Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Checklists

Dashboard Launch QA Checklist

A QA checklist for a marketing dashboard before it is shared: data sources and blends, totals and definitions, design, speed, access and governance.
Checklistintermediate7 min readUpdated

When to run this checklist

Run it before a new dashboard is shared beyond the people who built it, and again whenever a rebuilt version replaces an old one, a data source is swapped or reconnected, or a metric definition changes. A connector moved from one person's login to a shared account is a change; so is a blend that gains a new table. Either can alter every number on the page without a single chart looking different.

The builder runs the data sources and numbers groups, ideally with a second analyst checking the totals independently. The design group is best ticked with one member of the intended audience, who can say whether the first screen answers what they actually ask. This list is a one-off gate before launch. Producing the recurring monthly report from a finished dashboard is a procedure, covered by the monthly reporting pack SOP, and the standing rules for ownership, naming and retirement across all dashboards are set out in the dashboard governance guide. This checklist applies those rules to one dashboard on the day it launches.

Before you start

  • The brief: who the dashboard is for, which decisions it supports and which KPI figures those decisions depend on.
  • The agreed metric definitions, so names and formulas are checked against a reference rather than against each other.
  • Access to each source platform, such as GA4, the ad platforms and the CRM, to pull the same totals for the same period.
  • A known period with settled data, not the last few days, for reconciling totals.
  • The list of intended viewers and editors, by role.
  • The dashboard inventory, if the organisation keeps one, so the new entry can be added on the day.

The checklist

The groups are ordered from the foundations up. A wrong data source makes every later check meaningless, so the data sources group is finished before anyone reconciles a total, and the design group is only worth ticking once the numbers are right.

0/15

Data sources

Numbers

Design

Speed and access

Governance

Where each check is explained

The widget keeps each line short enough to tick; the reasoning sits in the guides. Each check links to the page that explains it.

CheckExplained in
Confirm each data source points at the right account, property and viewData sources and connectors
Check data source credentials belong to a shared owner rather than one personData sources and connectors
Check blends join on the right keys and do not duplicate rowsBlending data in Looker Studio
Match headline totals against the source platform for the same periodLooker Studio numbers don't match GA4
Check calculated fields against a hand calculationCalculated fields in Looker Studio
Use the agreed metric definitions and name them consistentlyMetrics layer and metric definitions
Explain known gaps between platforms on the page itselfWhy platforms and analytics disagree
Lead with the KPIs the audience makes decisions onDashboard design
Label every chart with its metric, date range and filtersDashboard design
Check the default date range and comparison periodDashboard design
Check pages load acceptably and caching suits how fresh the data must beLooker Studio performance and caching
Set viewer and editor permissions by roleSharing and embedding reports
Confirm viewers cannot see personal data they should not seeSharing and embedding reports
Name an owner and record the dashboard in the inventoryDashboard governance
Add a notes page with definitions, sources and the last review dateDashboard governance

If an item fails

A failure in the data sources or numbers group holds the launch. A headline total that does not reconcile with its source platform will be found by the first stakeholder who checks, and the explanation offered after the fact is never as convincing as one offered on the page. Fix the source or the formula, then rerun the whole numbers group, because a corrected blend changes every total built on it. The Looker Studio not matching GA4 guide walks through the usual causes, from date ranges and sampling to the way a connector aggregates metrics, and a chart that shows an error or no data at all is traced in the page on a Looker Studio chart not showing data. A dashboard that passed at launch but later keeps showing the same figures, or stops gaining its latest days, is traced layer by layer in the diagnosis for a Looker Studio report that is not updating.

Not every gap is an error. Ad platforms and analytics count differently by design, and some difference between them is expected; the fix there is a short note on the page explaining it, not a forced match. A permissions failure that exposes personal data is a different matter and should be fixed before anyone outside the build team opens the link.

Design, speed and governance failures can ship with a note when the launch date matters, provided they are logged with an owner and a date. A slow page is an irritation; a chart without a date range is a misunderstanding waiting to happen, so labelling is worth fixing first. A dashboard that leads with Vanity Metrics rather than the measures the audience decides on is better redesigned before launch than after, since people quickly learn to read the first screen and ignore the rest.

Illustrative example — not a real result

A physiotherapy group's bookings dashboard

Tarnfield Physio, a group of five physiotherapy clinics, rebuilds its marketing dashboard to show new patient bookings by channel. The numbers group fails on the first line: the dashboard shows about a fifth more bookings than the clinic system for the same month. A blend joins sessions to bookings on date and clinic but not on channel, so bookings repeat once for each channel row. The join key is corrected, and the total then sits close to the clinic system, with the remaining gap explained in a note about bookings made by phone. The data source credentials belong to the analyst who built it and is about to go on leave, so they are moved to a shared reporting account before the link goes to the clinic managers.

Common mistakes

  • Reconciling against the wrong period. The dashboard uses a rolling range, the platform export uses calendar months, and the mismatch is blamed on the data.
  • Credentials in one person's name. The analyst leaves, their access is removed, and every chart turns to an error message the following morning.
  • Blends nobody tested. The join looks right, the totals look plausible, and they are inflated by duplicated rows.
  • Metric names that drift. "Leads" on one page and "conversions" on another for the same number, or the same name for two different numbers.
  • Sharing by link to anyone. The link travels further than intended, and the page holding customer-level rows travels with it.
  • No notes page. By the next annual review, nobody remembers why a metric excludes one channel.

The monthly marketing report template gives the finished dashboard a written report to sit alongside, and the checklists library holds the rest.

Terms used on this page

Where this page is used

Also related

Sources

  1. Looker Studio Help — Looker Studio Help
  2. Digital Marketing and Measurement Model — Avinash Kaushik

Related