Dark AI: Why Half Your Buyers Research In ChatGPT But Show As Under 1% Of Your Traffic
AI now shapes half of purchase decisions, yet AI-referred traffic shows up as roughly 0.2 percent of visits in GA4. Analysts call this the dark AI attribution gap. Here is what it means for Indian brands and the exact steps to start measuring it this week.
Here is a question worth sitting with before your next budget review: if half your customers are researching their next purchase inside ChatGPT, Gemini, or Perplexity, why does your analytics dashboard barely show any traffic from them? That gap between influence and visibility has a name now. Analysts are calling it dark AI, and it is quietly distorting how Indian brands measure what actually drives sales.
The numbers frame the problem sharply. McKinsey finds that roughly half of consumers now use AI as their primary or preferred source for product research. Yet analysis of real web traffic puts AI-referred sessions at around 0.2 percent of total visits. Both figures are accurate, and the space between them is where your demand is being created without your tools seeing it.
// The Short Version
- Half of consumers now use AI tools as their main product research source.
- AI-referred traffic shows up as only about 0.2 percent of visits in analytics.
- ChatGPT-referred sessions convert higher than direct, paid search, or organic.
- The influence is real, the measurement is missing: that is the dark AI gap.
- Last-click reporting makes your strongest demand look invisible.
What the dark AI attribution gap means for Indian brands
For Indian marketers, this matters more each month as urban buyers move research into chat tools. A customer asks an AI assistant to compare options, reads the answer, then arrives at your site through a branded search or by typing your URL directly. Your analytics records that as direct or branded traffic, free and unattributed, when in reality an AI conversation did the persuading. The work happened in a room your dashboard cannot enter.
The conversion data makes the stakes concrete. Retail analysis shows ChatGPT-referred sessions converting at around 11.4 percent, ahead of direct traffic, paid search, and organic search. The visitors who do arrive tagged from AI are unusually high intent, which strongly implies they represent the visible tip of a far larger pool of AI-influenced journeys you are not capturing at all.
Your strongest demand channel may be the one your dashboard shows as nearly zero. That is not a data error. It is a blind spot.
Why your last-click reports are misleading you
The danger is not just missing data, it is making the wrong decision because of it. When AI-influenced buyers land as direct or branded traffic, last-click attribution hands the credit to the final touch and starves the channels that actually built the intent. In a budget review, your top-of-funnel and content work looks like it is underperforming, and the natural reaction is to cut it. You would be cutting exactly what feeds the demand you are trying to grow.
DO TODAY Your three-step move this week
- Ask the customer directly. Add a self-reported “How did you hear about us” field to your checkout and lead forms, with an explicit AI assistant or ChatGPT option, so you capture influence your tracking misses.
- Tag AI referrals in GA4. Identify and tag any AI-referred sessions you can capture as a distinct channel, so AI traffic stops hiding inside direct and referral buckets.
- Wait before you cut. Review the share of buyers citing AI after 30 days, and hold any channel-cut decision until you cross-check against that signal rather than last-click alone.
The bigger picture
This is the early stage of a structural shift in how discovery works. As AI assistants become the first place buyers go, the link between influence and a trackable click keeps weakening, and the brands that adapt their measurement first will read the market more clearly than competitors still trusting last-click dashboards. You do not need a perfect solution today. You need to stop assuming the traffic you can see is the whole story.
The Compass TakeYou cannot fully measure AI influence yet, and any vendor claiming a clean number is overselling. What you can do is stop making cut decisions on last-click alone. Add the self-reported field, tag what you can, and treat rising direct and branded traffic as a signal that AI is working, not as free traffic to ignore.