Skip to main content
WISEPIM Catalog Jobs review queue with run and attribute filters, High, Check and Unsure tabs, an Accept all High banner, keyboard hints and suggested attribute values with confidence scores WISEPIM Catalog Jobs review queue with run and attribute filters, High, Check and Unsure tabs, an Accept all High banner, keyboard hints and suggested attribute values with confidence scores Every Catalog Job ends in the review queue. Anything the job was not sure enough to apply on its own waits here: attribute values, categories, images, supplier codes, variant groups and more. Accept what looks right, change a value, or reject it. Nothing reaches your shop before you export. What this page enables: you can clear a pile of suggestions quickly by accepting the confident ones in bulk, spend your attention on the unsure ones, and take back a decision when you got it wrong. Where to find it: click Review now on the Catalog Jobs hub, or Review 32 suggestions (with your own count) on a run page. When you open it from a run, the header shows From: with the run name, so you only see that run’s suggestions.
The review queue only holds Catalog Jobs suggestions. AI suggestions from other sources, such as AI fill or web research, wait in the Review Inbox. Catalog Jobs suggestions show there too, marked From a Catalog Jobs run.

Confidence bands

Each suggestion carries a confidence score and falls into one of three bands: The threshold is the one you picked when you started the job (Careful, Balanced or Fast), or your project default. Click the score on a row to see what it means: “Out of 100 suggestions this sure, we expect about 94 to be right.”

Filter and sort

The toolbar narrows the list: Suggestions are grouped per field, with a count of what is still open in each group.

Accept all High

When there are High suggestions, a green banner says, for example, “17 High suggestions. We’re confident about these.” Click Accept all 17 to apply them in one go. The confirmation reminds you that you can still change any value on the product afterwards. To pick your own cut-off, click Choose a threshold. A small chart shows how your open suggestions spread over the confidence range. Pick a preset (Careful, Balanced, Fast or Recommended, which is based on what you accepted before) or drag to a value, then accept everything at or above it. Use the default goes back to your normal threshold.
One click accepts up to 2,000 suggestions. If there are more, click again after the first batch. Suggestions that fail to apply stay in the queue with the reason.

Decide on one suggestion

Each row shows the product, the field, where the value came from (for example “from the photo” or “from your rules”), the current value, the suggestion and its confidence.

Why we suggest this

The drawer explains the suggestion:
  • Why we think so: the reasoning, such as the value it matched or the rule it followed.
  • Sources: the photo, web page or supplier code behind it.
  • How sure we are: the score, and the Runner-up value it almost picked. A small gap is flagged as a close call.
  • Decided by: “our judge model”, “an AI model”, “a fixed rule, no AI” or “the source, copied as is”.

Image suggestions

Image rows show the current image next to the suggested one. Depending on the job, the actions are Replace placeholder, Replace image, Add to gallery, Add as main image or Remove: doesn’t match. Click Compare images to see them side by side.

Variant group suggestions

WISEPIM review queue filtered to variant groups, each row showing which products become one product with their SKUs and colour options WISEPIM review queue filtered to variant groups, each row showing which products become one product with their SKUs and colour options Turn look-alikes into variants suggests groups such as “4 products become one: Snap Retro”, with each member’s SKU and option value. Accepting a group creates one parent product with the members as its variants. See Product Variants. Click Edit group to fix a group before you accept it: “Untick a product that is a different model. At least two stay in the group.” You can also rename the new product.
A group with products that are already listed on your Shopify shop can’t be accepted yet. The row says so: “Can’t be accepted yet: some of these products are on your Shopify shop”.

Work in bulk

Tick rows, or use Select all 37 on this page (with your own count), and a dock appears: Bulk decisions are free. Large batches run in the background, so you can keep reviewing. One bulk decision covers up to 50,000 suggestions.

Keyboard shortcuts

The queue is built for the keyboard. Click Hide keyboard hints to hide the reminder bar.

Undo a decision

Undo on a row, or ⌘Z, takes back your last decision. After a bulk decision, the result message offers undo for the whole batch, for example “We took back 17 decisions”. The suggestions go back into the queue. You can undo attribute values, product details, categories and supplier codes. A value that changed on the product since you accepted it stays as it is. Photos can’t be undone from here: “17 photos stay as accepted. Change them on the product.”

When a suggestion can’t be applied

Sometimes an accept fails. The row stays in the queue with the reason:

Act on what you find

Start with the High tab and Accept all. Then sort the Check tab by Least sure first and work down with the keyboard. Leave Unsure for last: reject what looks wrong.
Open Why? to see the source. If a supplier code is misread, teach it in the supplier glossary. If the allowed values are too broad, tighten them on the attribute in Managing Product Attributes.
Press Undo on the row, or ⌘Z right away. For a bulk accept, use the undo in the result message.
Click Edit group, untick the product that doesn’t belong, and accept the rest. Or reject the whole group.

Catalog Jobs

Start jobs, set the confidence threshold and read run reports.

Review Inbox

The project-wide inbox for every AI suggestion.

Product Variants

What happens when you accept a variant group.

AI Review Settings

Decide which AI changes wait for approval.