UGC marketplaces reporting dashboard: a blank design

What to take away

Section 0: the contract table

Section 0 is one table plus one rule. The table fixes what each figure means. The rule says nothing is plotted until that row is complete. F-UGC-01, used throughout as the fictional example, is a 20 second video commissioned through a UGC marketplace for one English brand unit. Every cell starts as { }.

This blank design is for reporting on UGC marketplace commissions, not a live dashboard or a review of marketplace providers. Link the marketplace order or brief and creator record to the accepted asset, rights window and buyer-side outcome so exceptions can be traced across the commission.

A missing or failed approval keeps the release on HOLD. The blank budget template shows how cost states and approval gates sit beside a single creator asset.

These are general-purpose BI and data-quality tools, not UGC marketplaces or evidence of marketplace integrations. Looker Studio, Power BI and Metabase can each publish a page that holds a table and no chart. dbt attaches lineage and tests to modelled tables, while Great Expectations writes data checks as code.

One provenance schema on every card

Evidence class must be one of observed operational, platform-reported, rule-attributed, associated, causally estimated, assumed or forecast. A calculated field keeps the source and version of each input. ONS quality dimensions raise useful questions about relevance, accuracy, timeliness and comparability, though a buyer dashboard is not an official statistic.

A blank cell is a question, not a result.

A starter metric dictionary

The metric dictionary row in Section 0 is where most UGC marketplace dashboards go wrong, because a marketplace export, an ad platform and the buyer's order table each count different things. These are the metrics a commission dashboard usually needs, each with the source it should come from and the evidence class it carries:

Each metric gets one row in the dictionary with its numerator, denominator, period and owner filled in before it is plotted. Whichever marketplace the commission runs through, whether a UGC platform such as Billo or Insense or an agency, map its order or delivery export to these rows once, record the mapping version, and remap when the export format changes.

A filled-in dictionary row, illustrative only, reads: metric, days from brief to accepted asset; numerator, calendar days between the brief timestamp and the signed acceptance; denominator, one accepted asset; period, assets accepted between 1 and 30 September; source, the marketplace order export and the acceptance log; owner, measurement lead; illustrative value for F-UGC-01, 12 days.

Illustrative England asset card

Every value below is illustrative and describes the fictional F-UGC-01. None of it is a measured result.

Open the first screen on the exception queue

Queues to hold from day one:

Every item needs {detected time}, {source}, {affected records}, {owner}, {containment}, {next evidence} and {closure proof}. A queue stays visible even when it holds no value.

An illustrative worked item:

For a UK advertising-claim exception, check the ASA/CAP Code; the recognition and rights checks in our advertising rules walkthrough test one advert rather than a whole campaign.

Publish corrections as versions

Never overwrite a page silently. Store refresh time, changed source, reason, author, affected cards and prior version. OSR transparent revisions and corrections covers revisions to official statistics; voluntary use here improves traceability without conferring official status.

Late records wait in a visible pending queue until the next authorised refresh or a documented restatement. An attributed order and an incremental order need separate labels, and the attribution methods note explains why the counterfactual limit belongs beside the number.

Record access and permission rules

The data owner lists allowed users, purpose, necessary fields, joins, exports, retention and deletion. The UK GDPR and Data Protection Act 2018 are the relevant UK data-protection framework for personal data here. ICO data protection by design guidance covers privacy across a system lifecycle and limits personal information to what is necessary. It does not select the lawful basis for this fictional dashboard.

Separate read, edit, approval, export and publishing permissions. Measurement access must not grant the ability to post.

Common questions

Which fields must be complete before a chart appears?

The decision and authorised reader, the commission and accepted asset identifier, the period and cut-off, the metric dictionary version, the source snapshots and the privacy purpose. Anything missing keeps that cell blank.

How is a missing or late value shown?

As a blank cell beside a visible pending queue entry, never as zero. The card also carries a blank decision threshold, so the reader knows what absence means.

Why can an attributed order not share a label with an incremental one?

Watching change does not by itself establish what caused it, so the two sit on different pages under different labels. Causal analysis appears only once a qualified evaluator approves a defensible design.

Who controls access and who controls publishing?

The data owner sets users, purpose, retention and deletion. Publishing rights sit with a separate person, and every export is logged.