Frequently Asked Questions

SEO optimizes ranking on search result pages, where users click through to you. GEO optimizes the answer AI gives directly — users may never click a link and still make a decision based on that answer. The techniques overlap (structured data, authoritative sources, content quality) but the goals differ: SEO is measured by ranking, GEO by being mentioned and recommended by AI.

Yes, but it is not a rebuild. The structured data, authoritative sources and content quality you already invested in SEO carry straight over — that foundation is shared. GEO adds three things: writing content in the shape AI prefers to cite (clear definitions, verifiable numbers, comparisons and steps); placing trustworthy information on third-party sources beyond your own site (AI often cites sources that are not your website); and measuring "mentioned and recommended in AI answers" instead of rankings. Put simply: SEO opens the door, GEO is what makes AI willing to say your name.

The check report is visible immediately. Systematic improvement depends on industry competitiveness, content supply and knowledge-base quality; monthly reports show the trend without promising a timeline.

The free check is a snapshot: whether your brand is currently mentioned by mainstream engines and roughly where it stands — enough to judge whether there is a problem. The paid plan is continuous monitoring: fixed question sets sampled on a schedule, competitor comparison, negative-mention alerts, source tracing, plus the content production and website rework capabilities to act on what is found.

The platform can be configured for domestic models only; calling overseas models requires your explicit authorization and is logged. Knowledge base documents are vectorized locally and used only for retrieval and writing references within your own account. Tenants are strictly isolated — even agencies cannot see another agency's client data.

Every sample records how it was captured — official API, UI simulation, manual entry or a third-party source — and the report and analytics console disclose the source composition up front. Different capture methods are never merged into one trend line: averaging official API results with manual entries yields a curve that looks perfectly normal, corresponds to a measurement that does not exist, and will suddenly jump the day you switch capture methods. The same applies across markets: a market we do not cover is labeled "not covered" rather than filled in with data in another language.

Yes. The platform separates a domestic edition from a cross-border edition: the domestic edition monitors Chinese engines and distributes to Chinese platforms; the cross-border edition connects to OpenAI, Perplexity, Gemini, Claude and overseas content platforms. Overseas sampling runs through an overseas egress node, and the two markets are metered separately. Each sample records which egress it used — different egress regions return different regional search results, so the two sets of samples cannot be compared directly.

Yes. The platform uses "market packs" as the unit of market: each pack pre-loads the market's language and variants, its mainstream AI engines, a source matrix derived from real citation data, its legal jurisdictions and its marketing calendar. The same brand is set up as separate projects per market — Germany measured with German questions against European sources, Southeast Asia with local languages — and the data is reported per market, never blended. Competitor comparisons are split by market too. Markets we do not cover are explicitly labeled "not covered" rather than filled in with English samples — that would systematically flatten real per-market differences.

Two layers of control: first, the data region is configurable (domestic / cross-border) and can be locked to domestic-only models; second, calling overseas models requires your explicit authorization, and every call is logged (who authorized it, when, and what was called). Tenant data is strictly isolated — cross-tenant access is refused at the architecture layer. The full statement lives on our public Trust & Compliance page, and our privacy policy and DPA are available for your legal and security teams to review.

Reports support full brand customisation: your brand name, logo and watermark, exportable as PDF and PPTX. Chinese and English are both natively supported, and public share links switch language too. Every metric in the report carries its denominator and methodology notes, so it can go straight into HQ reporting — no reformatting, and no awkward moment when someone asks "where does this number come from".

Yes. The open platform provides three routes: a REST API (v2/v3 with tenant-scoped tokens and permission scopes), Webhook events (article published, lead created, sampling completed, report ready and more) and an MCP server that plugs into MCP-capable AI workflows. Most competitors only let you read data in their dashboard — for companies with a data platform, being able to take the data out is often a deal-breaker, so ours is open by default.

Role-based access control: each role sees only its scope — the European team sees European projects, the China team sees China. Key actions (logins, configuration changes, approvals) are written to an audit log. Permissions follow roles rather than individuals, so staff turnover is handled by changing a role, not by wondering who still has access. SSO and enterprise system integration (OA/CRM/DAM and more) are part of the strategic customisation service — contact us to assess.

Each product line gets its own project: separate knowledge base, monitoring keywords and brand profile. Content generation only injects that project's knowledge base and sampling is tracked per product line — isolation is mechanical, not a matter of human care.

Yes. The parent brand and each product can be separate projects, or monitoring keywords can be layered within one project — volume is tracked separately with a combined overview.

Local recommendations are one of the winnable battles in GEO: you will not outspend giants on national terms, but "city + service" questions are far less contested. Three layers: (1) structured facts — make your services, hours and contact details machine-readable. The platform does this by binding a store profile to a content task: every generated article carries that store's structured data (name / address / phone / hours) and outputs it on the website page; (2) local question monitoring — track real local phrasings to see who AI currently recommends and why; store-level monitoring composes queries from area + category + reputation words (e.g. "dental cleaning in Xuhui, Shanghai") and keeps them in a fixed question set for weekly retests; (3) sustained content supply — local explainers, cases and service notes that build up citable material. One precondition: consistency — AI cross-checks information across platforms, so your website, map listing and review profiles must align. The website side is handled automatically through structured data; maps and review sites need your own upkeep. Contradictory information is worse than none.

Yes, and more strictly than most industries: outcome promises, cure rates and absolute claims like "best / number one" are prohibited (pet healthcare has additional local rules in many regions). The platform ships a healthcare compliance pack that intercepts these phrasings before publishing and suggests compliant rewrites. Another way to look at it: compliance is not the enemy of traffic — medical advertising penalties run from fines to license suspension, and every sentence we block is one less risk you carry.

Yes — and the game is local questions, not national keywords. The platform closes the loop at store level: (1) keep address / phone / hours in the store profile; bind it to a content task and every generated article carries the store's structured data (LocalBusiness); (2) start monitoring from the store page — queries are composed from area + category + reputation words and synced into the fixed question set for weekly retests, so you can track visibility over time; (3) phone-click tracking and lead attribution show the full path from "AI mentioned us" to a call and a lead. The other quick win is usually information consistency: AI leans heavily on cross-checking multi-platform information for local recommendations, so aligning your website / map listing / review profiles helps — the website side is automated, maps and review sites need your own upkeep — and it often moves the needle faster than ten new articles.

They are not alternatives but links in the same chain: review platforms serve customers who browse, while AI serves customers who simply ask — and the answers AI gives quote exactly that kind of public content, including your review profiles. The real path is: AI generates a recommendation from public sources (including your reviews) → the customer follows it. So your existing review work is not wasted; what the platform adds is the missing segment: making your professional information readable and understandable to AI, and continuously monitoring how it describes you.

First, the mechanism: AI answers are grounded in public sources, so negative descriptions surface in proportion to negative material in those sources. The platform gives you negative alerts: when negative phrasing appears in mentions, you are notified with an evidence record (which source, at what time). The response has two directions — dilute the source pool with positive facts (credentials, cases, service commitments, dispute-handling records), and use platform appeals or corrections for inaccurate sources. We do not offer "post removal" style operations: they are unreliable, non-compliant and create new risks of their own.

Quite the opposite: AI answers are not yet monopolised by big brands — in long-tail questions, whoever states the facts clearly and gets indexed wins the slot. The real barrier for small companies was never budget; it is content supply, which is exactly what the platform handles for you: the industry question bank tells you what to answer, the content pipeline drafts from your materials, and compliance gates do the policing. No dedicated team needed — one person checking results and approving content weekly is enough.

Month-to-month billing is available on standard plans — no annual lock-in, no hidden fees. Features, quotas and prices for every tier are public on the pricing page. The more sensible path is to verify with facts first: a free signup gives full-feature trial access — run your own brand end to end (monitoring, content, reports all genuinely produced) and decide afterwards. If you do not even want to sign up, there is a registration-free AI check.

Straight talk first: anyone promising "top rankings within a week" or "guaranteed recommendations" is not to be trusted — AI answers are generated dynamically and no one can promise on the engines' behalf. What can be fast is the process: the check-up reports immediately, content production is measured in days, distribution completes the same day, and re-testing runs weekly. When AI starts citing you depends on crawl and indexing cycles; our monthly reports show the real trend. For a client meeting next month, we suggest a check-up plus content catch-up — present "where you stand and what is being done" honestly; it holds up better than a prettied-up number.

The risk structures differ. GEO buys no backlinks and stuffs no keywords — the core is stating true information clearly and distributing it through compliant channels, which has no "algorithm penalty" surface. On the contrary, the platform's layered compliance gates (advertising-law red lines, sensitive terms, industry rules) intercept violations before publishing, reducing your own legal exposure. Distribution goes through official APIs or your own computer (the desktop executor) rather than bulk operations from platform IPs, so platform anti-fraud systems are not triggered.

Daily use requires no AI knowledge: monitoring ships with sensible defaults, the question bank offers candidates you simply confirm, content generation is a one-click task, and progress plus approvals work on your phone. What genuinely needs your input is two things — confirming what to monitor, and approving what gets published. If you want hand-holding, Enterprise tiers and above include dedicated implementation and customer success: from account setup to the first monthly report, someone walks the first full cycle with you.

Yes — full brand customisation: brand name, logo, watermark and footer all become yours, and every page your client sees is your deliverable. Commercial data licensing (Research edition / agency partnership scope) also lets monitoring conclusions go into your published materials. Client monthly reports can be generated in bulk with public share links — your clients never need to log in anywhere; they just open the link.

The multi-client workspace is built on separate tenants: each client gets its own tenant, and data is isolated by tenant and by project — switching to a client shows only that client's data, never a blend; clients cannot see each other. Enterprise tiers and agency partnerships open up multi-client management, with client counts agreed per engagement. In other words: isolation is not "be careful" — it is enforced by the architecture.

Three verifiable artifacts: (1) effect-proof report — core metrics current period vs previous, with an improvement list; (2) article-level citation attribution — "was the piece we published cited by AI?", checkable article by article with sampling evidence for spot checks; (3) case-study pack — before-and-after re-test comparison as a one-page PDF, handy when your client forwards it to their boss. One more tip: precisely because the evidence is verifiable, we suggest you do not promise rankings either — showing data and process persuades better than promises, and that is how the platform treats its own customers.

No — that is a baseline of the agency partnership: clients are introduced by you and bound to your account; the platform provides tools and back-end capability without reaching around you. Rebates settle monthly by service month, percentages are transparent, and the ledger is always checkable. If your delivery requires clients to be unaware the platform exists, brand-customized reports and custom domains make that possible.

That is exactly what the platform is for: industry assets are included in every plan at no extra cost — 22 vertical industries, each with its own question bank, compliance rules and industry profile (including food & dining, pet services, dental care and local services). Pick the industry when setting up each client project and content generation applies the corresponding rules automatically; low-quality questions are filtered by the quality gate before entering the bank, so your sampling budget is not burnt on noise. Your job shifts from writing content to reviewing it and serving clients — the latter is worth far more.

Enterprise contracts are annual, with monthly or annual payment cadence; self-serve plans can be billed month-to-month. Invoices are issued after payment (VAT special or general). Final terms follow the signed contract; if your legal team needs to review clauses, we provide a draft for internal review.

You get console access: configure monitored questions, view sampling data, launch content production and distribution. Your side mainly confirms two things — (1) the monitored question list (we propose candidates, you pick the final set); (2) content review before publishing (if client approval is enabled). Everything else is run by the platform and our operations.

We do not promise specific results: AI answers are generated dynamically and shift with engine updates and competition — any vendor guaranteeing outcomes is not credible. What we do commit to is verifiable process: sampling data you can re-check, before/after re-tests, and every conclusion traceable to raw samples. Empirically, 4–8 weeks of continuous monitoring makes trends more meaningful (single runs fluctuate a lot — that is a property of the measurement itself, not a product defect).

No. Your data is used only for analysis, reporting and content generation inside your own account. The platform can be configured for domestic models only; calling overseas models requires your explicit authorization and is logged. Tenants are strictly isolated — agencies cannot see other agencies' client data. Knowledge base documents are embedded locally for retrieval and writing references within your account only. See the Privacy Policy and DPA for the full statement.

You do. The platform proposes candidates based on your industry and brand (including real search-suggestion collection and AI-expanded candidates), but the final monitored list is confirmed by you. You can add, edit or remove questions anytime from the question library page in the console — changes take effect immediately.

Yes. The export center exports sampling details, source details and report data as CSV/Excel, including the full methodology dictionary — ready to hand to your own analysts or consultancy. Exports are not gated on subscription status.

Yes. Overseas engines (OpenAI / Perplexity / Gemini / Claude, etc.) are reached through an overseas relay so you can monitor brand presence for English queries; content distribution covers Dev.to, WordPress.com, Medium and more. Note: overseas engine sampling requires the corresponding API keys, and the coverage scope is confirmed at onboarding.

Three verifiable differences: (1) we do not promise outcomes, and every report states what cannot be concluded from it (most tools give conclusions without boundaries); (2) every metric carries its denominator and confidence interval, so you can tell a 400-sample result from an 8-sample fluke; (3) when the data shows "not mentioned" or "mentioned but not cited", the platform not only alerts — it provides the content production and website remediation to act on it, diagnosis to execution in one chain. Our full measurement methodology is public on the Methodology page — compare it yourself.