Attribution Reconciliation
See the real business impact of AI mentions — how many visits each LLM platform drives to your pages, with statistical confidence.
What it measures
When AI platforms cite your pages in their responses, does that actually translate into real traffic? Attribution Reconciliation answers this question by connecting two data streams: the URLs cited in LLM responses (from our prompt monitoring pipeline) and the page-level traffic from your Google Analytics 4 account.
The system runs daily as part of the intelligence pipeline, comparing AI citation patterns against GA4 traffic baselines to identify pages where AI mentions are driving measurable visits. Each page receives a confidence score from 0 to 100, indicating how strongly the evidence supports AI-driven attribution.
How the pipeline works
Attribution reconciliation runs in five stages, each building on the previous.
Citation Extraction
Every time an LLM platform (ChatGPT, Perplexity, Gemini, Claude) responds to a tracked prompt and mentions your brand, we extract any URLs cited in the response. These URLs are normalised to page paths — stripping tracking parameters, lowercasing hosts, and removing fragments — so that `/services/consulting` and `/services/consulting?utm_source=linkedin` are treated as the same page.
GA4 Channel-Group Ingestion
Each day, the platform ingests page-level session data from Google Analytics 4, broken down by default channel group (Organic Search, Direct, Referral, etc.). This produces a daily snapshot that gives us the traffic denominator for each page.
Baseline Computation
We compute a rolling 28-day IQR baseline for each page's traffic by channel. The interquartile range filters out spikes and dips to produce a stable expected-traffic figure. When a page's AI citation count rises materially and its traffic simultaneously increases beyond the IQR baseline, we flag a potential AI-driven attribution.
Reconciliation Scoring
Reconciliation runs daily after data ingestion. For each page with both citation data and GA4 traffic, the platform computes a deviation score: how far above (or below) the baseline the current traffic is, weighted by the number and recency of AI citations. Pages are scored on a 0–100 confidence scale.
Branded Query Downgrade
If LLM responses primarily drive branded searches (e.g. 'Nous Group consulting') rather than direct page visits, the attribution confidence is adjusted downward. Our branded query matcher detects when GSC queries overlap with brand name variations and flags these as brand-halo traffic rather than direct AI attribution.
Confidence tiers
Every attribution result includes a confidence score. Here is what each tier means.
High
75–100Strong signal overlap: page was cited in AI responses, traffic increased beyond IQR baseline, and timing aligns within the measurement window. Non-branded query uplift detected.
Medium
40–74Moderate signal: citations present with some traffic uplift, but confounding factors exist — seasonal patterns, concurrent marketing campaigns, or branded query dominance reduce certainty.
Low
0–39Weak attribution signal: citations detected but traffic is flat or below baseline. The AI mentions may not yet be driving meaningful visits, or attribution is entirely branded.
Data sources
AI Citation Data
URLs extracted from LLM responses to your tracked prompts. Normalised to page paths and deduplicated. Updated daily as part of the prompt monitoring pipeline.
Google Analytics 4
Page-level sessions by default channel group. Ingested daily via the GA4 Data API. Requires an active GA4 OAuth integration in your organisation settings.
Google Search Console
Query-level click and impression data used for branded query detection. Pages where branded searches dominate have their attribution confidence adjusted downward.
Frequently Asked Questions
How long before attribution data appears?
Attribution requires at least 28 days of GA4 channel data to compute baselines. New organisations will see their first attribution scores approximately one month after connecting Google Analytics.
What if I don't have Google Analytics connected?
Attribution reconciliation requires a GA4 integration. Without it, you will still see citation data from AI platforms, but cannot reconcile those citations against real traffic. Connect GA4 in Settings > Integrations.
Does this double-count organic search traffic?
No. Attribution reconciliation specifically isolates the incremental traffic above baseline. If a page normally receives 200 organic visits per day and after AI citations begin it averages 260, the reconciliation scores the incremental 60 visits — not the entire 260.
How do you handle pages with zero prior traffic?
Pages with no historical baseline (new pages or pages that previously had negligible traffic) are flagged with lower confidence, since any traffic increase cannot be compared against a stable baseline. As data accumulates, confidence increases automatically.