The rise of generative search engines and conversational AI assistants has created a massive blind spot for digital marketers. While consumers increasingly rely on tools like ChatGPT, Google Gemini, Claude, Perplexity, and Grok to discover local businesses, connecting those AI-driven interactions to actual bottom-line revenue has proven notoriously difficult.
During a recent SEJ Live session, industry experts Sean McCrohan, Vice President of Technology at CallRail, and Steve Wiideman, an enterprise SEO advisor specializing in multi-location brands, tackled this exact challenge. Diving deep into the realities of modern search behavior, they explored the metrics that actually matter, the technical hurdles of tracking AI traffic, and the strategies local businesses must adopt to stay competitive in an AI-first world.
Main Facts: The Current State of AI Referral Traffic
At its core, the intersection of AI assistants and local lead generation is characterized by small, measurable metrics and immense untapped potential. According to data shared by McCrohan, AI citation clicks currently account for roughly 1% to 2% of total inbound calls for CallRail’s customer base. While that percentage may sound small, it represents a doubling of volume since January, signaling a steady, upward trajectory.
The primary takeaways defining this shift include:
The Measurement Gap: Traditional advertising networks provide robust, native tracking tools. In contrast, AI vendors have been reluctant to share granular data or develop advanced ad ecosystems for conversational interfaces, leaving marketers to rely on basic referral clicks and self-reported attribution.
The 24/7 Nature of AI Discovery: Traditional website traffic typically plateaus outside of standard working hours. However, data from CallRail shows that AI-driven search traffic peaks later in the day, with nearly two-thirds of interactions occurring outside regular business hours—stretching well past midnight.
Technical Tracking Constraints: Standard in-browser phone number swapping scripts fail to capture AI crawler activity because these bots typically do not execute JavaScript. Consequently, marketers must implement more complex server-side handling to accurately track AI-driven touchpoints.
Chronology: How the Conversation Around AI Search Has Evolved
To understand where local search optimization is heading, it helps to look at how the conversation has evolved over the past year:
Early January: AI citation clicks sat at a fraction of their current volume, treated largely as a novelty by multi-location and enterprise brands. Most web analytics tools struggled to categorize referral traffic coming from conversational platforms properly.
Mid-Year (The Shift): As platforms like ChatGPT, Gemini, and Perplexity integrated real-time web browsing and local intent into their core functionalities, businesses began noticing a disconnect between traffic reports and actual customer inquiries. CallRail’s internal data revealed a steady 65% growth trajectory over a compressed six-month window, ultimately doubling by late summer.
August 26 (SEJ Live Session): McCrohan and Wiideman took to the virtual stage to demystify AI visibility. McCrohan updated previous projections, highlighting that while growth is undeniable, the channels remain standard discovery tools rather than direct, trackable sales platforms.
Present Day: Brands are shifting away from chasing elusive, single-source vanity rankings. Instead, they are focusing on robust data governance, cross-platform reputation management (beyond just Google Business Profile), and mining first-party call and chat transcripts to align their content with actual customer language.
Supporting Data and Analytics: What the Numbers Tell Us
For marketers attempting to build a business case for AI optimization, the data presents a nuanced picture. When filtering Google Analytics 4 (GA4) for activity originating from ChatGPT, Gemini, Claude, Perplexity, and Grok, Wiideman’s multi-location and franchise clients see traffic hovering around the 1% mark. Rather than viewing this as a failure, Wiideman characterizes it as entirely normal for an emerging top-of-funnel discovery channel.
Additional data points shaping the landscape include:
Audience Awareness: During a live poll conducted during the session, approximately one-quarter of the audience reported that they could definitively tell when an assistant sent a lead, while nearly half were unsure. McCrohan noted that the "yes" responses were higher than anticipated, indicating that a subset of businesses are actively listening to call recordings and noticing new intake patterns.
Prompt Drift and Consistency: Data from Steady Demand highlights the volatile nature of generative search. When repeating the exact same local search query in Gemini, overlapping cited sources occur only 40% of the time, and the exact same top-recommended business appears in just 7% of results. This starkly contrasts with traditional Google local packs, which boast a roughly 90% consistency rate.
The Review Threshold: Drawing on insights from local SEO experts like Darren Shaw, Wiideman emphasized that recommendation engines heavily weigh aggregate review scores. Businesses failing to maintain an average score between 4.5 and 4.7 stars across platforms see a sharp drop-off in their chances of being recommended by an AI assistant.
Official Responses and Expert Insights
Both McCrohan and Wiideman offered critical perspectives on how businesses should adjust their technical setups and marketing strategies to accommodate AI bots.
Overcoming Technical Limitations
McCrohan stressed that standard in-browser scripts are useless for tracking AI agents because crawlers do not execute page scripts.
"The crawlers that AI agents are using to retrieve this data do not execute scripts on your page when they go to look at your website. They just don’t," McCrohan explained.
To combat this, he advocates for server-side number swapping. While more technical to implement, server-side handling ensures that AI bots and crawlers properly register tracked assets. However, Wiideman issued a word of caution regarding user-agent tests, reminding marketers to remain vigilant against accidental "cloaking" violations. Ensuring that Name, Address, and Phone (NAP) data remains strictly consistent across Maps and canonical local pages is essential to avoid algorithmic penalties.
Mining First-Party Data
A recurring theme of the session was the untapped goldmine sitting inside existing customer interactions. McCrohan noted that call transcripts capture the raw, unscripted language customers use to describe their pain points. If a business relies solely on corporate jargon while consumers search using conversational, problem-focused phrasing, the brand will inevitably miss out on AI citations.
Wiideman echoed this sentiment regarding site chat logs. Many businesses overlook initial prompts typed into website chat widgets. By utilizing chat tools that highlight specific keyword conversations, brands can quickly reverse-engineer customer intent and build more effective content roadmaps.
Implications for Local Businesses and Multi-Location Brands
The transition from traditional search engine result pages (SERPs) to generative AI answers carries profound implications for digital marketing strategies:
The Death of the Single-Keyword Strategy: Because of "prompt drift," optimizing for a single, static keyword or ranking position is no longer sufficient. Brands must instead focus on semantic triples—specific claims and factual attributes they want to be recognized for—building a comprehensive "prompt library" of 100 to 125 entries to monitor baseline trends.
The Imperative of Centralized Governance: Whether managing 40 locations or 4,000, multi-location brands must enforce centralized data control. Rogue listings, unverified manager profiles, and inconsistent schemas dilute a brand’s authority, making it difficult for an AI assistant to confidently recommend a local branch.
Optimizing for Ecosystems, Not Just Google: AI assistants pull data from a diverse web ecosystem. Wiideman urges brands to diversify their review-generation efforts across Yelp (which feeds Apple Maps, Bing, and ChatGPT integrations), TripAdvisor, and Reddit, rather than relying solely on Google Business Profile.
Embracing 24/7 Operations: With nearly two-thirds of AI-driven research occurring outside normal business hours, missing a call or failing to provide an immediate digital touchpoint means losing the lead to a competitor. As McCrohan bluntly advised: "Get something [like a voice agent], because more and more of this business is a 24-hour business."
Looking Ahead
As conversational AI continues to evolve, the consensus among industry leaders is that the future will likely feature autonomous agents that not only verify contact numbers for users but execute calls and bookings on their behalf—often without the consumer ever visiting the target website.
Despite these seismic shifts, marketers should avoid panicking. As McCrohan succinctly summarized:
"It is an exciting new world, but it is not a 100% new world. There is carryover. It will be OK."
By focusing on robust data governance, maintaining stellar multi-platform reviews, implementing server-side tracking, and listening closely to the actual words customers use, local businesses can position themselves to win citations and drive revenue in the age of AI search.
