Simple GA4 Dashboard: Just the Numbers That Matter

A simple GA4 dashboard does one thing: it strips Google Analytics 4 down to the handful of numbers a publisher actually reads each morning. That means sessions, active users, engagement, your top pages, where traffic comes from, and now AI referrals. You do not need a new tracking script to get any of this. Every one of those numbers already exists inside your property. The problem is that GA4's default interface is built around an event data model and two report collections that make a quick read genuinely hard, which is why most site owners end up in the Explorations builder or an external dashboard just to answer basic questions. This guide covers exactly which numbers matter, how to surface them without adding any tags, and why GA4 still cannot show you AI traffic cleanly on its own.

Why is GA4 so hard to read at a glance?

GA4 is not broken, but it is built for a different job than the one most publishers have. By default it surfaces only two report collections, Lifecycle and User, and everything sits on an event-based data model that is hard for a non-technical site owner to interpret quickly. The moment you want a slightly deeper answer, GA4 pushes you into Explorations, its free-form, funnel and path template area, which most publishers find intimidating. This is well documented: GA4 ships with limited out-of-the-box customisation and steers you toward Explorations for anything beyond the basics (Coupler.io).

A simple dashboard fixes this by pre-selecting the load-bearing metrics so you never have to build them. For a publisher, those are:

• Sessions and active users, your raw audience size
• Engaged sessions and engagement rate, the honest read on quality
• Top pages by views, so you know what is actually carrying the site
• Session source and medium, so you know where that audience comes from
• AI referrals, now a category that deserves its own line

That is it. Everything else in GA4 is detail you can pull when you have a specific question, not something that belongs on a daily view.

Do I need to add a tracking script to get a simpler GA4 dashboard?

No. This is the single most important point, so it is worth being blunt. None of the visibility you gain from a simple dashboard requires a new tag on your site. A cleaner view of sessions, users, engagement, top pages and traffic sources is just a re-selection of metrics GA4 already records. The AI-referral improvements are also read-only reinterpretations of existing data: a channel grouping, and a source or medium pattern applied on top of what GA4 has already logged.

The only thing that genuinely sits outside GA4 is AI-crawler measurement, and that is not solved by a tag either. It is solved by reading your server logs, which every host already writes. So the honest positioning is this: no new script, just a cleaner read of your existing GA4, plus separate server-log analysis for the crawler question. If a product tells you it needs to install another JavaScript tag to simplify your dashboard, it is adding weight you do not need. Worse, layering a second analytics container on an ad-funded site can interfere with the measurement your ad network relies on, so a read-only approach is the safe one.

Ramprt is built on exactly this principle. It reads your existing GA4 read-only and gives you a clean, Plausible-style view plus an AI tab, with no new tracking script. You can try the live demo free at ramprt.io/demo to see the read before connecting anything.

Which two GA4 reports should I actually use?

If you only ever open two reports in the native interface, make them these.

1. Traffic acquisition (Reports, Acquisition, Traffic acquisition)

This is where you answer "where is my audience coming from?" Change the primary dimension from the default channel view to Session source / medium. That single change is what lets you see AI referrals in their raw form, entries like chatgpt.com / referral, perplexity.ai / referral and gemini.google.com / referral. GA4 files these under the Referral channel, not Organic Search, because they carry another site's domain as the referrer. If you stay on the default channel grouping you will miss this nuance entirely, which is exactly the digging a simple dashboard removes for you.

2. Pages and screens (under Engagement)

This answers "what is actually working?" Sort by views and engagement to see which articles carry your traffic and which hold attention. For a publisher living on ad revenue, page-level performance is the report that maps most directly to income, because impressions follow pageviews.

Between those two reports you have the spine of a publisher dashboard: where people come from, and what they read. Everything in Lifecycle and User beyond that is secondary. The reason a purpose-built dashboard still beats even these two reports is grouping and history, which the next section explains.

Why does GA4 not show AI traffic clearly?

There are four separate mechanisms working against you, and it helps to name them, because they are not configuration mistakes you can simply switch off.

1. AI referrals are filed as Referral, not Organic

When someone clicks through from ChatGPT, Perplexity or Gemini, they arrive carrying that tool's domain as the referrer. GA4 classifies this as Referral traffic. The catch is that it then sits mixed in with every other referring site, so unless you look at Session source / medium it does not read as "AI" at all.

2. The native AI Assistant channel is forward-only

GA4 now has a native fix for the grouping problem. On 13 May 2026 Google added an AI Assistant default channel: when GA4 detects a referrer matching a recognised AI assistant it assigns the medium value ai-assistant and buckets the session under the AI Assistant channel automatically, with no configuration required. The launch named ChatGPT, Gemini and Claude as examples (Search Engine Journal; Google Analytics Help). The problem is that this channel is forward-only. It does not reclassify historical data, so any AI trend from before 13 May 2026 stays buried in the Referral channel, and you cannot draw a clean year-on-year line from the native reports alone. A dashboard that back-fills and groups AI sources itself is more useful precisely because it can reconstruct that history from the raw Referral data.

3. GA4 materially undercounts AI referrals

Even going forward, GA4's raw numbers run low, for two reasons. First, "(not set)" bucketing: in a study of 181.6 million GA4 sessions across 22 clients over 12 months of 2025, roughly 22% of ChatGPT sessions and 32% of Perplexity sessions were dumped into the "(not set)" medium, while Claude and Gemini were attributed correctly (Workshop Digital). Second, AI traffic that arrives without a referrer header, common from in-app browsers, mobile apps and copy-pasted links, lands in Direct rather than being attributed to its source (Search Engine Journal). Because of both gaps, practitioners recommend adding a custom regex channel group, matching sources such as chatgpt, openai, perplexity, gemini, copilot, claude, grok and deepseek, placed above the Referral rule, to catch what the native channel misses (Authority Tech).

It is worth keeping perspective on scale. In that same Workshop Digital dataset, known AI sources peaked at just 1.1% of total organic traffic in July 2025 and settled around 0.3% to 0.4% by Q4 2025. AI referral traffic is real and growing, but for most publishers it is still a small slice of visits. The click-suppression story below is arguably the bigger revenue issue.

4. AI crawlers are completely invisible to GA4

This is a structural blind spot, not a setting. GA4 is JavaScript-based: the gtag tag only fires in a real browser. AI crawlers such as GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and ChatGPT-User fetch your HTML server-side and never execute the tag, so they generate no sessions and no events (AI+Automation). If you want to know how heavily AI is scraping your content, GA4 cannot tell you, ever. The only reliable method is parsing raw server access logs for those user-agent strings and inspecting the HTTP status codes they receive.

What about Google AI Overviews eating my clicks?

AI referrals and AI crawlers are only two-thirds of the story. The third, and for many publishers the most painful, is Google's own AI Overviews suppressing clicks before a visit ever happens. The Pew Research Center analysed 68,879 Google searches from 900 US adults during March 2025 and found that users clicked a traditional search-result link on only 8% of visits where an AI summary appeared, versus 15% where it did not. They clicked a link inside the AI summary itself on just 1% of visits (Pew Research Center). About 18% of searches produced an AI summary that month, and users ended their browsing session more often after a page with an AI summary (26%) than after one with only traditional results (16%). It is worth noting Google disputes that this reflects real-world traffic loss, so treat the figures as one strong dataset rather than the final word.

Here is the trap: none of this is visible in GA4 alone. The impression-with-no-click happens inside Google's search results page, so GA4, which only records sessions that actually land on your site, has no way to show it. To diagnose AI Overview click erosion you have to pair Google Search Console impression, click and position data with GA4 session data. When impressions hold flat or rise while clicks fall, that pattern points to AI Overview suppression rather than a ranking drop. A combined dashboard that puts Search Console and GA4 side by side is the only way to catch it without manual cross-referencing.

Is Looker Studio still worth it for publishers?

Looker Studio remains a legitimate option: it is free, it connects natively to GA4, and it lets you build charts you fully control. For a technical publisher who enjoys building and maintaining reports, it is a reasonable home for a simple dashboard.

But be clear-eyed about what it does not solve. Looker Studio pulls from the same GA4 data, so it inherits every gap described above. It will not back-fill the forward-only AI Assistant channel, it will not recover the ChatGPT and Perplexity sessions lost to "(not set)" or Direct, and it cannot show you crawler activity that never entered GA4 in the first place. To handle the AI questions in Looker Studio you would have to build custom channel groups, blend in a Search Console data source to catch AI Overview click erosion, and separately process server logs somewhere else entirely. That is a real project to build and an ongoing one to maintain.

So the honest answer is this. Looker Studio is worth it if you want a bespoke build and have the time. If you want the AI-aware read working out of the box, without wiring up channel-group regex, a Search Console blend and log parsing yourself, a purpose-built publisher dashboard gets you there faster. The decision is build-versus-buy, not free-versus-paid, because both can be free.

A worked example: what a clean publisher read looks like

Imagine a mid-sized content site checking its numbers for a month. In the native GA4 default view they see healthy Organic Search and a large, undifferentiated Referral bucket, and they move on. Nothing looks wrong.

Now apply the simple-dashboard lens. Switch Traffic acquisition to Session source / medium and the Referral bucket breaks open: chatgpt.com, perplexity.ai and gemini.google.com each appear as a distinct line. Add the custom regex group above Referral and the sessions GA4 had scattered into "(not set)" and Direct get pulled back where they belong, lifting the AI figure by a meaningful margin. The native AI Assistant channel confirms the forward trend from 13 May 2026, and a back-filled view reconstructs the months before it.

Then blend in Search Console. Impressions on several big evergreen articles are up year on year, but clicks are flat or down. That is not a ranking problem, it is the AI Overview suppression pattern, in line with the 8%-versus-15% click gap Pew measured. Finally, a glance at the server logs shows GPTBot and ClaudeBot hitting those same articles hundreds of times, traffic GA4 never recorded at all. Same underlying data, but now the publisher can see the whole picture: what AI sends them, what it takes from them, and what it is quietly suppressing.

If you want to see this read applied to your own property without setting any of it up by hand, the free demo at ramprt.io/demo shows the AI tab in action.

The bottom line

A simple GA4 dashboard is not about collecting more data. You already have almost everything you need. It is about selecting the six numbers that matter, grouping AI sources correctly, and pairing GA4 with the two things it structurally cannot see on its own: Search Console for AI Overview click erosion, and server logs for AI crawlers. Do that, and you replace a daily wrestling match with the Explorations builder with a read you can take in over a coffee. No new tag, just a cleaner view of what you were already recording.