Brand Visibility in AI search became a practical concern for Stakewise after its referral traffic from large language models grew during 2025. The New Zealand-based investment platform saw ChatGPT send roughly 700 referral sessions each month by midyear, up from only a handful a year earlier, according to research published October 13, 2026.
The visitors arriving through ChatGPT had lower bounce rates and viewed more pages per visit than visitors from other referral sources. Yet Stakewise’s conventional search measures, including keyword rankings, impression volume and organic click-through rates, could not show what AI systems told prospective customers before they clicked.
Brand Visibility Required an Answer Audit
The company’s digital strategy team began querying ChatGPT with questions real customers asked about fees, fund choices and comparisons with competitors. The review found confident answers that were often wrong. One repeated response said Stakewise offered five funds in its KiwiSaver product, although the actual number was seven.
Other responses mischaracterized fee structures, described a protective investment safeguard as restrictive, or combined features from the platform with a separate product. For a financial services provider, the mistakes meant prospective customers could encounter inaccurate information while researching retirement savings options.
The researchers describe two challenges for generative engine optimization, or GEO: whether a brand appears in AI answers and whether those answers represent it accurately. GEO is also called answer engine optimization, or AEO. The article says AI-generated responses can replace direct links on search results pages, while conventional analytics record arriving traffic but not the conversation that preceded a visit.
Brand Visibility Has Two Measures
Stakewise’s example separates referral volume from the content of the referrals. Its analytics showed ChatGPT visits increasing, and those visitors showed lower bounce rates and more pages per visit than other referral visitors. Those measures described activity after a click; the team still needed to inspect answers about the business itself.
The research says teams can ask the questions customers actually ask, then review how AI describes the brand. For Stakewise, those questions covered fees, available funds and comparisons. The audit exposed errors across the fund count, fee descriptions, a safeguard and distinctions between products. The account does not report a corrected answer or a later measurement, so the audit is the reported development.
The research frames this as a measurement problem: as zero-click LLM responses replace direct links on search results pages, new metrics are replacing traditional traffic analytics.
Readers tracking the broader traffic question can also see coverage of AI search and website visits. A separate report examines AI Overviews in brand searches, while this guide discusses answer engine optimization for businesses.
The researchers frame visibility and accuracy as distinct strategic challenges as generative AI search grows. Stakewise’s team moved from watching referral sessions to examining the answers customers might receive. The article’s account leaves the practical next step clear: audit those answers alongside traffic measures to see both whether AI brings visitors and what it tells them first.
The source describes the case and its findings in its research on brand visibility, published October 13, 2026.
