Looker Studio / Data Studio marketing dashboards

Design the decision before the dashboard

A dashboard is useful when someone knows what decision to make from it. Emerge starts by identifying the audience and the operating question: which acquisition sources create accepted opportunities, which product categories lose customers during checkout, or which markets need a measurement investigation. Those questions determine the data model and report layout.

The deliverable is a maintained reporting service with agreed definitions, trustworthy source relationships and an owner. Google renamed Looker Studio to Data Studio in April 2026. This page retains the familiar Looker Studio search term; the product remains distinct from the Looker platform and its modelling tools. We scope the product actually required and do not imply that a reporting project includes a separate enterprise semantic platform by default.

Build a metric dictionary people can challenge

Every headline metric receives a definition, source, grain, date rule and owner. A lead may mean a submitted enquiry in one system and an accepted sales record in another. Revenue may be gross order value, recognised revenue or net value after refunds. The report names the chosen meaning rather than leaving readers to infer it.

Ratios need particular care. We document which numerator and denominator belong together and how filters affect them. Combining total campaign cost with only part of the corresponding lead population can produce a precise-looking but misleading cost-per-lead figure. The metric dictionary gives analysts and commercial users a shared place to resolve these questions.

Currency, time zone and reporting-period rules are recorded before sources are combined. Market comparisons retain enough context to distinguish a genuine performance change from exchange-rate treatment or a local calendar difference.

Join sources at a defensible grain

GA4, advertising exports, CRM records and order tables describe different entities and time periods. Emerge identifies the keys and aggregation level needed for each view. A campaign-day table should not be joined directly to many lead records in a way that repeats spend once per lead.

Where source preparation is modest, a direct connector may be sufficient. More complex joins, reusable calculations or business rules belong in a governed warehouse model, such as BigQuery, before the report reads them. This makes the calculation testable outside the dashboard and reduces the risk that several reports implement different versions of the same rule.

The model retains useful diagnostic fields: source system, last successful refresh, rejected-record counts and mapping completeness where relevant. These do not need to dominate the executive view, but operators should be able to investigate why a number changed.

Make the report usable in the operating meeting

Report pages are organised around decisions, with a concise overview followed by the detail required to investigate a change. Filters use terms the team understands and begin from a sensible default. A regional director and a campaign operator may need different views of the same underlying model.

We test the layout with realistic amounts of data, long labels and mobile or smaller-screen access where required. Charts use readable scales and labels, while tables retain the dimensions needed for action. The design avoids decorating a report with metrics that have no owner or decision attached.

A report should also communicate uncertainty. Incomplete CRM status updates, delayed exports or unclassified campaign names are visible quality conditions. Hiding those gaps can make the dashboard look polished while making the decision less reliable.

Treat credentials and refresh as part of the product

The data-source credential model determines whose access is used when a viewer opens a report. Emerge reviews the selected connector and credential options, the report audience and the underlying dataset permissions. Sharing a report must not accidentally expose a broader source than intended.

Refresh behaviour depends on the source, connector and caching configuration. We document the expected freshness and how a viewer can recognise stale data. A scheduled report delivery is not proof that the underlying source refreshed successfully, so operational checks inspect the data pipeline as well as the report’s availability.

Validate from source record to rendered metric

Acceptance uses sample transactions, leads and campaign records that can be followed through transformation and aggregation into the report. Tests cover missing joins, duplicate keys, currency changes and empty periods. We reconcile totals with the relevant source system and explain any intentional scope differences.

Users then complete the actual operating task: identify a source with poor lead quality, investigate a revenue change or locate an unassigned opportunity cohort. Their ability to reach an interpretable conclusion is part of acceptance alongside numerical correctness.

Handover includes the metric dictionary, data-source map, access roles and change process. Each report has an owner and a review date. The published retail analytics case provides related context for connecting GA4, warehouse analysis and Looker Studio; new dashboards receive their own definitions and validation evidence.