AI search visibility dashboard in Looker Studio
Build a Looker Studio board for AI Assistant referrals, Search Console AI data and MentionX-style prompt tracking in Sheets.
An AI visibility dashboard is the board leadership asks for after they hear "ChatGPT is the new homepage." Build it as three honest lanes — GA4 AI Assistant referrals, Search Console Generative AI surfaces, and a prompt-tracking Sheet for citation checks — never as one fake "AI traffic" number. This lesson wires the board in Data Studio and links the Academy measurement track.
- Three lanes, three scorecard groups, one cover that defines them.
- GA4 AI Assistant ≠ Search Console Generative AI ≠ prompt mention rate.
- Put MentionX-style scans in Sheets or BigQuery, then connect.
- Weekly cadence beats real-time theatre.
- Link deep dives: Track AI traffic in GA4, monthly AI visibility report template.
Checked 5 November 2026 (CET) against Google Cloud Data Studio documentation, the Data Studio product page, the Looker Studio-is-Data-Studio blog post, connector help pages and the Analytics Help pages cited in Sources. Google renames and moves features often, so confirm labels and limits in your own account. "Our view" marks MentionX opinion. Ledgerline is a fictional company used for illustration.
01What are the three lanes?
| Lane | Source | Primary metrics | Common mistake |
|---|---|---|---|
| Assistant referrals | GA4 | Sessions, engaged sessions, key events by session source (chatgpt.com, perplexity.ai, gemini.google.com, …) | Mixing into Organic Search |
| Search generative | Search Console GenAI export or connector fields | Impressions, clicks by AI Mode / AI Overviews (use columns your property shows) | Calling these "ChatGPT clicks" |
| Prompt visibility | Sheets / BigQuery from scans | % prompts mentioned, share of voice vs competitors A–D, cited URL rate | Inventing daily ranks without a scan log |
02How do you build lane 1 (GA4)?
- Reuse the GA4 owner data source.
- Scorecard: sessions where Session default channel group = AI Assistant.
- Table: Session source, Sessions, Engaged sessions, Key events, filtered to AI Assistant.
- Time series of AI Assistant sessions.
- Optional: landing page table filtered to AI Assistant for content winners.
- Caption: "Excludes AI Mode and AI Overviews (those are Organic Search / Search Console)."
Cross-check GA4 AI Assistant channel regex and GTM AI referral traffic if tagging is incomplete. Note Direct undercount for app clicks (AI referral traffic undercounted).
03How do you build lane 2 (Search Console Generative AI)?
- Prefer a weekly Sheet export from the Generative AI performance report until you confirm connector fields in-account.
- Columns to aim for: Date, Search appearance / surface, Impressions, Clicks (match your UI).
- Scorecards split by surface when available.
- Caption linking Search Console generative AI performance report and Gemini Search Console AI reports.
04How do you build lane 3 (prompt tracking Sheet)?
Suggested schema for ai_prompt_tracking:
| Column | Type | Example |
|---|---|---|
| scan_date | Date | 2026-11-10 |
| engine | Text | chatgpt / perplexity / gemini / claude |
| prompt | Text | best ap automation for mid market |
| brand_mentioned | Text YES/NO | YES |
| position | Number or null | 2 |
| cited_url | Text | https://ledgerline.example/ap |
| competitor_mentioned | Text | A,C |
| notes | Text |
Connect Sheets with 15-minute or longer freshness. Scorecards:
- Mention rate = COUNT of YES / COUNT of rows (chart-level).
- Mentions by engine (stacked bar).
- Top cited URLs table.
- Competitor co-mention table (keep competitors as A–D in Academy examples; real brands OK on client boards).
Automate scans with n8n AI visibility monitoring or MentionX exports — agents draft, humans approve outbound.
05Page layout
| Page | Content |
|---|---|
| Cover | Definitions of the three lanes + owner + cadence |
| Assistants | GA4 lane |
| Search generative | GSC lane |
| Prompt tracker | Sheet lane + engine comparison |
| Actions | Top 5 URLs to improve + owners (manual text or Sheet) |
Mirror the executive GEO dashboard template for the one-pager version leadership wants.
06Worked example: Ledgerline
Ledgerline is a fictional accounts-payable software company. Details are illustrative.
| Lane | 28-day snapshot (illustrative) |
|---|---|
| AI Assistant sessions | 1,940 |
| chatgpt.com share | 58% |
| GenAI export clicks | 620 |
| Prompt mention rate (40 prompts × 4 engines) | 34% |
| Competitor A mention rate | 41% |
Walkthrough
- Marketing tried one "AI traffic" scorecard; the CMO rejected it after a confusing QBR.
- Three-lane cover fixed the vocabulary in a week.
- Prompt Sheet highlighted missing
/integrationscitations. - Weekly Slack used a scheduled PDF of the Assistants page only.
- They aligned KPIs with AI visibility KPIs for marketing teams.
07Sample weekly narrative (for Slack or PDF note)
AI visibility week {n}
- Assistants: {sessions} sessions ({wow}% wow); top source {source}
- Search generative: {genai_clicks} clicks (export)
- Prompt mention rate: {rate}% on {n_prompts} prompts × {n_engines} engines
- Action: {owner} to ship {url} fixes by {date}Keep the narrative next to the dashboard so screenshots do not lose the story.
08Engine coverage vs traffic coverage
Prompt tracking might include Claude and others that send little or no web referrer. That is fine — lane 3 is about answers, lane 1 is about visits. See Claude visibility, ChatGPT track, Perplexity track, Gemini track.
09How do you keep sample sizes honest?
Small prompt sets swing wildly week to week. Publish the denominator next to every mention-rate scorecard (n = prompts × engines × days). Link interpreting week-over-week visibility swings and statistical confidence for AI visibility tracking from the appendix. If n is tiny, show a qualitative table of prompts instead of a precision percentage.
10Sheet automation options
| Method | Pros | Cons |
|---|---|---|
| Manual weekly paste from MentionX | Simple | Humans forget |
| n8n workflow writing rows | Repeatable | Needs maintenance (n8n AI visibility monitoring) |
| BigQuery load then Data Studio | Scales | Needs eng |
Agents draft; humans approve anything customer-facing. Do not auto-tweet scan results from a dashboard webhook.
11What decisions does this board unlock?
| Signal | Decision |
|---|---|
| Assistant sessions up, key events flat | Fix landing pages cited by assistants |
| GenAI impressions up, clicks flat | Improve titles/snippets for AI Mode surfaces |
| Mention rate down vs competitor A | Ship comparison and integration pages |
| Cited URLs concentrate on /blog only | Push product and docs pages into the citation set |
If the board does not change a backlog item within two weeks, it is decoration. Tie each lane to an owner in the Actions page.
12Common mistakes
- One number for all AI.
- Prompt scores without timestamps or engine labels.
- Treating MentionX position like Google rank.
- Daily panic on small prompt samples — see interpreting week-over-week visibility swings.
- No link to content owners.
13Try it yourself
Create the three-lane cover. Wire GA4 AI Assistant. Add a 10-row illustrative prompt Sheet. Add a GenAI Sheet stub. Schedule a weekly email to yourself.
14Quick quiz
1. Name the three lanes.
Answer: GA4 AI Assistant referrals; Search Console Generative AI; prompt-tracking Sheet/warehouse.
2. Where do AI Mode impressions belong?
Answer: Search Console Generative AI lane (not GA4 AI Assistant).
3. Why timestamp every prompt row?
Answer: So week-over-week mention rate is computable and auditable.
15What should you check before the next lesson?
- Cover definitions live on every page.
- Prompt schema includes engine + date + YES/NO.
- No single "AI traffic" mega-metric.
Previous: Cross-channel ROI with blended Looker Studio data. Next: Executive reporting with Looker Studio. Or return to the Looker Studio for Marketers track. Related: Monthly AI visibility report template, AI visibility tracker setup, Perplexity for Marketers, ChatGPT for Marketers, Claude for Marketers.
16Common questions
What should an AI visibility dashboard include? Three lanes: GA4 AI Assistant referrals (ChatGPT, Perplexity, Gemini app and similar), Search Console Generative AI performance for AI Mode and AI Overviews, and a prompt-tracking Sheet for citation and recommendation checks.
Can MentionX scan results feed Looker Studio? Yes. Export or sync scan results into Google Sheets or BigQuery, then connect that table. Keep engine, prompt, date, mentioned (yes/no), position and cited URL as columns.
Should AI Mode sit in the same scorecard as ChatGPT referrals? No. AI Mode and AI Overviews are Search generative surfaces (Search Console). ChatGPT and similar assistants are referral traffic (GA4 AI Assistant). Report them in separate scorecards.
Questions people ask
3 answersWhat should an AI visibility dashboard include?
+
Three lanes: GA4 AI Assistant referrals (ChatGPT, Perplexity, Gemini app and similar), Search Console Generative AI performance for AI Mode and AI Overviews, and a prompt-tracking Sheet for citation and recommendation checks.
Can MentionX scan results feed Looker Studio?
+
Yes. Export or sync scan results into Google Sheets or BigQuery, then connect that table. Keep engine, prompt, date, mentioned (yes/no), position and cited URL as columns.
Should AI Mode sit in the same scorecard as ChatGPT referrals?
+
No. AI Mode and AI Overviews are Search generative surfaces (Search Console). ChatGPT and similar assistants are referral traffic (GA4 AI Assistant). Report them in separate scorecards.
Sources
- Analytics Help, Default channel groupsupport.google.com
- Search Console Help, Generative AI performance reportsupport.google.com
- Google Cloud, Connect to Google Sheetscloud.google.com
- All accessed 5 Nov 2026.