As AI tools become more common, the amount of artificial traffic hitting websites has exploded, and most analytics platforms aren’t built to keep up. AI traffic can distort your reports, tests, and decisions. To understand the real impact, I analyzed traffic across multiple tools and uncovered how much came from bots.
In the video below, we’re looking at how bots, scrapers, and large language models can pollute your analytics data.
Recently, a client noticed an inconsistency in the data they saw in Google Analytics. After digging into my own site, I saw the same issue and decided to figure out what was happening.
As the internet moves toward agentic systems, we’re seeing more crawlers, scrapers, and AI-driven agents doing work across the web. The imbalance between AI traffic and human traffic is growing, and much of that traffic is “dark”; it’s unclear where it’s coming from and often ends up in a direct traffic bucket. I wanted to understand how much of our traffic was actually human and how much was driven by bots.
Red Flags in Google Analytics
Inside Google Analytics, I noticed random traffic spikes. We don’t do much PR or run events, so our traffic shouldn’t jump like that. High-level metrics looked positive, with users up over 100% and new users increasing, but something still didn’t feel right.
One of the biggest indicators was the spike in direct traffic. Direct was up 302% while organic was down 55%. Referrals dipped slightly. Unassigned traffic was up 1,000%, though it recorded only 22 visits. Some fluctuations made sense, such as seasonality, search shifts, or video work, but the direct spike stood out.
Cross-Reference with Google Search Console
GA4 and Search Console were relatively aligned. Sessions and clicks were close, though not identical. A few gaps existed because Search Console backfills data.
Organic traffic had dropped compared to the previous year. We removed older blogs that drove traffic but not business value, so I expected a decline in traffic. Still, the imbalance between organic and direct traffic suggested something else was happening.
Track Bots with Dark Visitors
To dig deeper, I used a tool called Dark Visitors, which tracks agent activity on websites. I noticed similar spikes there. Over the last 28 days, it showed 33,000 agent visits, about half of all traffic.
Googlebot made sense, but other crawlers were less noticeable. Meta’s AI scraper appeared frequently, along with ChatGPT, SEO tools, and various RAG-based assistants.
The “AI data scraper” bot matched the suspicious patterns I was seeing in GA. After pulling everything together, I started filtering behavior signals to differentiate humans from bots.
Custom Analysis Tool
A lot of the direct traffic was likely bots. To analyze this, we built a small React tool that uses engagement triggers such as visiting multiple pages, scrolling, and general on-page activity.
I fed all of the data into BigQuery–if you’re not using BigQuery, I highly recommend you set it up. It gives you 100% of your data and lets you query with SQL, including using an LLM to generate queries.
Bots Dominate Your Direct Traffic
The analysis showed that 68% of direct traffic was likely bots, 16% was human, and the rest was uncertain.
Bot activity wasn’t limited to direct traffic; it spread across other channels. Google Analytics struggles with attribution, so misclassified bot visits end up everywhere. Bot sessions averaged one page and 0.5 seconds of engagement, while human sessions averaged around 43 seconds.
Organic traffic was also affected, with about 53% predicted to be bot-driven.
Once you remove bots, actual engagement is far higher than the GA reports show. My real engagement averaged 43 seconds, not the much lower number shown in GA when it included bot traffic.
The Impact on Your Marketing Strategy
If you aren’t filtering bot traffic, your A/B tests and optimization decisions are based on flawed data. After comparing trends, it was clear how much bots were inflating metrics.
The biggest offender was Meta’s AI research scraper. Others included Google Search, uptime monitoring tools, ChatGPT, HFS, Yandex, Baidu, and Huawei’s SimrPet Bot. GA filters some of this out, but far from all of it.
Reporting becomes difficult when you can’t trust traffic numbers, and leadership expects clarity and honest metrics. Artificial traffic isn’t going away, so marketers must get smarter about separating AI visits from human visits.
Microsoft Clarity: A More Accurate Alternative
Microsoft Clarity did a better job filtering bot sessions. In the last 30 days, it showed 1,513 real sessions and filtered out nearly 1,000 bot visits.
Engagement metrics aligned with my own calculations, with an average engagement of around 49 seconds. Clarity also provided scroll depth, pages per session, smart events, and channel insights.
Even with Clarity, a lot of traffic still shows up as “other.” AI bots don’t send referral signals, so attribution remains a challenge. Still, Clarity helped better separate AI activity from real users than GA alone.
How To Set Up Custom Audiences in GA4
In GA4, you can build custom audiences to isolate human traffic. I created a “direct human” audience using filters such as “direct traffic,” “more than 15 seconds on site,” “more than one page viewed,” and users who came through a browser. This helps surface real visitors.
I plan to build similar audiences for organic and referral traffic. Blanket analytics can’t differentiate AI from humans; you need custom setups to get accurate insights.
Understanding the Rise of Artificial Traffic
AI-driven traffic is increasing, and we can’t rely on standard analytics reports. You need the proper setup and custom filtering to make sense of your data.
Artificial traffic will only grow. Marketers must learn how to separate AI activity from actual visitors to make smart decisions.
If you’re an agency or business owner looking for a partner to understand your analytics better or need help setting this up, please contact us. And until next time, happy marketing.
www.smamarketing.net (Article Sourced Website)
#Traffic #Real #Diagnose #AIBot #Traffic #Google #Analytics #SMA #Marketing