Quality

Quality checks assets against rules you write about their metadata, such as "Product assets have a part number" or "Every photo has a credit line", and grades the library from A to E. Each rule says when an asset fails it, which assets it applies to, and whether it is a Must or a Should rule. Freedam checks every asset in the rule's scope, counts the assets that pass and fail, and lists the rules that fail most at the top of the page, with a link to those assets in the gallery.

The page is under Administration > Content Quality > Quality. It needs the Manage quality rules permission and a plan that includes Quality (Business and Tailored). Quality covers metadata rules and brand rules alike.

Every Quality page has the same tabs:

• Overview: the grade, the fixes, the AI brand checks and the rules that fail.
• Rule sets: metadata and brand rule sets in one list.
• Review: what waits for a decision, in two tabs: Suggested values and Brand checks. People with the Review quality results permission see it.
• Activity: what Quality checked and changed.
• Settings: the automatic fix user, the instructions sent with every AI brand check, and the daily allowance.

The overview

The page shows:

• Library grade: the letter, the score out of 100, and the share of the assets in scope that were checked. When a Must rule holds the letter down, the page names that rule.
• Rule sets: each active set with its own grade and its last run.
• Rules that fail: every metadata rule that some assets fail, wait for, or that could not be checked, and every brand rule that reviewed checks found failing (marked Brand rule), with Must rules first and the most failures first. View assets opens the gallery filtered on the assets that fail a metadata rule. A brand rule counts the latest reviewed check of each asset; a check a reviewer set to Compliant fails no rule.
• Fixes, last 7 days: how many values were written and how many suggestions wait for review, with links to the review and to the activity.
• AI brand checks: how many checks wait for review, how many assets were found not compliant, how many checks of the daily allowance were used today, and how many wait for tomorrow's allowance. Review checks opens Review > Brand checks and Brand settings opens Settings.
• Related tools: filename parsing rules, which fill fields from file names and also give Quality values to suggest, and the duplicate scan, which finds files uploaded more than once. Duplicates do not count in the grade.

Run now runs every active metadata rule set again. People with the Run quality checks permission see it.

Brand rule sets count in the grade too. See Brand Rules.

How the grade is computed

Each rule has a pass rate: the assets that pass it, out of the assets that pass or fail it. The score is the average of the pass rates, where a Must rule counts three times and a Should rule once. The letter follows the score:

Letter	Score
A	90 and above
B	75 to 89
C	60 to 74
D	40 to 59
E	below 40

Two rules keep a library from a good letter while an important rule fails often:

• A Must rule that passes on less than 95% of its assets holds the letter at B at best.
• A Must rule that passes on less than 80% holds it at C at best.

A grade is provisional while fewer than 90% of the assets in scope are checked, for example right after a rule set is saved. Without any checked asset there is no grade.

Rule sets

The Rule sets tab lists every set, metadata and brand, with its grade, status, result and how many assets it applies to. New rule set asks which kind: Metadata rules (fields assets must fill, checked on the whole library) or Brand rules (what images must look like, checked asset by asset). Both editors look the same: each rule is one line (Must or Should, the rule, how it is checked, its result), and selecting it opens it in a side panel.

A rule set has one of three statuses:

Status	What it does
Draft	Never runs. Use it while you write the rules.
Active	Runs when its rules or its scope change, when it becomes active, every night, and on the assets that change. Counts in the grade.
Disabled	Kept with its rules, but never runs and does not count in the grade.

Getting started

The first time an administrator opens Quality, Freedam creates a disabled set named Getting started from the requirements it finds:

• a Must rule for each required field, applied to the field's asset class, for example "SKU is filled"
• a Should rule for each field rule that makes a field required, applied when that field rule's conditions match, for example "Caption is filled when Published needs a caption"

Each rule fills its field, with no source chosen yet. A rule that an active set already has, with the same field and the same Applies when, is left out, and so is a field rule whose conditions cannot be checked. A requirement that a later field rule can make optional again is left out too, since the rule would fail assets the form accepts. Review the rules, choose sources for the fixes, then make the set active. The set is created once: deleting it does not bring it back.

Writing a metadata rule set

• Name, Description and Status.
• Applies to: the conditions that select the assets the whole set checks, with the same editor as gallery filters. The count under the editor shows how many assets match now. Without conditions, the set checks every asset.
• Rules: at most 50 per set.

Saving a set with changed rules creates a new version, listed under Versions, and runs the set on the library.

A rule

• Rule: the requirement in plain language. It is how the rule appears on the overview and in the gallery filter.
• Must: on for a Must rule, off for a Should rule.
• Fails when: the conditions under which an asset fails the rule, for example "Part number is empty".
• Applies when (optional): the rule checks only the assets of the set that match these conditions, for example "Asset class is Product".
• Depends on (optional): the rule checks only the assets that pass another rule. Assets that fail the other rule are held back: they are neither passed nor failed by this rule.
• Give time before failing (optional): an asset that fails the rule counts as waiting, not failing, for the given number of days, counted from a date field such as the creation date. Use it for information that arrives after the upload.

Under each rule, the editor shows how many assets the rule checks and how many fail it now.

Examples

A product catalog:

Rule	Fails when	Applies when	Other settings
Product assets have a part number (Must)	Part number is empty	Asset class is Product
Products with a part number name their device (Should)	Device name is empty	Asset class is Product	Depends on the part number rule
Products have care instructions (Should)	Care instructions is empty	Asset class is Product	30 days from the creation date

A photo library:

Rule	Fails when	Applies when
Photos credit their photographer (Must)	Credit is empty	File type is image
Advertising photos have a model release (Must)	Model release is empty	Usage is Advertising
Photos have keywords (Should)	Keywords is empty

A marketing team:

Rule	Fails when	Applies when
Campaign assets name their campaign (Must)	Campaign is empty	Collection is Campaigns
Agency assets say when their rights end (Should)	Rights end date is empty	Source is Agency

Rules on the import catalog

The condition Missing from the import catalog selects the assets an import source lists as unmatched: files whose key (for example the part number) has no row in the source's catalog, or that have no key. Use it in Fails when, for example "Missing from the import catalog is Products feed", to count those assets as failing.

• The rule shows Open the unmatched list on the overview and in the editor. It opens the source's Unmatched view, where you can fix the key or dismiss the asset.
• A dismissed key or asset counts as dismissed, not failing, and leaves the rule's score. Rules that depend on this rule hold those assets back.
• The rule sets that use the condition run again after each import of the source, and after each dismissal or undismissal.

Quality fields (the gallery filters on failing rules) cannot be used inside a rule, and rules cannot depend on each other in a circle. Saving refuses such a rule with a message on it.

Conditions from other tools

Besides metadata fields, a rule's conditions can read what other tools keep on each asset:

• Has an exact duplicate: another asset that is not in the trash has exactly the same file. Resized or re-saved copies are not exact duplicates; find those with the Duplicate Scan. When a copy is uploaded or deleted, the assets with the same file are checked again.
• Brand compliance status, Brand compliance review pending and Brand compliance reviewed: the status a reviewer approved for a brand check. When a reviewer settles a check, the asset is checked again. See Brand Rules.

Ready-made rules

A new rule offers ready-made rules above its text: Assets have no exact duplicate, and with brand compliance Assets are on brand (fails when the approved status is non-compliant or partially compliant) and Brand checks are reviewed. Choosing one fills the text and the Fails when conditions, which stay editable.

Fixes

A rule that fails when a field is empty can name that field as the field to fix, and the sources a value may come from:

• Filename parsing rules: the value your active filename parsing rules read from the file name.
• Import catalog: the value the import would write, for assets whose key field (for example the part number) matches a row of an import source's catalog, including assets uploaded outside the import.

Each source has a mode:

Mode	What happens
Apply automatically	The value is written as the automatic fix user (see Settings).
Ask me	The value is suggested on the Review page.
Off	The source is not used.

After each run, Freedam asks the sources in order for every failing asset and keeps the first value that:

• makes the rule pass
• makes no other rule of an active set fail, for example a value that brings an asset into the scope of a rule whose own field is empty

Every value is checked again just before it is written. If the field was changed since, or the catalog row or filename rule changed, nothing is written and the fix is marked so. A rejected value is not suggested again until its source changes. Fixes are not undone; each fix keeps the value it replaced.

Review

Review > Suggested values lists the suggested fixes, grouped by rule and source, on the assets you can see. Review > Brand checks lists the brand checks waiting for a reviewer (see Brand Rules). Each tab shows how many items wait in it. Approve writes a value as you, so it needs the right to edit the asset and the field; Reject drops it. In a group, Approve all on this page approves the suggestions shown. The page needs the Review quality results permission.

Activity

Activity is a timeline, newest first, of the metadata checks and, for each day, the brand checks it saw (how many were compliant, partially compliant, non-compliant, waiting or not checked, and how many wait for review, with See the checks). Metadata runs shows the metadata checks alone; Brand checks lists the brand checks one by one. Each check shows how many assets it checked with how many rules, its failures (one failure is one asset that fails one rule), and whether that is more or fewer than the previous check of the same set. Under a check, each rule it found values for says how many values were written, how many wait for review (with a Review button), how many were rejected, and how many were not used because the asset or the rule changed. See the values lists each value under what happened to it (waiting for review, written, not used, rejected), with its asset, the field, the value it replaced, where it was found, and who wrote or rejected it and when.

Settings

Settings holds the user automatic fixes are written as: an active user allowed to edit metadata. Without one, automatic fixes wait for review. It also holds the instructions sent with every AI brand check and the daily allowance (see Brand Rules).

Runs

A run checks every asset of a rule set and replaces the set's results at once, so the overview never mixes two runs. Sets run:

• when they are saved with changed rules, when they become active, or when their scope changes,
• every night,
• with Run now on the overview.

Between runs, the assets that are created, edited, archived or restored are checked again about a minute later, so the counts follow the library during the day.

A rule that cannot be checked, for example because a field it uses was deleted, shows "This rule could not be evaluated" on the overview. Its assets count as not checked, and the other rules of the set still run. While a set waits for its first run after a change, its rules show "Not checked yet" instead.

In the gallery

Three gallery filters read the results: Quality rule failed (the assets that fail a rule), Quality rule waiting (the assets that wait for a rule's deadline) and Fails a quality rule (the assets that fail any rule). View assets on the overview opens the first one.

Permissions

Permission	Gives access to
Manage quality rules	This page, the rule set list and the editor. With brand compliance, the brand rule sets and settings too.
Run quality checks	Run now on the overview, and Check again for brand compliance on the asset page.
Review quality results	The Review page: approving or rejecting suggested fixes (writing still needs the right to edit the asset). Approving brand check statuses, and the brand checks in Review and Activity.

The three permissions are given to the Administrator role; give them to other roles in Administration > Roles.