TL;DR
- On average, 2.9 brands in the top five changed when the study moved beyond the New York and English baseline.
- Payroll, accounting, payments and HR recommendations changed more across markets than project management, video conferencing and cloud storage.
- Of the 206 brands recorded, 159 appeared in only one country, while just 14 appeared across all eight countries.
- ChatGPT and Gemini shared an average of 3.2 brands in their top five for the same question and city.
- Only 35% of tested cases kept the same top brand across all five identical runs, showing why a single answer cannot establish consistency.
New York’s payment recommendations were Stripe, Square, PayPal, Helcim and Shopify. Mumbai’s were Razorpay, Cashfree, PayU, PhonePe and Instamojo.
The buying question stayed the same. The two lists shared no brands.
FTA Global’s September 2026 Context Report examines how far AI recommendations move when geography, language and engine change. Across 1,200 answers, the research found distinct local competitors, different platform preferences and variation even when the question was repeated unchanged.
How Did FTA Global Test AI Brand Recommendations?
The study covered ten software categories, ten cities across eight countries, and seven language variants, including Hinglish. Questions ran on ChatGPT and Gemini five times each, producing 1,200 answers.
The buyer remained a small business owner throughout. Questions used neutral sources rather than a brand’s own website, and no FTA client was included.
Holding the buyer persona fixed allowed the study to examine other contextual differences. Edition 01 did not test how recommendations change between buyer roles.
Our explanation of what Context Authority measures provides the background to the metric introduced in the report.
How Much Do AI Recommendations Change Across Cities?
An average of 2.9 of the top five brands changed against the New York baseline.
Berlin showed the greatest average difference, with 3.3 brands changing across the ten categories. London showed the smallest difference at 2.3. Bengaluru recorded 3.0, Delhi 2.9 and Mumbai 2.8.
Stripe’s payment rankings illustrate how sharply a brand’s position can vary. It ranked first in New York, London, Madrid, Paris and Berlin, and second in Mexico City and Dubai. It didn't appear in the displayed top five in Mumbai, Delhi and Bengaluru.
A strong position across several markets therefore did not establish visibility in India within the study.
Which Software Categories Change Most by Location?
The report groups payroll, accounting, payments and HR as money tools. Across markets, an average of 4.4 brands changed in their top five, compared with 1.5 for everyday tools such as project management, video conferencing and storage.
The category breakdown shows how large that difference is.
HR and recruiting had a completely different top five in every other city tested. Project management retained much more of the baseline list.
Geographic stability in one software category could not be assumed for another.
Do Global Brands Dominate AI Recommendations Everywhere?
The study recorded 206 brands. Of these, 159 (about 77%) appeared in the top five in only one country. Just 14 appeared across all eight countries.
The globally recurring names included Zoho, HubSpot, Google Workspace, Microsoft 365, Zoom, Asana and Mailchimp.
Local recommendations told a different story. New York’s accounting list included QuickBooks, Xero, FreshBooks, Zoho and Wave. Berlin’s included Lexware, sevDesk, BuchhaltungsButler, DATEV and Papierkram.
Mumbai’s accounting list contained Zoho, TallyPrime, Vyapar, Busy and myBillBook.
A competitor set built only from the New York results would miss many of the brands recommended elsewhere.
Does Hindi or Hinglish Change AI Recommendations?
The Mumbai payment test compared English, Hindi and Hinglish while keeping the buyer and buying need fixed.
Razorpay and Cashfree remained first and second in all three lists. Hindi introduced Paytm in fourth place, replacing PhonePe. Hinglish retained the English list’s five brands but moved PhonePe above PayU.
Language refined an already local answer. None of the three lists returned to the New York payment recommendations.
The observed changes involved both brand inclusion and ordering, which explains why AI search visibility depends on context beyond prompts.
Do ChatGPT and Gemini Recommend the Same Brands?
ChatGPT and Gemini shared an average of 3.2 brands in their top five for the same question in the same city.
Agreement varied by category. Project management shared 4.1 brands, and CRM shared 4.0. Cloud storage shared only 2.1, while web hosting shared 2.7.
Cloud storage is especially revealing. Its geographic variation was relatively low, yet its overlap between engines was the lowest shown in the report’s platform comparison.
For CRM in Paris, Gemini placed Sellsy first while ChatGPT placed Pipedrive first. The same market and question produced different leaders.
How Consistent Are AI Recommendations Across Repeated Queries?
Only 35% of tested cases produced the same first-ranked brand in all five identical runs.
In the remaining cases, the leading brand changed at least once, even though the question was unchanged.
The finding concerns consistency across five runs. It does not mean a brand had a 35% probability of appearing in any individual answer.
Repeated testing captures whether a position holds. A single result cannot reveal that pattern.
Our comparison of Context Authority and AI visibility shows why consistency matters as much as presence.
Which Local Sources Did AI Cite?
Local domains and media accounted for 42% of cited sources in Germany, 34% in France and 17% in India.
German sources included fuer-gruender.de, trusted.de and datev.de. Indian sources included itforsme.in, technologycounter.com and kryptonforge.in.
Global review sites such as G2 and Capterra also appeared. The report nevertheless found that the mix of cited sources changed with the market, along with the recommended brands.
What Does the Study Establish About Context Authority?
The evidence identifies separate dimensions of AI recommendations: geographic variation, language differences, engine disagreement, repeated answer consistency and local sources.
Context Authority is FTA’s metric for measuring how strongly a brand holds its position across relevant buying contexts. The report places repeated testing and market-specific measurement at its centre.
A brand’s position needs to be read in the markets, languages and engines where it competes. The study shows how much a single baseline can leave unseen.
Do you want more traffic?
.jpg)