> ## Documentation Index > Fetch the complete documentation index at: https://docs.vortexiq.ai/llms.txt > Use this file to discover all available pages before exploring further. # AI Engagement, Google Analytics 4 > AI Engagement for Google Analytics 4 stores. Tracked live in Vortex IQ Nerve Centre. How to read it, why it matters, and how to act on it. **Metrics type:** [Supporting Metrics](/nerve-centre/overview#metrics-types-explained) • **Category:** [Analytics](/nerve-centre/connectors#connectors-by-type) ## At a glance > **Engagement rate of AI-referred sessions.** Measures the depth of interaction (10s+ time-on-site, 2+ pages, or conversion) for users arriving from AI platforms. **High AI engagement means the AI's representation matched what the user found.** Low AI engagement means a mismatch: AI promised something the site didn't deliver, and users bounced. The card complements `ai_conversion` (purchase rate) with the pre-conversion engagement signal. | | | | ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **What it counts** | Engagement rate for AI-referred sessions over the rolling 30-day window. GA4's engagement-rate definition: percentage of sessions that lasted 10+ seconds, viewed 2+ pages, or fired a conversion event. | | **Sample basis** | Sessions filtered to AI source domains (chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, copilot.microsoft.com, you.com, etc.). | | **Sampling threshold** | GA4 sampling kicks in above 10M events; small AI-traffic stores are well below this and unsampled. | | **Bot traffic filter** | GA4 default bot filter applied; additional Vortex IQ bot detection layered. | | **Time zone** | GA4 property time zone (matches merchant config). | | **Time window** | `30D` | | **Alert trigger** | `engagement_rate < 40%` | | **Sentiment key** | `ai_engagement` (HIGHER\_IS\_BETTER; GOOD ≥ 30%, BAD \< 10%) | | **Roles** | owner, marketing | ## Calculation ``` ai_engagement = COUNT(ai_sessions WHERE engaged = true) ÷ COUNT(ai_sessions) × 100 ``` Where engaged = (session\_duration ≥ 10s) OR (page\_views ≥ 2) OR (conversion\_event\_count ≥ 1). ## Worked example A UK-based BC store, AI engagement reading on Wednesday 15 May 26. | Channel | Sessions | Engaged sessions | Engagement rate | Notes | | ------------------- | -------: | ---------------: | --------------: | ------------------ | | Site-wide | 108,142 | 43,900 | 40.6% | Baseline | | AI traffic combined | 327 | 164 | **50.2%** | Above baseline | | Organic search | 42,500 | 18,700 | 44.0% | - | | Direct | 14,800 | 4,700 | 31.8% | Below baseline | | AI: ChatGPT | 184 | 96 | 52.2% | Above baseline | | AI: Perplexity | 78 | 45 | 57.7% | Highest engagement | | AI: Gemini | 38 | 14 | 36.8% | Below baseline | What the engagement reading is telling us: 1. **AI engagement at 50.2% materially exceeds the site-wide 40.6%.** Pre-qualified AI users spend more time and view more pages, consistent with the higher AI conversion rate. 2. **Perplexity sends the most engaged AI traffic at 57.7%.** Perplexity users explicitly clicked through to source, they're already pre-engaged. ChatGPT users vary depending on whether they accepted the in-line answer or clicked through. 3. **Gemini engagement is lower (36.8%).** Likely because Gemini's AI Overviews surface short answers without strong incentive to click through; the small share of Gemini users who do click through include some lower-intent browsers. 4. **What "low AI engagement" means.** Below 30% suggests AI users are bouncing on landing, the AI promised something the site doesn't deliver. Most common causes: outdated AI representation (price, availability), broken landing page, missing product copy. 5. **Recommended actions:** * **Maintain or improve content quality**, engagement is downstream of content match. * **Periodic prompt-test**, verify AI's representation matches site reality. * **Cross-reference `ai_conversion`**, engagement and conversion should track. ## Sibling cards merchants should reference together | Card | Why merchants reach for it | | ------------------------------------------------------------------------------------- | -------------------------------------------- | | [`ai_traffic`](/nerve-centre/kpi-cards/google-analytics/ai-traffic) | AI traffic volume. | | [`ai_conversion`](/nerve-centre/kpi-cards/google-analytics/ai-conversion) | AI conversion rate; should track engagement. | | [`ai_revenue`](/nerve-centre/kpi-cards/google-analytics/ai-revenue) | AI revenue contribution. | | [`ai_traffic_overview`](/nerve-centre/kpi-cards/google-analytics/ai-traffic-overview) | Detailed AI breakdown. | | [`ai_vs_search`](/nerve-centre/kpi-cards/google-analytics/ai-vs-search-traffic) | AI vs traditional search comparison. | | [`ga_engagement_rate`](/nerve-centre/google_analytics/ga_engagement_rate) | Site-wide engagement baseline. | ## Reconciling against the vendor's own dashboard **Where to look in Google Analytics 4's own dashboard:** GA4 → Reports → Engagement → Engagement overview, segmented by AI source. **Why our number may legitimately differ from the vendor's:** | Reason | Direction | What to do | | ----------------------------------------------------------------------------------- | --------- | ---------------------------------- | | **AI source list.** Vortex IQ maintains curated list. | Variable | Confirm list. | | **Engagement definition.** GA4 default 10s+/2+ pages/conversion; this card matches. | Match | n/a. | | **Sampling.** Above 10M events, GA4 may sample. | Variable | Use unsampled report if available. | **Cross-connector reconciliation:** complement with `ai_traffic` and `ai_conversion` for the full AI quality picture. ## Known limitations / merchant FAQs **Q: AI engagement is 50%, is that good?** Above the 40% baseline; healthy. AI users self-select for engagement; rates of 45-60% are typical for stores with rich content + valid structured data. **Q: Our AI engagement is 25%, what does that mean?** AI users are landing and bouncing. Likely causes: AI is promising something (specific product, price, availability) that the site doesn't show; broken landing page; thin product copy. Run prompt-tests to verify. **Q: Why does AI engagement vary across platforms?** Different AI platforms send different intent profiles. Perplexity (purpose-built for source-attribution) sends the most engaged users. ChatGPT mixes in-line and click-through users. Gemini's brief AI Overviews send fewer click-throughs. **Q: How does this differ from `ai_conversion`?** Engagement is the pre-conversion quality signal; conversion is the bottom-line. They should track each other; if engagement is high but conversion is low, the friction is at checkout (cross-reference `ga_checkout_completion`). **Q: How do we improve AI engagement?** Improve content quality (rich, accurate descriptions); validate structured data; ensure landing pages match the AI's representation; minimise PDP load time for the click-through experience. **Q: GA4 sampling, does it affect AI engagement readings?** At small AI traffic volumes (under 1K sessions/month), no sampling. At larger volumes, GA4 may sample but the engagement rate as a percentage remains stable. The card flags sampling explicitly. **Q: Engagement is high but AI traffic volume is small, should we focus elsewhere?** Volume is the constraint, not quality. Investments to grow AI traffic (rich content, structured data, brand authority) preserve the high engagement rate while raising volume, both lift AI revenue. **Q: Does the card include AI bot crawlers like GPTBot?** No. GPTBot, ClaudeBot, PerplexityBot crawler traffic is filtered as bots. The card measures human users referred from AI conversation interfaces. *** ### Tracked live in Vortex IQ Nerve Centre *AI Engagement* is one of hundreds of KPI pulses Vortex IQ tracks across Google Analytics 4 and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English. [Start for free](https://app.vortexiq.ai/login) or [book a demo](https://www.vortexiq.ai/contact-us) to see this metric running on your own data.