Looker Studio + GA4: build a client-ready dashboard in 2026

Looker Studio is free, connects to GA4 in two clicks, and can replace 90% of the reasons anyone logs into GA4 manually. The tool is not the problem. The problem is that most Looker Studio dashboards I see in client accounts are dense, chaotic, and unread after the second week.

I have been building analytics dashboards for clients since 2014. Across 120-plus containers I have converged on six charts. Not six favourites. Six that answer the questions CMOs actually ask, fit on one page, and still leave room for the date picker.

This is the step-by-step Looker Studio tutorial I wish existed when I started: GA4 connector in sixty seconds, six charts with copy-paste configs, three silent failure modes nobody warns you about, and the line where you stop blending data and move to BigQuery.

Why build the dashboard in Looker Studio, not in GA4

GA4 has its own Reports tab. It also has Explore. Both are reasonable. Neither is a good client dashboard.

The Reports tab is governed and fine for analyst peers. It lives inside GA4, which means the client needs a Google account with the right GA4 role, needs to know where to click, and needs to tolerate the GA4 interface. Most CMOs I work with will open GA4 exactly twice a quarter, realise they forgot which report they wanted, and close the tab.

Explore is a sandbox. It auto-saves for seven days and is great for ad-hoc questions. It is terrible as a standing report. Sharing is awkward, scheduling does not exist, and the sampling warnings appear in small grey text that nobody reads.

Looker Studio sits outside GA4. You can share a dashboard with a link, no login required for a view-only URL, and schedule a monthly PDF to the stakeholder inbox. It reads GA4 through a native connector and blends in Google Ads, Search Console, Sheets, and BigQuery without a plugin marketplace. It is also free, which matters more than it should.

The one case where Looker Studio is the wrong tool: compliance reporting where every number must survive an audit. Looker Studio caches, blends client-side, and shows sampled data without always flagging it. If you need numbers that will be defended in a regulator meeting, run SQL against BigQuery directly and export from there.

Connect GA4 to Looker Studio in sixty seconds

Open lookerstudio.google.com with the Google account that has at least Viewer access to the GA4 property. Click Create, then Data source, then pick the Google Analytics connector.

The connector will ask for an account, property, and data stream. Pick the GA4 property. Not a Universal Analytics property. UA stopped collecting data on 1 July 2023 and a UA connector will cheerfully return zero for every metric without any error message. I have debugged this three separate times for clients who insisted their dashboard was broken. It wasn't. It was connected to a dead property. If you want the official reference, Google maintains the GA4 connector documentation.

Once connected, Looker Studio pulls the full GA4 schema in about five seconds: default channel grouping, source / medium, session attributes, user attributes, every event, every event parameter, every custom dimension. You can now drag any of them onto a chart.

One gotcha: there are two similar connectors. "Google Analytics" is the native one you want. "Google Analytics (BigQuery)" is a separate connector that reads GA4 data after it has been exported to BigQuery. That one is faster and lets you query historical data beyond GA4's retention, but it only works if you already set up the BigQuery export. If you have not, see my GA4 + BigQuery setup guide for the wiring.

The six charts a CMO actually reads

The temptation with Looker Studio is to show everything. Resist it. The dashboards I have inherited from agencies typically have 20 charts spread across three tabs. They look thorough. Nobody opens them after month two. A six-chart dashboard with the right six charts gets opened every Monday morning.

Here is the layout I have converged on. Page one, top to bottom.

1. Scorecard row: users, sessions, conversions, revenue

Four scorecards across the top. Last 30 days vs previous 30 days, percentage delta visible. Active users, sessions, key events (or conversions if you still use that label), and total revenue if the property has ecommerce tracking.

Dimension: none. Metric: each scorecard gets one. Comparison: Previous period.

This is the first thing the CMO reads. If all four are green, the month is fine. If revenue is red while sessions are green, something in the funnel changed. If sessions are red while revenue is flat, traffic quality shifted. You can answer half the opening questions of a quarterly review from this row alone.

2. Time series: sessions by default channel grouping, last 90 days

One line chart. X axis is date. Y axis is sessions. Broken down by Default channel grouping. Seven lines: Organic Search, Paid Search, Direct, Organic Social, Paid Social, Email, Referral.

Chart type: Time series. Dimension: Date. Breakdown dimension: Default channel group. Metric: Sessions. Date range: Last 90 days.

Ninety days is the sweet spot. Thirty days is too narrow to see trends, and twelve months looks busy. Ninety days shows a campaign shift without overwhelming the eye.

3. Table: top ten landing pages by conversions

A ten-row table. Landing page, sessions, conversions, conversion rate. Sorted by conversions, not sessions. This is the chart that separates an analyst dashboard from a client dashboard.

Chart type: Table. Dimension: Landing page + query string. Metrics: Sessions, Key events (sum), Conversion rate (sessions) calculated as key_events / sessions. Sort: Key events, descending. Row limit: 10.

Sorting by sessions surfaces the homepage and whatever vanity URL has the most traffic. Sorting by conversions surfaces the pages that actually make the business money. These are almost never the same list. A well-trafficked homepage is often a poor converter; a thin content page with narrow-intent traffic often punches far above its weight.

4. Bar chart: conversions by source / medium

A horizontal bar chart, ten rows, showing which source / medium combinations actually drive key events. This is where attribution conversations start.

Chart type: Bar chart (horizontal). Dimension: Session source / medium. Metric: Key events. Sort: Key events descending. Row limit: 10.

If (direct)/(none) is the top row and you are spending on paid media, you have an attribution problem (usually Consent Mode, usually sitewide, usually fixable). See my Consent Mode v2 guide for the diagnostic.

5. Pie or donut: device category share

A three-slice donut. Desktop, mobile, tablet. One sentence above it: "What device your buyers use."

Chart type: Donut. Dimension: Device category. Metric: Users. Date range: Last 30 days.

Pie charts are unfashionable. I keep this one because it answers a question executives ask repeatedly: "are our customers on mobile?" The answer is almost always yes, and the chart reminds them.

6. Scorecard: average engagement time per session

Not a KPI. A sanity check. If this number suddenly halves, something in the tracking broke. If it suddenly doubles, something in the tracking broke in the other direction. I put it at the bottom right, deliberately out of the way.

Chart type: Scorecard. Metric: Average engagement time per session. Comparison: Previous period.

The optional seventh and eighth charts, if the client has the traffic to justify them: a revenue-by-country map if they sell internationally, or a funnel chart if the commerce flow is short enough for GA4 to track cleanly. Anything beyond that starts to double up and the dashboard loses focus.

Three things that silently break every Looker Studio + GA4 dashboard

Looker Studio will show you numbers. It will not always tell you when those numbers are wrong. There are three failure modes I now check on every new dashboard before I hand it over.

Sampling above roughly 10M events per date range

GA4 reports can sample when a query covers more than about 10 million events for the selected date range. When sampling is active, the numbers are extrapolated from a subset and can swing by double digits between refreshes. Looker Studio does show a little blue "reduced data" icon when this is happening, but it is easy to miss. Google documents the threshold in the Looker Studio data freshness reference.

The fix is not clever. You either narrow the date range, or you move the data source to the "Google Analytics (BigQuery)" connector, which reads the full export and does not sample. For a property doing 3M events a month, sampling will not kick in on a 90-day dashboard. For a property doing 50M events a month, it will. Check the dashboard after the first render and again after the first real data spike.

The (other) row, or why source / medium suddenly looks wrong

GA4 caps cardinality on high-cardinality dimensions. When the cap is hit, the overflow is bucketed into a row labelled (other). On a source / medium chart this can eat 30 to 50 per cent of traffic. The dashboard still shows numbers. The numbers are just collapsed into a bin called (other) and anything useful about attribution is lost.

A Polish B2B client of mine had 40 per cent of their traffic in (other) and nobody on the team had noticed for three months. Three months of channel attribution, gone. The fix is not at the Looker Studio layer. It is at the GA4 layer: clean up your UTM taxonomy so the cardinality stops blowing up.

If (other) is meaningful on a chart you care about, move that chart to a BigQuery source, where the full cardinality is preserved.

Cache freshness

Looker Studio caches report data aggressively. Default cache is twelve hours, which is usually fine for a dashboard read once a week. It is a problem when you make a GA4 change at 9am, look at the dashboard at 10am, and see yesterday's data.

There is a manual refresh button (three-dot menu, "Refresh data"). There is also a data freshness setting per data source (Resource menu, "Manage added data sources"). For dashboards that need near-real-time numbers, set freshness to 15 minutes. Everything else, leave it at twelve hours and get some sleep.

Blended data, and when to stop blending and use BigQuery

Looker Studio lets you blend up to five data sources into one chart. Blending is how you get a chart that shows GA4 sessions next to Google Ads cost next to Search Console clicks in the same grid. For some common questions, it works well.

Use blended data for:

• GA4 + Google Ads, joined on Session campaign, to reconcile conversions and cost in one view. This is the single highest-value blend I build for clients.
• GA4 + Search Console, joined on Landing page, to cross-reference organic clicks with on-page engagement.
• GA4 + Sheets, for a manual CRM merge when the client wants deal-value attribution but has no CRM connector. This is fragile and I treat it as a temporary bridge until BigQuery.

Stop blending and move to BigQuery when:

• The dashboard needs historical data beyond GA4's 14-month default retention. GA4's retention is short by analytics standards. BigQuery exports keep everything.
• You need cohort analysis. Looker Studio cannot do meaningful cohort logic on blended data. BigQuery can, in roughly 30 lines of SQL.
• You need custom attribution. Looker Studio can apply simple lookback windows. Anything non-trivial (position-based, data-driven, custom weighting) needs to happen in BigQuery first, then Looker Studio reads the output.
• The blend takes longer than five seconds to render. Large blends slow dashboards to a crawl. A Krakow SaaS client of mine had a Looker Studio dashboard blending GA4 + Sheets + Ads + Search Console. It took 30 seconds to open. We moved the four key charts to BigQuery-backed tables, and the dashboard now renders in 1.5 seconds with more historical depth.

Make the dashboard self-explanatory

A client dashboard needs to be readable without the analyst in the room. This is not a soft rule; it is the only rule that separates a dashboard from a vanity screen.

A few things that help:

• Every chart title is a question, not a label. "Which channels drove conversions this month?" instead of "Conversions by channel". The question forces you to think about what the chart is for.
• Date range control is visible at the top. Not hidden behind a menu.
• The first row of scorecards gets the company's most-asked numbers. If the CEO asks about revenue, revenue goes top-left. If nobody asks about bounce rate, bounce rate is not on the dashboard.
• Annotations for campaign launches, tracking changes, and outages. Looker Studio supports annotations on time series charts. Use them. "Black Friday campaign starts here." "Consent Mode v2 deployed here." Future-you will thank present-you.
• Sampled charts get a one-line caveat in the chart description: "Above 10M events, numbers are extrapolated." Honesty in a caption beats silence.

And one opinion: bounce rate is not a KPI. Average session duration is not a KPI. If your dashboard leads with either, it is not a dashboard. It is a screen to make the analyst look busy.

Schedule delivery, or the dashboard dies

A dashboard that nobody opens is a dashboard that does not exist. The single highest-value move, once the dashboard is built, is to schedule a monthly PDF export.

In Looker Studio: Share, Schedule email delivery, PDF attachment, monthly on the first working day, send to the stakeholder list. Pair the PDF with a one-paragraph written commentary from whoever owns the number. The commentary is what turns a dashboard into a living document. The dashboard does not explain itself. Numbers without interpretation are not insight.

This is how I run monthly monitoring for GA4 clients. Automated dashboard, written commentary, one email, no call. That is the whole product of monthly GA4 monitoring at €150 a month. If your own tracking is likely contaminated before any of this matters, start with a free GTM audit first.

FAQ

Is Looker Studio free?

Yes. The consumer version is free with no user limit, no credit card, and no usage cap that anyone reasonable will hit. There is also Looker Studio Pro (a paid tier with team assets, service-level guarantees, and admin features) but most small and mid-sized businesses do not need it.

What is the difference between Looker Studio and Looker?

Looker (the one without "Studio") is Google Cloud's enterprise BI platform, a separate product with its own modeling language (LookML) and a paid licence. Looker Studio is the free dashboarding tool that used to be called Data Studio. They share a name, they share almost nothing else. For a GA4 dashboard, you want Looker Studio.

Can I use Looker Studio without a GA4 property?

Yes. Looker Studio connects to 1,000-plus data sources including BigQuery, Google Sheets, Google Ads, Search Console, and a long tail of third-party connectors. GA4 is one option among many. If you only have BigQuery data, skip the GA4 connector and go straight to BigQuery.

How do I fix (other) showing up everywhere in source / medium?

(other) is a GA4-level cardinality cap, not a Looker Studio bug. You cannot fix it in the dashboard. You fix it in the data by cleaning up UTM tagging (stop auto-generating unique UTMs per session), or by moving the chart to a BigQuery source which preserves full cardinality.

Why don't my Looker Studio numbers match GA4 directly?

Usually sampling, cache freshness, or a dimension scope mismatch. Check the three in that order. If all three look clean and numbers still disagree, the cause is almost always a blended data source where one side has a filter the other does not.

Do I need Looker Studio Pro?

Probably not. Looker Studio Pro adds team workspaces, dashboard-level permissions, delivery SLAs, and Google Cloud admin integration. Worth it for agencies and large enterprises. Overkill for most SMBs and for every solo consultant I know.

Next step

If your GA4 numbers are off, no dashboard will save you. The fix is upstream. Start with a free GTM audit to see whether your tag firing, consent wiring, and event schema are actually healthy. Once the data is clean, the dashboard builds itself in an afternoon.

If you already know the data is clean and you want someone to build, maintain, and monitor the Looker Studio + GA4 setup monthly, GA4 Monitoring is €150 a month. Written report, no calls, no dashboards of mine for you to log into.