Build an AI Search Traffic Dashboard in Looker Studio (Free 2026 Guide)

GA4 hides most AI search traffic inside Direct and Referral. A free Looker Studio dashboard fixes that in about 15 minutes. This guide builds one from scratch on a real B2B SaaS property, with a screenshot for every step, the exact AI source regex, and a calculated field that splits sessions by engine.

The measurement gap is large. Loamly’s benchmark of 446,405 visits found that 70.6 percent of AI visits arrive with no referrer header. GA4 files those sessions under Direct. Meanwhile Microsoft Clarity’s study of 1,200 publisher sites measured AI visitor sign-up conversion at 1.66 percent versus 0.15 percent for organic search. AI traffic converts 11 times better than organic search. Reporting that channel as “Direct” is a budget allocation error, as we showed in our GEO ROI analysis.

What you will build

The dashboard takes 15 minutes and costs nothing. It contains four blocks:

• Three scorecards: AI sessions, AI users, engaged AI sessions.
• A time series: daily AI sessions over 90 days.
• A donut chart: traffic split by AI engine (ChatGPT, Perplexity, Gemini, Claude, Copilot).
• A landing page table: which pages AI engines actually send people to.

We built this walkthrough on a live GA4 property of a B2B SaaS client (name blurred in every screenshot). Over 90 days the property logged 441 AI sessions from 285 users. Every number you see is real.

Why GA4 hides your AI search traffic

GA4’s default channel report routes most AI clicks into Referral or Direct. Here is the same client property in the standard Traffic acquisition report. Seven channels, no AI anywhere:

The traffic exists. Switch the dimension to Session source and search for chatgpt, and 231 sessions appear:

Google did ship a native “AI Assistant” channel in late 2025. It helps, but it has three problems:

• It is not retroactive. Historical AI sessions stay in Direct and Referral.
• It does not fire on every property. The client property above still shows zero AI channel rows.
• The recognition list lags. Newer engines and regional assistants are missing.

The full fix on the GA4 side is a custom channel group, which we covered step by step in AI Search Referral Attribution in GA4. The Looker Studio dashboard in this guide works without touching GA4 admin settings, and it gives you a view you can share with clients and executives.

Step 1: Connect GA4 to Looker Studio

1. Go to lookerstudio.google.com and click Create, then Report.
2. The “Add data to report” panel opens. Pick the Google Analytics connector.

3. Select your GA4 account, then the property, and click Add.

4. Optional but recommended: rename the data source. Go to Resource, then Manage added data sources, then Edit, and give it a clean name. Charts, filters, and tooltips will all display this name.

Step 2: Create the AI Engine calculated field

GA4 reports sources as raw hostnames: chatgpt.com, perplexity.ai, gemini.google.com, copilot.com. A calculated field groups them into readable engine names. One engine can also have several hostnames (chatgpt.com and openai.com both carry ChatGPT clicks), and the field merges those for you.

1. While editing the data source, click Add a field, then Add calculated field.
2. Name it AI Engine.
3. Paste this formula and click Save:

CASE
  WHEN REGEXP_CONTAINS(Session source, "(?i)(chatgpt|openai)") THEN "ChatGPT"
  WHEN REGEXP_CONTAINS(Session source, "(?i)perplexity") THEN "Perplexity"
  WHEN REGEXP_CONTAINS(Session source, "(?i)(gemini|bard)") THEN "Gemini"
  WHEN REGEXP_CONTAINS(Session source, "(?i)copilot") THEN "Copilot"
  WHEN REGEXP_CONTAINS(Session source, "(?i)claude") THEN "Claude"
  WHEN REGEXP_CONTAINS(Session source, "(?i)grok") THEN "Grok"
  WHEN REGEXP_CONTAINS(Session source, "(?i)deepseek") THEN "DeepSeek"
  WHEN REGEXP_CONTAINS(Session source, "(?i)(mistral|lechat)") THEN "Mistral"
  ELSE "Not AI"
END

Why split by engine at all? Because the engines behave differently. When we decoded 42,971 AI citations, ChatGPT and Google’s AI surfaces shared almost no cited domains, and each engine picked sources by different rules. Engine level reporting tells you where your citation velocity is actually building.

Step 3: Create the AI Traffic Only filter

The filter is the heart of the dashboard. It keeps sessions whose source matches a known AI platform and drops everything else.

1. Select any chart, scroll to Filter in the Setup panel, and click Add filter, then Create a filter.
2. Name it AI Traffic Only.
3. Set the clause to Include, Session source, RegExp Contains and paste this pattern:

(?i)(chatgpt|openai|perplexity|gemini|bard|copilot|claude|grok|deepseek|mistral|lechat|meta.ai|poe.com|you.com|phind)

Save it once and reuse it everywhere. Every later chart attaches the saved filter with two clicks. Revisit the pattern quarterly: new engines launch, and a stale regex silently undercounts. This is the June 2026 list.

Step 4: Add the four report blocks

Use Add a chart from the toolbar and draw each block on the canvas. Configure them like this:

Two shortcuts save time. Build one scorecard fully, then copy and paste it twice and swap the metric; the copies keep the filter. And the doughnut’s dimension is the AI Engine field you created in Step 2.

Step 5: Set a 90 day date range and read the results

AI referral volumes are still small for most sites, so a 28 day window looks noisy. Select all charts, set Default date range to Custom, and use Advanced with Today minus 90 days. Here is the finished dashboard on the client property:

What 90 days of real data showed:

• ChatGPT drove 52.8 percent of AI sessions. Below its 74.78 percent global share measured by SE Ranking across 101,574 websites, because this client ranks well in Gemini and Perplexity.
• Gemini took 15.6 percent and Perplexity 15.2 percent. Gemini referrals grew 388 percent year over year per Similarweb data via Digiday, so that slice will keep widening.
• Claude contributed 11.3 percent. Worth watching if your audience is technical.
• The cited pages were almost all deep guides. Not the homepage. AI engines send visitors to specific answers, which is exactly the behavior we map in our zero-click strategy and AI Overviews CTR analysis.

What this dashboard still misses

Referrer based tracking has a hard ceiling, and you should know where it is. Loamly’s February 2026 data puts 70.6 percent of AI visits in the “dark” bucket: no referrer header, classified as Direct. Mobile AI apps strip referrers, and many users copy URLs from answers instead of clicking. SE Ranking’s study of 101,574 websites pegs visible AI referrals at just 0.15 percent of global traffic in 2025, up from 0.02 percent in 2024, a sevenfold rise.

So treat the dashboard as your floor, not your ceiling. To close the gap:

• Run AI crawler log file analysis to see GPTBot, ClaudeBot, and PerplexityBot activity that never appears in GA4.
• Track brand visibility across the AI engines directly, since citations precede clicks.
• Measure AI search share of voice against competitors, not just your own sessions.
• Watch Direct traffic growth. If Direct grows faster than brand awareness, dark AI traffic is the usual cause.

The 15 minute investment that reframes your reporting

AI search traffic is small, fast growing, and converts far above organic. A dashboard that isolates it changes the conversation from “why is Direct up” to “ChatGPT sent 231 visitors to our pricing guides”. If you are new to the discipline, start with what GEO is and the GEO KPI framework. And if you would rather have the whole measurement stack built for you, that is literally what we do for clients.