At a glance
AI traffic plotted over time, 90-day rolling. Volume shifts week-to-week are noisy; the 90-day trend is the canonical signal of whether AI-discoverability work is paying back. The healthy pattern is steady right-trending growth: AI traffic compounds as content + structured data + brand authority improve. Flat or declining trend means content/structure work is needed; sudden spikes need investigation (a viral mention, a new AI platform launching, or a structured-data fix unlocking visibility).
| What it counts | Daily AI-referred sessions plotted across the rolling 90-day window. The trend slope is the headline signal. |
| Sample basis | GA4 sessions filtered to AI source domains, daily aggregation. |
| Sampling threshold | GA4 sampling above 10M events; small AI populations unsampled. |
| Bot traffic filter | Default + Vortex IQ. |
| Time zone | Property time zone. |
| Time window | 90D rolling, daily granularity. |
| Alert trigger | no growth in 90D (slope ≤ 0 over the window) |
| Sentiment key | ai_traffic |
| Roles | owner, marketing |
Calculation
Worked example
A UK-based BC store, AI traffic trend over 90 days reading on Wednesday 15 May 26.| Time bucket | AI sessions | Notes |
|---|---|---|
| 90 days ago (15 Feb) | 95 | Baseline |
| 60 days ago (15 Mar) | 145 | +52% |
| 30 days ago (15 Apr) | 240 | +66% |
| Today (15 May) | 327 | +36% from prior month |
| 90D total | 6,200 | - |
| 90D trend slope | +4.2 sessions/day | Steady positive growth |
- Slope is positive at +4.2 sessions/day across 90 days. Traffic is roughly 3.4x what it was 90 days ago. The compounding pattern is consistent with the early-growth stage of AI-search adoption.
-
Inflection points to investigate:
- 15 Mar (60 days ago): jump from 95 to 145 likely tied to either a content-quality improvement (description refresh, schema fix) or an AI platform’s algorithm update making the brand more visible.
- 15 Apr (30 days ago): another acceleration to 240, could be Perplexity’s expanded user base, ChatGPT’s improved web-browse, or a brand-mention event.
- The forward projection. At current slope, AI sessions hit 1,000+/month within 6 months and 3,000+/month within 12 months. Revenue contribution rises proportionally; per-session value of £6.96 means £20K+/month within 12 months at current trajectory.
- What flat or declining trend looks like. A 90-day slope at or below zero indicates: (a) no investment in content or structured-data improvement, (b) competitor AI-discoverability outpacing yours, (c) an AI platform’s algorithm change that reduced your visibility. Investigation needed.
-
Recommended actions for sustained growth:
- Maintain content depth, every product page should answer the AI’s likely query.
- Validate structured data quarterly, Product schema changes can break AI extraction.
- PR + brand mentions, AI training data prefers authoritative sources.
- Track per-platform trends via
ai_traffic_overview.
Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
ai_traffic | Headline volume; this card is the trend. |
ai_traffic_overview | Per-platform breakdown. |
ai_conversion | Conversion quality as volume grows. |
ai_revenue | Revenue trend; should track traffic trend. |
ai_engagement | Engagement quality. |
ga_traffic_trend | Total traffic trend; AI is a subset. |
Reconciling against the vendor’s own dashboard
Where to look in Google Analytics 4’s own dashboard: GA4 → Reports → Acquisition → Traffic acquisition; segment by AI sources; view as time series. Why our number may legitimately differ:| Reason | Direction | What to do |
|---|---|---|
| AI source list curation. Vortex IQ maintains updated list. | Variable | Confirm. |
| Period boundary. GA4 default period may differ. | Variable | Match. |
| Sampling at scale. Above 10M events GA4 may sample; small AI traffic typically unsampled. | Variable | Confirm. |
ga_traffic_trend (total traffic) for context.
Known limitations / merchant FAQs
Q: Our AI traffic is flat at 0 for 90 days. Is that a problem? Yes, it means AI platforms aren’t surfacing the merchant in product recommendations. Investigate: structured data validity, content depth, brand authority signals, robots.txt blocking AI crawlers. Each is fixable; AI traffic typically begins within 30-60 days of remediation. Q: We see steady growth at +5%/week. Is that good? Excellent. 5%/week compounds to 13x in a year, well ahead of the typical AI-channel growth curve. Likely indicators: rich content, strong PR / mentions, valid structured data, AI-favourable robots.txt. Q: AI traffic spiked 5x in one day. Glitch or real? Usually one of: (a) a major AI platform updated its index and your site got re-included, (b) a viral mention on Reddit/X drove AI agents to recommend you, (c) a competitor went down and AI platforms re-routed recommendations. Checkai_traffic_overview to see if one platform spiked or all of them.
Q: How long does it take for content/schema improvements to show in this card?
Typically 14-60 days. AI platforms re-crawl periodically (some weekly, some monthly), and training data updates lag further (3-6 months for major models). Don’t expect immediate uplift; track quarterly.
Q: Is there a seasonal pattern in AI traffic?
Less than traditional channels, but yes, gift-buying questions to AI platforms spike in November-December for ecommerce. Plan content for seasonal queries (gift guides, comparison content) ahead of the spike.
Q: How does this differ from ga_traffic_trend?
That card is total sessions across all sources; this card is AI sources only. Total can be flat while AI grows, that’s healthy mix shift toward a higher-quality channel.
Q: Should we set custom alerts for AI traffic milestones (e.g., first 100 sessions/month)?
Yes if useful for your team. Vortex IQ supports custom alerts via Vortex Mind. Common milestones: first 100/month (channel established), first 1,000/month (channel material), first 10,000/month (channel becoming dominant).
Q: We blocked GPTBot in robots.txt. Should we unblock?
Almost always yes for ecommerce. Blocking saves on the bot’s training-data scraping but excludes you from AI search. The discoverability benefit dwarfs the IP concern for most ecommerce.