Method
What we measure, and how
An audit measures one business across channels, kinds of search, places and languages. Here is exactly what that means.
1. Searches customers actually make
We build the list of searches from what people type for your kind of business — the service, an urgent need, a comparison — plus your own name, in each language your customers speak. You can edit the list at any time.
Each search is made from your area on a mobile device, and we record the first 20 results: whether your site appears, which platforms and directories show your business, and which competitors rank instead. Monthly search volumes, when shown, are estimates from a data provider at country level.
2. Google Maps from many points
Maps results depend on where the phone is. We repeat the same search from a grid of points around your location (or across your service area) and record your position at each point. The result is a map of where you appear in the top 3, lower down, or not at all.
We never recommend showing an address your business does not staff. Advice about address settings follows Google's rules for shops versus businesses that travel to customers.
3. AI assistants, sampled
We ask the questions a customer might ask an assistant, through official APIs with web search enabled: OpenAI, Anthropic and Perplexity. Each question is asked more than once in a fresh session because answers vary. We report how often you were named, the uncertainty of that rate, which sources were cited and who was named instead.
API answers are not identical to what the consumer apps show, and Google Gemini with Search grounding is not tested because its terms do not allow analysing grounded answers. Reports say this explicitly.
4. Reviews and platforms
We record your Google rating and review count and compare them with the businesses that appear around you. Other platforms are recorded as they appear in search results. We do not scrape review sites.
5. Your website, page by page
Our crawler reads your public pages politely: it respects robots.txt, waits between requests and stops at a page budget. It checks more than 50 things, grouped as technical health, mobile speed (Google Lighthouse and, when available, real-user data from the Chrome UX Report) and conversion — can a visitor on a phone call, message or book in one tap?
About FarolumaBot
Our crawler identifies itself as FarolumaBot and only visits a website when an audit of that business was requested. It follows robots.txt (user-agent token farolumabot, including Crawl-delay), loads one page at a time with a pause in between, slows down when a server answers 429 or 503 and stops after the third such answer. To keep it out, add User-agent: farolumabot and Disallow: / to your robots.txt.
What we never do
- Invent data. If something could not be measured, it is shown as not measured, with the reason.
- Turn estimates into facts. Estimates are labelled as estimates.
- Let an AI write findings. Findings come from measurements and fixed rules; AI may only rephrase them, and its text is checked against the evidence.
- Promise rankings or AI mentions.
- Show history we don't have. Trends start from your first measurement.