GA4 Dashboards: Build Client KPI Views Without Leaving Google Analytics

For years the answer to "the client wants a one-page KPI view" was Looker Studio. A second product, a connector to authorise, a separate set of sharing permissions, and one more thing that breaks when Google moves something. On 9 September 2026 Google put dashboards inside GA4 itself.

The release note is three sentences long, which is why most people will miss it. The official documentation fills in the mechanics. What it does not tell you is what to build, what to avoid, and where this leaves Looker Studio. That is what follows.

What Google shipped on 9 September

The GA4 changelog entry reads: "Dashboards are now available in Google Analytics. Dashboards are a highly flexible way to help businesses view their KPIs on a single report." It describes the new interface capabilities as drag-and-drop visualisation creation, manual chart placement, and new report visualisations.

The workflow sits under Reports. Click the create button at the top of the report left navigation, choose Dashboard, and you get a blank canvas. Save keeps your work. Publish adds the dashboard to the left navigation so anyone with access to the property can open it.

One detail matters more than the rest: publishing no longer requires a trip through the Library. In the older custom-report flow you built the report, then went to Library, then added it to a collection before anyone else could see it. For dashboards that middle step is gone. Build the page, publish it, done.

Access follows the property roles. You need Editor or Administrator to create and publish a dashboard. Anyone who can open the property can view whatever has been published.

The six card types

A dashboard is a grid of cards. Google supports six visualisation types, and each one has a job:

• Score cards show a single KPI, and they can display a percentage change when you apply a date comparison.
• Table charts show dimension rows against metrics, with shaded bar graphs and pagination for long lists.
• Line charts plot a metric over time, at daily, weekly, or monthly granularity.
• Bar charts compare values across dimensions, horizontal or vertical.
• Donut charts show how parts contribute to a whole.
• Funnel charts track the steps in a conversion path and the drop-off between them.

That set covers most of what a KPI page needs. Score cards carry the headline numbers. A line chart carries the trend. A funnel carries the conversion story. Tables carry the detail for whoever wants to dig.

The canvas runs on a grid, so cards snap into a layout and resize cleanly. You drag dimensions and metrics straight onto a chart instead of filling in a settings panel, and a right-hand panel edits whichever card is selected. Custom dimensions are supported, so the event parameters you set up in your taxonomy can appear as cards.

The limits come first, because they shape every decision

Read the limitations before you promise a client anything. Four of them change how you use the feature:

• Card limits. A standard GA4 property allows 15 cards per dashboard. A premium (360) property allows 30. Fifteen sounds generous until you try to fit acquisition, engagement, and revenue on one page and run out at the checkout funnel.
• No private dashboards. Every published dashboard is shared with the whole property. There is no per-user sharing, no share link, and no permission model beyond the property roles.
• No API access. You cannot create, read, or manage dashboards through the Admin API or the Data API. Everything is manual, which means a dashboard built for one client has to be rebuilt by hand for the next one.
• No segments and no card-level comparisons. Segments are not supported on dashboard cards, and card-level comparison is not supported either. The percentage change on a score card is the main comparison you get.

The sharing point deserves emphasis for anyone managing other people's properties. If you build a dashboard on a client property, that client's whole team sees it, along with every other agency, contractor, or freelancer who holds access. It is not a private workspace. Treat every published dashboard as a public artifact of that property, and put nothing on it that you would not want the client reading without context.

Where this leaves Looker Studio

GA4 dashboards do not replace Looker Studio, and Google is not pretending they do. The two tools answer different questions.

Looker Studio still wins on anything that involves more than GA4 data. Blending GA4 with Google Ads, Search Console, a CRM export, or a spreadsheet is routine there and impossible in a GA4 dashboard. Looker Studio also gives you scheduled delivery, PDF export, calculated fields, parameters, and full control over branding. If the output leaves your desk and lands in a client's inbox, Looker Studio remains the tool.

GA4 dashboards win on speed and on staying inside the property. There is no connector to authorise, no data freshness gap beyond normal GA4 processing, and no second product to check when a number looks wrong. For an internal view the account team opens every morning, or a quick way to stop tab-hopping between three reports, it is faster than anything Looker Studio offers.

The sensible split: GA4 dashboards for internal working views owned by the people who use GA4 daily, Looker Studio for client deliverables. Write that split down somewhere, because the alternative is two people building the same report in two tools and neither of them trusting the other's numbers.

How to build one without wasting the canvas

Fifteen cards is the constraint, so treat it like one. Start from the questions the page has to answer, not from the metrics you happen to have. A dashboard that shows everything answers nothing.

A workable rhythm:

1. Write down three questions the page has to answer. For an e-commerce client: how much did we make, where did it come from, and where does the checkout leak.
2. Pick the card type that answers each question most directly. Revenue is a score card with a comparison. Channel performance is a table or a bar chart. The checkout leak is a funnel.
3. Drag the cards onto the canvas in the order the reader will scan them. Headline numbers top left, trend beside them, detail below.
4. Apply a date comparison to the score cards, or the percentage change means nothing.
5. Publish only when the page reads cleanly. Once it is published, everyone with property access sees it.

Keep each dashboard to one theme. A property can hold several, and the left navigation is where they pile up. "E-commerce performance" and "Paid media" as two clean dashboards beat one crowded page every time.

Three client dashboards worth building first

Different client types need different cards. These three shapes cover most of the work.

E-commerce. Purchase revenue, transactions, average purchase value, session conversion rate, revenue by channel group, top products by revenue, and a funnel from view_item through to purchase. Seven or eight cards. The funnel is the reason to build the page, because it is the view people never bother to assemble by hand.

Lead generation and B2B services. Sessions, key events, form submissions by landing page, conversion rate by channel, and top landing pages with engagement rate. Add one caution card to the conversation: GA4 attributes by last non-direct click by default, so the table shows where your property credited the lead, not a modelled view of every touch. A dashboard does not fix an attribution model.

Local and small business. Sessions, calls and direction clicks marked as key events, top pages, branded against non-branded search, and the new versus returning split. Six cards is plenty. These clients open the page on a phone, so keep the top row to four score cards that fit a narrow screen.

What this changes in the reporting workflow

Dashboards add a third surface next to the standard reports and Explorations. Teams need a rule for which one owns what. Standard reports answer "what is going on" for anyone who opens GA4. Explorations answer a specific question once and are then abandoned. Dashboards hold the questions you ask every week.

Two habits keep it from drifting:

• Name dashboards by the decision they support, not the data they hold. "Weekly trading" tells you when to open it. "GA4 overview 2" does not.
• Record which dashboard is the source of truth for which number. When a client asks why the GA4 dashboard and the Looker Studio report disagree, the answer should be a known difference rather than a discovery.

The measurement plan is the right place for that note, next to the KPI definitions and the conversion windows. A dashboard without a documented owner becomes another stale page nobody trusts.

Consent mode still shapes what the cards show

Dashboards read the same data as every other GA4 report, so consent mode v2 applies to them. Where analytics_storage is denied for a share of visitors, part of what you see is modelled rather than observed, and a score card inherits that blending without saying so. When a headline number moves, check the modelled share before you credit marketing. The UK consent guide covers the plumbing.

What to do this week

1. Open each client property, go to Reports, and read the left navigation. If someone on the account already published a dashboard, find out before you build a second one.
2. Check who holds Editor or Administrator on the properties you manage. Any Editor can publish a property-wide dashboard. If that is wider access than you intended, tighten it now.
3. Build one dashboard for the client who asks the most reporting questions. Keep it to eight cards and one theme.
4. Write the internal-versus-client reporting rule into the measurement plan.

The tool is small. The value sits in deciding what deserves a card and who is allowed to publish it. Google put the canvas inside GA4. The judgement is still yours.