At a glance
Percentage of AI-referred sessions that result in a purchase. AI visitors typically convert at 1.5-3x the site-wide conversion rate because they arrive after asking a specific buying question and seeing the merchant recommended as the answer. The card answers “are AI users who land on our site actually buying?” If yes, AI is a quality channel worth amplifying via better content and structured data. If no, the AI conversation may be misrepresenting the merchant’s offering.
| What it counts | Conversion rate (sessions with a purchase event ÷ total sessions × 100) for sessions where session_source matches a known AI platform domain. Computed across the current 30-day period. |
| Sample type | GA4 session and purchase data, refreshed on the standard data refresh. |
| Why AI conversion matters | (1) Channel quality verification: traffic without conversion is vanity; AI traffic with healthy conversion is a real growth channel. (2) Content alignment proof: high AI conversion means the AI’s representation of the merchant matched the on-site reality; low AI conversion means a mismatch (AI promised something the site doesn’t deliver). (3) Investment justification: a 4% AI conversion rate vs 2% site-wide means doubling AI traffic is worth roughly 4x the revenue lift of doubling generic organic. |
| Reading the value | (1) Below 0.5%: AI quality issue; investigate content-claim alignment. (2) 0.5-1%: emerging; small sample size may distort the figure. (3) 1-3%: typical site conversion-aligned. (4) Above 3%: high-quality AI traffic; this is the typical pattern as AI users self-select for purchase intent. (5) Compare to site-wide ga_conversion_rate. |
| Currency | percent. |
| Time window | 30D vsP. |
| Alert trigger | ai_conversion < 0.5 (BAD threshold). |
| Sentiment key | ai_conversion (HIGHER_IS_BETTER; GOOD ≥ 0.5%, BAD < 0%). |
| Roles | owner, marketing |
Calculation
Worked example
A UK-based ecommerce merchant on BC, AI Conversion reading on Wednesday 15 May 26.| Channel | Sessions (30D) | Purchases | Conversion rate | Notes |
|---|---|---|---|---|
| AI Traffic (combined) | 327 | 11 | 3.4% | Above site-wide |
| Site-wide (all channels) | 108,142 | 2,103 | 1.94% | Baseline |
| Organic search (non-AI) | 42,500 | 893 | 2.10% | Slightly above baseline |
| Paid search | 38,200 | 1,005 | 2.63% | Best non-AI channel |
| Direct | 14,800 | 220 | 1.49% | Below baseline |
| 6,400 | 380 | 5.94% | Highest converter (own audience) | |
| AI: ChatGPT | 184 | 6 | 3.2% | - |
| AI: Perplexity | 78 | 3 | 4.1% | Highest AI converter |
| AI: Gemini | 38 | 1 | 2.6% | - |
| AI: Copilot | 18 | 0 | 0.0% | Small sample |
| AI: Claude | 9 | 1 | 11.1% | Tiny sample, noisy |
- AI traffic converts at 3.4% vs site-wide 1.94%, 1.75x the baseline. AI is a quality channel. The pre-qualified intent (“the customer asked a specific question and your site was the answer”) translates directly to higher conversion.
- Among AI platforms, Perplexity is the highest-quality referrer at 4.1%. Perplexity is purpose-built for source-attributed search; users explicitly click through to see the source. ChatGPT users sometimes click through, sometimes accept the in-line answer; the 3.2% reflects this mixed pattern.
- Small-sample noise warning. The Claude bucket has 9 sessions and 1 purchase showing 11.1%, but with only 9 sessions, a single conversion swings the rate by 11 percentage points. Don’t make decisions on samples below 50 sessions per platform. Aggregate across platforms or wait for sample to grow.
-
What healthy AI conversion looks like by site type:
- B2C ecommerce: AI conversion typically 1.5-3x site-wide (this merchant is at 1.75x, healthy).
- B2B / considered-purchase: AI conversion typically 0.5-1.5x site-wide; AI users still need consideration time.
- High-AOV / luxury: AI conversion may be lower than baseline because purchase decisions involve more touchpoints; track AI-attributed assists.
-
What “low AI conversion” reveals:
- Content-claim mismatch: AI promised “best garden hoses” but the site shows generic descriptions. Audit AI-prompt-test results vs your actual product copy.
- Pricing surprise: AI gave a price range; site has higher prices. Audit pricing transparency.
- Inventory mismatch: AI recommended a product that’s now OOS. Cross-reference
out_of_stock_count. - Brand recognition gap: AI recommended you but the user bounces because the brand is unknown. Build brand-authority signals (PR, reviews).
-
Recommended actions:
- Maintain or amplify: at 3.4%, AI is performing well. Continue investing in content + structured data to grow volume.
- Periodic prompt-test: monthly, ask ChatGPT, Perplexity, Claude for product recommendations in your category. Verify the AI representation matches reality.
- Cross-reference
ai_engagementandai_revenuefor the full quality picture.
- Read the conversion rate. Compare to site-wide.
- Decompose by AI platform (sample-size permitting).
- If low: prompt-test to verify AI’s representation.
- If high: sustain via content + schema investment.
| Time horizon | Action |
|---|---|
| First 1 hour | Read AI vs site-wide rate. |
| First day | Decompose by platform; identify quality issues. |
| First week | Prompt-test AI representations vs site reality. |
| Quarterly | Track AI conversion rate trend; rising rate signals improving content. |
Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
ai_traffic | AI traffic volume; complements quality. |
ai_revenue | Revenue contribution; volume × CR × AOV. |
ai_engagement | Engagement; pre-conversion quality. |
ai_traffic_overview | Detailed AI breakdown. |
ai_vs_search | Compare AI vs traditional search. |
ga_conversion_rate | Site-wide conversion rate baseline. |
Reconciling against the vendor’s own dashboard
Where to look in GA4: Reports → Acquisition → Traffic acquisition; segment by AI source; check Engagement/Conversions section for purchase events. Why our number may differ:| Reason | Direction | What to do |
|---|---|---|
| Conversion definition. GA4 may use multiple conversion events; this card uses purchase events specifically. | Variable | Match definition. |
| Sample size. Small AI samples produce noisy GA4 segments. | Either | Aggregate across platforms. |
| Attribution model. Last-click vs data-driven changes which sessions get credit. | Variable | Confirm model. |
ai_revenue for total channel value.
Quick rule: confirm conversion definition and attribution model first.
Known limitations / merchant FAQs
Q: AI conversion is 3.4% but my site-wide is 2%. Why is AI better? Self-selection. AI users explicitly asked a buying question (e.g., “what’s the best 30m garden hose for hard water?”) and were directed to your site as the answer. They arrive pre-qualified and ready to convert. Site-wide includes all sources, many of whom are top-funnel browsers without immediate purchase intent. Q: AI conversion dropped from 5% to 1%. Did the channel break? Likely a sample-mix shift. Either: (a) one AI platform’s algorithm changed and it now sends a different (lower-intent) audience, (b) a competitor’s content improved and is now winning the high-intent queries, (c) seasonal: query patterns shifted. Decompose by platform to identify which one moved. Q: We have AI traffic but 0 AI purchases. What’s wrong? At low volumes (under 50 sessions/month), 0 purchases is a sample issue, not a quality issue. Above 100 sessions/month with 0 purchases signals a real problem: prompt-test to see what AI is saying about you (likely making promises the site doesn’t keep, pricing, availability, or product claims). Q: AI conversion is 0.5%, below the threshold. What do I do? Audit AI’s representation of your offering. Run 5-10 prompt-tests asking AI agents about your product category, brand, or key SKUs. Compare what AI says against what your site shows. The mismatch (price, availability, key features) is likely the cause. Fix the underlying content or schema; AI conversion should recover within 2-4 weeks. Q: Should I optimise this card directly? Indirectly. The drivers of AI conversion are the same as the drivers of organic conversion: clear value props, accurate descriptions, valid Product schema with prices and availability, fast PDPs. Investing in those lifts AI conversion (and SEO conversion, and direct conversion) simultaneously. Q: How does this differ fromga_conversion_rate?
ga_conversion_rate is site-wide (all sources). This card is AI-source only. Use both: site-wide for baseline; AI-source for channel quality verification.
Q: What’s the threshold of significance for AI conversion?
Roughly 100 sessions/month before single-period readings are reliable. Below that, aggregate across 60-90 days for a stable signal. Many merchants are still in the sub-100 phase as of 2026; the metric becomes more actionable as AI traffic volume grows.
Q: Are paid AI ads (e.g., Bing Chat sponsored placements) included?
No, paid AI placements are ad-source-tagged differently (utm_medium=cpc with the ad platform) and roll up under paid search rather than AI traffic. This card is specifically organic AI referral traffic.