A digital marketing agency had no systematic way to track whether their clients' brands appeared in AI-generated answers from ChatGPT, Gemini, and Perplexity. Kovil AI built a GEO monitoring platform that executes 50 AI calls per scheduled run across multiple LLMs, detects domain mentions, stores citation URLs, tracks visibility trends over time, and runs on a fully automated schedule.
50 AI calls
Per scheduled run
Across web + non-web LLMs
5×5
Keyword-query matrix
25 tracked queries per client
4 LLMs
Monitored simultaneously
ChatGPT, Gemini, Perplexity, Claude
Zero effort
After initial setup
Fully automated scheduling
Tech Stack
"We could see our clients ranking on page one of Google but disappearing entirely from AI answers. The GEO platform turned that into a measurable, trackable metric — and it became the foundation of an entirely new service line for the agency."
Generative Engine Optimisation — GEO — has emerged as a critical discipline alongside traditional SEO. As AI-powered answers from ChatGPT, Perplexity, Google AI Overviews, and Gemini answer more search queries directly, whether a brand appears in those answers has become a real commercial concern for marketing teams. Brands with strong organic search rankings are finding they have little or no visibility in AI-generated responses to the same queries — and vice versa.
The client is a digital marketing agency managing brand visibility for a portfolio of B2B clients. They had noticed several clients' competitors appearing consistently in AI-generated answers for high-intent industry queries — but had no systematic way to track this visibility, measure changes over time, or attribute appearances to specific content or citation sources. They approached Kovil AI to build a monitoring platform that could answer a commercially important question: is our brand appearing in the AI answers that matter for our clients' target queries?
The GEO Monitoring Platform tracks brand visibility across AI-generated answers from multiple LLMs, using a configurable keyword and query matrix that each client manages independently.
Each monitored brand can configure up to 5 keywords, each with up to 5 distinct query formulations — creating a matrix of up to 25 tracked queries per client. Keywords represent core topics (e.g. "project management software") while queries represent specific questions an AI user might ask (e.g. "what is the best project management software for remote teams?"). Each query can be individually enabled or disabled, and run frequency is configurable per client account.
Each scheduled run executes up to 50 AI calls in parallel — querying multiple LLMs across the full query matrix. The platform covers both web-browsing models (where the AI cites live sources) and non-web models (where answers are drawn from training data). Calls are batched to stay within API rate limits, with retry logic for failed calls and structured parsing of every response to extract domain mentions and citation URLs.
For each AI response, the platform detects whether the client's domain appears — as a direct citation, a named brand mention, or a URL reference in the answer. Detection uses both exact and fuzzy matching to handle variations in how AI models reference brands. Every mention is stored with full context: the query, the model, the position of the mention, and the cited URL if present.
The dashboard gives the agency's team — and, via a white-labelled view, their clients — a time-series view of visibility trends per keyword and per LLM. Key metrics include mention rate (percentage of runs that detected a mention), first-seen and last-seen dates per query, source breakdown (which URLs are cited when the brand appears), and a side-by-side comparison against up to two configured competitors.
Runs can be triggered manually or scheduled at daily or weekly intervals per client. Scheduled runs execute automatically — Slack and email notifications are sent when a run completes, with a summary of visibility changes since the last run. Zero human effort is required after initial setup.
The platform gave the agency a capability they could not find anywhere else: systematic, repeatable tracking of AI answer visibility across multiple models, with trend data that made it possible to measure the impact of content changes on AI appearances over time.
Within the first month of using the platform, the agency identified that three of their clients were appearing in fewer than 10% of monitored AI answers — despite having strong organic search rankings. This gap between SEO visibility and GEO visibility became the basis for a new service offering: GEO optimisation sprints, where targeted content changes were made and the platform was used to measure impact on AI appearances over 60-day windows.
For one B2B SaaS client, a structured GEO programme increased AI answer visibility from 8% to 41% of monitored queries over 90 days. The platform made that result measurable, attributable, and demonstrable to the client — turning an emerging discipline into a quantified, repeatable service.
Start Your Project
See the engagement model that fits your situation.