No matching terms. Try another word or clear the search.
- AEO
- Answer engine optimization — improving content so answer engines can understand it and use it as a source. Mentions and citations are outcomes to measure, not guarantees.
- Tracked prompt
- A test question written for a client and sent to selected AI assistants on a schedule. The results show mentions and cited pages. Nobody’s private conversations are read.How Preferium handles this →
- AI referral
- A visit that arrives from an AI assistant’s answer, such as a click from ChatGPT or Perplexity. Counted apart from crawler visits and from tracked answers.How Preferium handles this →
- Search crawler
- A crawler that collects pages for the index an assistant searches when it answers, such as OAI-SearchBot for ChatGPT search or PerplexityBot. A visit does not guarantee indexing or citation.How Preferium handles this →
- Answer-time fetch
- A request an assistant makes while it answers someone, e.g. ChatGPT-User opening a page a person asked about. Triggered by a person, not by a crawl schedule.How Preferium handles this →
- Training crawler
- A crawler that fetches pages for model training, e.g. GPTBot (OpenAI training) or ClaudeBot (Anthropic training). Does not produce a citation by itself.How Preferium handles this →
- Google-Extended
- A control token, not a crawler: a robots.txt name that tells Google whether content its crawlers already fetch may be used for Gemini training and grounding. It does not change Google Search or AI Overviews.
- The edge
- The layer between the website and the visitor where Preferium delivers changes in the HTML response — without touching the CMS.How Preferium handles this →
- Control level
- How much the engine may do on its own for a client site: Full autopilot, Invisible-only or Review everything. Set per client site.How Preferium handles this →
- Health check
- One of 88 automated checks across 13 areas that find what is wrong on a client site, from missing titles to pages AI crawlers cannot read.How Preferium handles this →
- Rollback
- Returning a change to what the page had before. After every deploy a real browser re-checks the title, meta description, H1 and structured data, and a change that made things worse is rolled back automatically; every other change can be withdrawn with one click.How Preferium handles this →
- Kill switch
- One switch that stops the engine for a client site and serves the original site everywhere within about a minute.How Preferium handles this →
- White label
- Reselling the platform under your own brand: your name, logo and colours in the panel, your own dashboard domain, reports and emails from your address, and your own prices through your own Stripe.How Preferium handles this →
- GEO score
- One score from 0 to 100 for AI visibility, built on seven weighted signals. The same calculation for every site, so the numbers can be compared.
- llms.txt
- A file that gives language models a guided overview of the site, in a short and a full version. Served by the engine.How Preferium handles this →
- agents.txt and agent.json
- Machine-readable manifests that describe the site to AI agents — what exists, and where.How Preferium handles this →
- Content signal
- The declaration of what AI may do with the content: show it in search, answer questions with it, and whether it may be used for training.
- IndexNow
- A protocol that notifies participating search engines the moment a page changes, instead of waiting for the next crawl.
- Canonical link
- A tag that identifies the preferred URL when pages have duplicate or very similar content. Search engines treat it as a signal when choosing which URL to index.
- Hreflang
- The links between language and country versions of the same page. Eight checks make sure they point the right way.
- Structured data
- A machine-readable description of what the page is about. The engine writes 31 schema types and merges in what the site already has.How Preferium handles this →
- AI Overview
- Google’s generated summary at the top of the results. Its own surface, with its own sources — tracked separately from AI Mode.
- Cloaking
- Showing machines something different from what people see. Not permitted under the acceptable use policy.
- Parity guard
- The check against cloaking: each optimized page is fetched both as a person and as GPTBot, and any serious difference is flagged.How Preferium handles this →
- Prompt injection
- Hidden text on a page that tries to hijack a language model. Neutralized before the text reaches the AI that writes about the site.