Chatbots, Assistants & AI Search: Tracking AI-Driven Website Traffic

Plus, Eight Ways to Analyze LLM-Referred Visitors With TWIPLA

Chatbots, assistants, and AI search are changing how people discover websites since these tools often bypass traditional search and referral patterns.

This is no small thing, with IT consultancy Gartner predicting that organic website traffic from search engines will fall by around 25% between now and 2026 as more internet users turn to these Large Language Models (LLMs).

Businesses must adapt to this new reality by rethinking both their website optimization strategies and how they analyze traffic.

As a website intelligence provider, we've been closely monitoring the situation.

Below, you'll find insights into this technological shift, along with practical guidance on how analytics platforms - specifically TWIPLA - can be adapted to track and analyze AI-driven traffic referrals.

From Search to AI: The Rise of AI-Driven Referrals

Despite data privacy concerns, ChatGPT, Claude, Copilot, and Gemini are transforming how visitors find and access websites.

These AI chatbots and assistants act as direct referrers, offering different link suggestions than traditional search engines and shaping intent before they even visit a site.

AI search tools like Perplexity AI and Google AI Overviews are having a similar impact, pulling information from multiple sources and linking directly to relevant webpages.

Adoption of these technologies has exploded in recent years.

According to Ahrefs, 63% of websites are receiving traffic from AI-driven sources while Search Engine Land claim that over 70% of people now use AI search.

This isn't to say that traditional search is dead, but these technologies are disrupting traditional referral models - making it essential for businesses to optimize their content to capitalize on the opportunities provided by AI-driven website discovery.

How AI-Driven Sources Are Changing Website Referral Tracking

Adapting to these new referral traffic patterns means measuring their impact, identifying which AI tools are driving visitors to your website, and understanding how this traffic behaves.

These new patterns are also both cause and consequence of a broader shift in behavior towards informational search.

AI agents are increasingly providing direct, comprehensive answers within their own interfaces, meaning many internet users never leave the AI’s UI to visit external websites. This trend reduces traditional search-driven web traffic, forcing businesses to rethink how they structure and distribute content to ensure it is surfaced by AI models.

Conversely, navigational and transactional searches still generate direct website visits. When people seek a specific product, or service, AI-powered assistants and search tools often provide direct links, making AI-generated referrals an increasingly important traffic source.

Tracking AI-Driven Traffic: A New Challenge for Analytics

Analyzing AI-driven website referrals presents a unique challenge for businesses.

For example, when people click links from ChatGPT or Perplexity AI, the referral source may appear as a distinct AI referrer - or, in some cases, as direct traffic if no explicit referral data is passed. AI-generated responses often lack structured referral information, making it harder to classify and measure their impact.

To track and analyze LLM-referred visits, website analytics platforms must go beyond traditional referral tracking and incorporate segmentation, behavioral insights, and engagement analytics.

This can be achieved through a combination of:

Referral Tracking: Identifying AI-driven traffic sources.
UTM Parameters: Isolating AI-generated visits from other referral sources.
Behavior Analytics: Understanding post-click engagement through session recordings and interaction metrics.

Additionally, monitoring engagement trends, conversion pathways, and drop-off points allows businesses to assess how AI-referred visitors interact with their sites.

Using TWIPLA to Analyze AI-Driven Traffic

TWIPLA provides a complete suite of tools to help businesses holistically isolate, measure, and interpret AI-driven traffic. From tracking referral sites and UTM parameters to analyzing visitor behavior and eCommerce impact, businesses can gain deeper insights into how AI-generated visitors engage with their sites.

In the following section, we’ll explore eight ways TWIPLA empowers businesses to analyze AI-driven referral traffic, understand visitor behavior, and refine their marketing strategies:

Use Referral Sites in Pages to track AI-driven sources.
Filter the Master or Pages Dashboard by AI Referrer for a more focused view.
Filter Traffic by UTM Parameters to isolate AI-generated visitors.
Watch Session Replays for engagement and behavioral insights
Monitor Alarming Behavior Events to identify visitor frustrations.
Calibrate eCommerce Stats to measure AI-driven customer interactions.
Create a Custom Dashboard for one or multiple referrers.
Adopt TWIPLA's upcoming AI search optimizatioin guidance tool.

With these tools, businesses can track AI-generated visitors and turn AI-driven discovery into an advantage.

Let's take a closer look at these eight TWIPLA use cases.

Analysing AI-Driven Website Traffic Sources in Pages — This list is sorted by visitor count and shows the percentage of total sessions, helping you quickly identify key traffic sources - including AI-driven referrals.

#1 Use Referral Sites in Pages

Jump into the Referring Sites tab in Pages to see a breakdown of where your traffic is coming from over a selected time period:

Analysing the Pages Visited by AI-Driven Website Traffic — This list shows the pages that ChatGPT-referred visitors accessed, ranked by the number of visits.

As shown in the screenshot, ChatGPT and Perplexity AI drove 497 website sessions in the selected period of time.

Identifying which AI referrers drive the most website traffic helps you understand how these tools reference your content. Since ChatGPT and Perplexity AI generate responses differently, tracking referrers shows which AI models favor your site. You can then refine your content strategy to maximize visibility in the AI tools that are proven traffic drivers.

Clicking on these sources reveals a list of the specific pages visitors viewed during their sessions:

Knowing which pages AI-referred website visitors view helps your business identify the content these tools are surfacing and how people engage with it. If certain pages are attracting significant AI-driven traffic, optimizing them can improve visibility in AI-generated answers - helping you attract more high-intent visitors from these sources.

Tracking Ai-Driven Website Traffic Metrics — You can save multi-layer filters as templates, making it faster and easier to apply them in the future.

#2 Filter the Master or Pages Dashboard by AI Referrer

TWIPLA's dashboards gather insights from across the website intelligence platform, offering a deeper understanding that complements Pages data.

Let’s start with the Master Dashboard, where you can apply one or more referral site filters to multiple report blocks.

You can also dive deeper into the data by layering additional filters, allowing you to analyze AI referral traffic from multiple angles:

Analyze Top Webpages and Bounce Rates for AI-Driven Traffic — Click into either of these report blocks to go deeper into the data.

As shown in the screenshot, the Page Views, Sessions & Visitors report block displays the following key website metrics:

Overall Page Views.
Overall Visitor Sessions.
New Visitors.
Returning Visitors.
Total Visitors.
Converting Visitors.

These metrics help you assess the impact of AI-driven traffic. For instance, tracking new vs. returning visitors shows whether AI referrals generate one-time clicks or recurring engagement, while conversion data reveals if this traffic leads to meaningful actions.

The Master Dashboard includes two more report blocks that are filterable by referral site:

Analyzing AI-Driven Webpage Traffic Data — Select one or more pages to visualize any metric in graph form.

While the Overall Page Views report block mirrors data from Referral Sites in Pages (discussed in Option 1), Bounce Rate is an additional and crucial metric for understanding whether this visitor category engages with content or leaves immediately.

A high bounce rate may indicate that AI-generated answers provide too much information upfront, while a low bounce rate suggests AI-referred visitors find the content valuable. By analyzing this metric, you can optimize landing pages to better retain AI-driven traffic and encourage deeper engagement.

Next up, let's look at the Pages Dashboard.

This report lists all your website pages and can be filtered by referral site to analyze performance insights for AI-driven traffic:

As shown in the screenshot, the Pages Dashboard displays the following page metrics:

Page Views.
Percentage of Total Website Page Views.
Direct Page Views.
Visitors.
Average Page View Duration.
Outgoing Visitors.
Bounce Rate.

Each of these metrics provides a different lens on AI-referred visitors, helping you measure reach, engagement, and behavior.

Page Views and Direct Page Views highlight how frequently AI-driven traffic lands on specific content, while Visitors indicate the unique audience size. Average Page View Duration helps assess whether AI-referred visitors stay to consume content or leave quickly. Lastly, Outgoing Visitors reveal whether they click external links, helping you track post-visit behavior.

Together, these insights allow you to fine-tune content, enhance visitor experience, and adapt to AI-driven discovery patterns.

Using UTM Parameters to Isolate AI-Driven Traffic Data — Select one or more pages to visualize any metric in graph form.

#3 Filter Traffic by UTM Parameters

Filtering traffic by UTM parameters is an effective way to gain focused insights into AI-driven visitors.

Notably, when ChatGPT includes a link in its responses, it often appends a UTM source tag (e.g., utm_source=chatgpt.com). Other AI tools may also include UTM tags to facilitate tracking.

Filtering by UTM Source in the Pages Dashboard or elsewhere in TWIPLA allows you to isolate and analyze AI-driven traffic more effectively:

As shown in the screenshot, activating the UTM Source filter enables you to examine the behavior, engagement, and conversion potential of this visitor segment.

While filtering by referral URL allows you to track visits from AI tools, filtering by UTM Source provides more granular insights by isolating specific referral instances.

Since AI-generated links often include UTM parameters, this method ensures that visits from ChatGPT or other AI-driven sources are recognized even if they don't pass standard referral data. This is particularly useful for distinguishing AI-driven visitors from direct traffic, allowing for more precise attribution.

This enables you to compare AI-driven visitors with other traffic sources, measure the true impact of ChatGPT referrals, and refine your content strategy accordingly.

Using Session Replays to Understand AI-Referred Visitor Behavior — Adding additional filters will let you dig even deeper into the specific website issues faced by different AI-referred visitor segments.

#4 Analyze AI-Driven Visitor Behavior with Session Recordings

While tracking referral data helps you understand where AI-driven visitors come from, TWIPLA Session Recordings provide deeper behavioral insights, revealing how these visitors engage with your site.

In Overview, you can filter videos by one or more AI referral sources and watch how AI-driven visitors interact with your content and move around the site during a session:

This filtering system lets you quickly create a curated list of session recordings for AI-driven referral visitors, allowing you to:

Observe navigation patterns: See whether AI-referred visitors explore multiple pages, follow expected visitor journeys, or behave differently from search or social media traffic.
Measure engagement: Determine whether these visitors read content, interact with CTAs, or leave quickly.
Identify friction points: Spot usability issues, dead clicks, or hesitation that may indicate confusion.
Analyze drop-offs: Understand where AI-driven visitors abandon their sessions and compare their behavior to other referral sources.

Additionally, by filtering recordings by multiple referral URLs simultaneously, you can compare AI-driven traffic to organic search, paid ads, and social media, uncovering patterns that other analytics tools might miss.

Identifying Alarming Behaviors in AI-Referred Visitors — Additionally, each recording includes timestamps for alarming behaviors, allowing you to jump directly to specific behavior types of interest.

#5 Monitor Alarming Behavior Events for AI-Driven Visitors

A key differentiator of TWIPLA’s Session Recordings is the ability to track Alarming Behavior Events for AI-referred visitors. Unlike other tools that simply show referral data (GA4) or offer filtering for session replays (Hotjar), TWIPLA combines both:

As you can see from the screenshot, the filtered Session Recordings overview provides a snapshot of any alarming behavior events that have been identified in AI-referred visitor sessions, including:

Rage Clicks: Website visitors rapidly clicking on elements, indicating confusion or frustration.
Dead Clicks: Clicks on non-interactive elements, suggesting unclear UI.
Excessive Scrolling: Visitors struggling to find relevant information.
Rapid Page Reloads: Indicating poor UX, slow load times, or unfulfilled expectations.

This extra layer of behavioral insight makes TWIPLA uniquely positioned to help you understand not just where AI-driven visitors come from, but also how they behave and what pain points they experience.

eCommerce Metrics on AI-Driven Traffic — Activate additional features to dig even deeper into your eCommerce data.

#6 Calibrate eCommerce Stats for AI-Driven Customer Insights

AI-driven referrals are increasingly shaping online shopping behavior, making it essential for eCommerce businesses to track their impact on conversions and customer journeys.

With TWIPLA’s eCommerce Statistics, this can be done in several ways.

Firstly, Overview data can be segmented using one or more referral URL filters:

Product-Related Metrics on AI-Driven Traffic — See how your products perform in the eyes of AI-referred customers.

As shown in the screenshot, Overview provides the following AI-driven traffic eCommerce KPIs for the selected time period:

Overall Visitor Sessions.
Total Customers.
Total Orders.
Sold Products.
Gross Revenue.

These AI-driven eCommerce metrics provide critical insights into how visitors from AI tools behave and whether they contribute to revenue growth.

Overall Visitor Sessions indicate the total number of AI-referred visitors, helping you gauge how much traffic is coming from chatbots, assistants, and AI search tools. If sessions are high but sales remain low, you may need to adjust your approach to better convert AI-driven intent into sales.

Total Customers and Total Orders reveal whether AI-referred visitors are converting into paying customers. A high number of sessions with low conversions could indicate that AI-generated visitors are more informational than transactional, and mean that you should refine your content strategy or promotional offers to increase sales.

Sold Products highlights which specific items resonate with AI-driven shoppers. If certain products consistently attract AI-generated traffic, you can optimize descriptions, adjust pricing, or sustain availability to better capitalize on demand.

Gross Revenue quantifies the direct financial impact of AI-driven traffic. By comparing revenue from AI referrals to other sources, you can measure the profitability of this traffic segment and refine marketing, pricing, and inventory strategies accordingly.

Taken together, insights like these will help you to fine tune your approach to AI-driven traffic, ensuring that you can maximize both conversions and revenue potential.

Next up are the Product-Related, Sales-Related, and Orders & Events by Traffic Channels subsections of Sales Charts, which can all also be filtered by one or more AI referral URL:

Product-Related

In the Product-Related subsection of eCommerce Statistics, all the report blocks can be filtered by AI-referral URL to analyze LLM-driven webstore activity:

Sales Metrics on AI-Driven Customers — See sales-based metrics for AI-referred customers.

Together, these blocks provide the following eCommerce metrics:

Revenue by Product.
Units Sold per Product.
Product KPIs (name, quantity sold, SKU, category, views, wishlist additions, cart removals, cart additions).
Product Lists, with number of views.

These metrics provide valuable insight into how AI-referred visitors engage with products, helping businesses optimize pricing, marketing, and inventory strategies.

Analyzing Revenue by Product (€) helps determine whether AI-referred visitors are making high-value purchases or primarily engaging with lower-margin items. Identifying these patterns can help you to prioritize profitable products and refine their positioning to attract the right customers.

Similarly, tracking Units Sold Per Product reveals which items are most frequently bought by AI-driven visitors, offering actionable insights for inventory planning and marketing optimization. If AI referrals consistently drive demand for specific products, you can strategically promote, bundle, or upsell them to increase conversions.

Product KPIs offer a more detailed breakdown of AI-driven shopping behavior by tracking specific product interactions.

Metrics such as quantity sold, SKU, and category help you understand demand trends, while views, wishlist additions, and cart interactions indicate whether AI-referred visitors are actively considering a purchase. A high number of wishlist additions or cart removals could signal pricing concerns or hesitation, prompting you to adjust discount strategies, improve product descriptions, or streamline the checkout experience to reduce drop-offs.

Product Views provide insight into whether AI-referred customers are seriously considering purchases or dropping off before completing a sale. A high number of product views with low conversions may signal a need for better pricing strategies, more compelling descriptions, or stronger CTAs to encourage purchases.

Sales-Related

The Sales-Related subsection of eCommerce Statistics can equally be filtered by one or more referral URLs to analyze AI-driven customer behavior:

Orders and Events Insights for AI-Driven Customers — Learn how to analyze orders and events around AI-referred traffic.

As shown in the screenshot, this submodule provides the following key insights into AI-referred visitors:

Sales Conversion Rate.
Overall Shop Sessions.
Number of Purchases.
Checkouts Initiated.
Cart Views.

Filtering Sales-Related metrics by AI-referral URLs helps businesses assess whether AI-driven visitors actively engage with their online store and, more importantly, whether they convert into customers.

Sales Conversion Rate, Overall Shop Sessions, and Number of Purchases are all shown in the Ratio Sessions/Sales report block. This graph allows you to visualize key trends, and quickly identify whether AI-driven traffic translates into meaningful sales volumes. If overall sessions from AI referrals are high but purchases remain low, it may indicate that AI-generated visitors are browsing but not converting, prompting a need for better product positioning, stronger CTAs, or pricing adjustments.

Checkouts Initiated and Cart Views provide deeper insight into purchase intent. These report blocks, which can be expanded for more detailed analysis, reveal whether AI-referred visitors are adding products to their cart and progressing toward checkout. If Checkouts Initiated are low compared to Cart Views, this could suggest that users are interested but hesitant to complete their purchase, potentially due to unclear pricing, complex checkout flows, or lack of trust signals.

Orders & Events by Traffic Channels

Finally to the Orders & Events by Traffic Channels subsection of eCommerce Statistics, which can also be filtered by referral URL to report on AI-driven traffic sources:

As you can see from the screenshot, this Orders & Events by Traffic Channels subsection provides a breakdown of how AI-driven visitors contribute to eCommerce orders and key shopping events.

Filtering this data by AI-referral URLs lets you compare AI-driven purchases with those from other traffic sources, such as organic search, social media, or paid ads.

This submodule consists of two key report blocks:

Number of Orders by Traffic Channel: This report shows the total number of completed purchases per traffic source, visualizing whether AI-driven referrals result in actual transactions. By analyzing AI-attributed orders alongside other channels, you can determine if AI-generated traffic is a strong revenue driver or primarily informational traffic.
Number of Orders Over Time: This graph provides a time-based view of AI-driven purchases, allowing you to track seasonal trends, campaign performance, and fluctuations in AI-generated orders. If AI referrals show a spike in purchases after a chatbot or AI search recommendation, try adjusting your promotions, content strategy, or engagement tactics to maximize these traffic surges.

Taken together, these insights into order trends will help you to evaluate the long-term sales impact of AI-driven referrals and optimize your acquisition strategy to drive more high-value customers from AI-powered sources.

Custom Analytics Reports for AI-Driven Traffic — Use Custom Dashboards to set up reports that apply all the AI-referral tracking methods outlined above, so they're ready to use whenever you need them.

#7 Create a Custom Dashboard for One or Multiple AI Referrers

If AI-driven traffic is a growing focus, a Custom Dashboard in TWIPLA can streamline analysis by consolidating key AI referral insights in one place.

You can choose to create separate boards for each AI referrer, or design a single dashboard that aggregates all AI-driven traffic insights for broader comparisons.

Custom Dashboards can incorporate any of the elements outlined in this article, including Dashboards, Pages, eCommerce Statistics, Alarming Behavior Tracking, and Session Recordings.

This will allow you to tailor TWIPLA to AI-related visitor behavior, and they can be set up with pre-activated filters for ease of use.

Beyond AI-specific insights, Custom Dashboards offer all the standard benefits of TWIPLA’s dashboarding system - from speeding up analytics workflows to controlling access to sensitive data. By centralizing AI referral data in a dedicated dashboard, you can improve efficiency, enhance reporting, and make faster, data-driven decisions.

#8 Adopt the Upcoming AI Search Optimization Tool

We’re currently building an AI search guidance tool to help businesses improve their visibility in LLM-generated responses and ensure AI-driven traffic is accurately tracked.

Many AI-powered search tools and assistants do not execute JavaScript, relying instead on server-rendered HTML to fetch content. This means that pages relying on JavaScript for key content may be invisible to AI search engines, leading to lost visibility and missed referral opportunities.

Most analytics platforms also depend on JavaScript-based tracking, meaning they fail to detect AI-generated visits accurately. This creates a blind spot in referral tracking, making it difficult to assess how AI-driven sources contribute to website traffic and engagement.

TWIPLA’s tool solves this by identifying pages that AI search engines may overlook, allowing businesses to:

Fix hidden SEO gaps by ensuring critical content is accessible to AI search engines.
Recover lost AI-driven referral traffic that would otherwise go untracked due to JavaScript limitations.
Optimize for AI discovery and maintain visibility in AI-generated responses.

By addressing both content visibility and analytics accuracy, this tool helps businesses stay competitive in the evolving AI search landscape.