

Where results appear

Once a product has a selected Equivalent Match, it feeds into any surface that offers the Exact only / Exact + Equivalent toggle. By default these surfaces show Exact Matches only; switching the toggle to Exact + Equivalent folds Equivalent Matches into what’s displayed.

On workspace-level surfaces, the toggle is disabled until your workspace has at least one Equivalent Match, with a tooltip pointing you to Market Benchmarks to set one up:

Dashboard. Coverage, price position, availability, and the other dashboard statistics.
Prices. The price matrix.
Competitors & Retailers. The Competitors/Retailers pages and their Common Products drill-down.

On a product’s own Product Detail page, the toggle unlocks earlier, as soon as that product has at least one Comparable Candidate at a competitor without an Exact Match, even before you’ve selected any of them as an Equivalent Match. If none is available yet, the toggle stays disabled with a note that there are no equivalents for that product yet.

A selected Equivalent Match appears on the product’s Equivalents and Matches tabs without needing the toggle. At a retailer that also has an Exact Match, the Exact Match takes over the row on the Equivalents tab and the equivalent is not shown. On the Matches tab, that retailer shows no candidates and reads “No product match under the configured match rules”. Market Benchmarks reflects a selected Equivalent Match directly:

Market Benchmarks. The stat cards and Comparison Set table reflect the current counts of Equivalent Matches and Approved Market Comparables.

The Coverage Overview and Needs Review tabs on the Equivalents page are reserved for future in-page coverage metrics and candidate review; in the meantime the Match rules page’s Review badges and the Market Benchmarks stat cards and its Review Comparable Candidates dialog are the primary surfaces for coverage and review.