Help & FAQ
Straight answers about SEO vs GEO, where the data comes from, accuracy and honesty, billing, and privacy. Every answer documents a real feature.
Ask a product question
Instant answers come from the verified FAQ — no AI model call and no credits. Account-specific or long-tail questions can be escalated to the signed-in assistant.
Getting started
8What the platform is, what you can do without connecting anything, and the core workflows.
NimbleSEO is one platform for both classic SEO and GEO (Generative Engine Optimization). You write and optimize content, audit pages, track rankings, and monitor whether AI engines like ChatGPT, Perplexity, and Google AI Overviews cite your brand — all on a single engine that powers both the search lane and the AI lane.
Last updated · 2026-07-10
You can write, score an article, audit a URL, or improve a draft without connecting anything. For dashboards and recommendations based on your business, connect two performance sources: Google Search Console, Google Analytics 4, Ahrefs, or Semrush. Examples stay clearly labelled until then. Add publishing tools such as a CMS only when you need them.
Last updated · 2026-07-23
Open Integrations and connect two performance sources: Google Search Console, Google Analytics 4, Ahrefs, or Semrush. After the first source, setup guides you to add a second so recommendations can cross-check results. Add a crawler, CMS, alerts, or automation only when a workflow asks for it. Every connection stays within the active workspace.
Last updated · 2026-06-30
In the Technical surface, the platform finds on-page issues, lets you preview the exact change, applies it, and — crucially — lets you undo it. Every fix is reversible: you can revert any applied change, so you can experiment without risk. This reversible find → fix loop is the core of the on-page engine.
Last updated · 2026-06-30
Existing customers can use the Feedback button in the app shell or sign in and open Support to file and track an account-linked ticket. If you cannot sign in or are still evaluating NimbleSEO, email support directly.
Last updated · 2026-07-10
AI Visibility samples a set of prompts across multiple AI engines and extracts whether — and how — your brand is cited. Each result shows the engine, the prompt, and the citation evidence, with a confidence band. It is a sample across prompts and engines, not a complete census of every AI answer, and it is labelled that way.
Last updated · 2026-06-30
Paste or open a draft in the Humanizer and choose a mode (Standard, Creative, Academic, or Casual) with adjustable intensity to match your brand voice. It rewrites the text to read more naturally. It does not promise to defeat any specific AI detector — it improves the writing, and you review the result before publishing.
Last updated · 2026-06-30
Add the keywords you want to track in the Rankings surface and the platform records their positions over time so you can see rank history and movement. Position data comes from the configured SERP source; until a run completes, new keywords show as pending rather than a fabricated position.
Last updated · 2026-06-30
Product & workflows
8How NimbleSEO finds what matters, recommends the next action, tracks the work, and measures the impact.
NimbleSEO starts with your goal, records source evidence, identifies the most valuable opportunity, recommends the next action, and tracks the work through completion. It then repeats the same measurement, compares before and after, and reports the impact with an attribution-confidence label. What the product finds appears as “What we found,” the prioritized recommendation appears as “What to do next,” active work appears as “Work in progress,” and measured change appears as “Results.”
Last updated · 2026-07-17
The command deck ranks concrete moves using the evidence available in your workspace: AI-answer presence and share of voice, ranking signals, technical findings, and content readiness. Inputs can be measured, sampled, or estimated, so every recommendation keeps its provenance and uncertainty instead of hiding them inside one magic score. You remain the decision-maker.
Last updated · 2026-07-17
Yes. The content workflow covers research and briefs, drafting, brand-voice and originality checks, SEO/GEO optimization, internal-link opportunities, decay detection, and refresh planning. Generated work is a draft for human review; factual claims and citations still need editorial verification before publication.
Last updated · 2026-07-17
Supported CMS connections can stage or publish approved content and on-page changes, but the safe path is preview → approve → deploy → read back → recheck. A database-only suggestion is not presented as a live CMS change, and an integration failure is shown as partial or failed rather than silently marked complete. CMS availability depends on the connected provider and plan.
Last updated · 2026-07-17
Yes. Workspaces keep client data separate, portfolios roll up multiple workspaces, roles control access, and agency workflows include client-ready evidence reports and white-label delivery. Limits for seats, workspaces, prompt checks, crawl pages, and credits vary by plan, so the pricing page is the source of truth for capacity.
Last updated · 2026-07-17
The product includes live crawling and technical findings, reversible on-page fixes, internal-link and topic-architecture analysis, provider-backed backlink analysis, and local-search workflows. Coverage is not identical in every area: backlink and local metrics depend on configured external providers, and a feature visible in the wider product may be plan-gated or outside the default launch workflow. The UI states when a connection or entitlement is required.
Last updated · 2026-07-17
You can retain before-and-after evidence, re-run the relevant checks, and compare changed pages with a matched control set. That supports a directional or matched-control conclusion, not proof of causation. NimbleSEO never upgrades an observational comparison into a randomized-experiment claim.
Last updated · 2026-07-17
The interface ships in 18 languages, while research and content generation support 36 language and market variants. Interface language and content language are separate: for example, you can use the product in Dutch while researching and writing for Japan. Search locale, script direction, and generation instructions follow the selected content market.
Last updated · 2026-07-17
SEO vs GEO
13Classic search optimization, Generative Engine Optimization, and how the two share one engine here.
GEO is the practice of optimizing content so AI engines — ChatGPT, Perplexity, Google AI Overviews, Gemini — cite it in their generated answers. Where classic SEO competes for a blue link on a results page, GEO competes to be the source an AI quotes. Techniques include clear BLUF answers, data density, structured formats, and source attribution.
Last updated · 2026-06-30
Classic SEO optimizes for ranking in a list of links; GEO optimizes for being cited inside an AI-generated answer. SEO success is a position on a SERP; GEO success is a citation in ChatGPT or an AI Overview. They overlap — good structure and authority help both — but they are measured differently, which is why this platform gives every surface an SEO lane and a GEO lane.
Last updated · 2026-06-30
Standard SEO is still foundational, but it cannot tell you whether an AI answer mentions your brand, cites your page, or sends the source opportunity to a competitor. GEO adds that evidence layer across tracked prompts and engines; NimbleSEO keeps both lanes together so one technical or content improvement can be evaluated in search and in AI answers without maintaining two toolchains.
Last updated · 2026-07-17
There is no universal refresh schedule: source selection can change with the provider, model, interface, region, account state, and time. NimbleSEO does not claim to control that cadence. It reruns prompts on the schedule included in your plan, timestamps every capture, records the capture method, and labels unstable results so you can distinguish a durable change from normal answer variance.
Last updated · 2026-07-17
Yes. Add the rivals you care about and NimbleSEO compares mentions, citations, answer position, engine coverage, and the third-party sources that influence the same prompts. It uses public or connected evidence rather than private competitor data, and every sample result remains explicitly labelled until your own captures are connected.
Last updated · 2026-07-17
Yes. Prompt gaps, cited-source evidence, page findings, and competitor patterns feed a ranked list of next steps and content briefs with the reason and evidence attached. You review every recommendation before publishing, applied technical fixes are reversible, and a follow-up run can check whether the observed visibility changed without claiming that the action caused it.
Last updated · 2026-07-17
We track brand mentions and citations across ChatGPT, Claude, Gemini, Perplexity, Grok, Microsoft Copilot, DeepSeek, Meta AI, You.com, and Phind. Google AI Overviews and AI Mode presence is tracked separately in the search-results lane, not as an answer-engine adapter. Each engine’s capture method (bare API model, search-grounded API, or manually seeded) is disclosed rather than hidden behind one platform count, and coverage is a sample of prompts per engine, not a census of every possible AI answer — the methodology page documents exactly how the sampling works.
Last updated · 2026-07-10
A citation is an instance where an AI engine references your brand or domain in its generated answer to a sampled prompt. We extract the engine, the prompt, and the matched evidence so each counted citation is traceable to a real result. We do not infer citations you cannot inspect — if we cannot show the evidence, we do not count it.
Last updated · 2026-06-30
No. A bare provider API model, a search-grounded API, and a real consumer browser session can return different answers to the same prompt. NimbleSEO keeps those populations separate and publishes the capture mode for each engine rather than hiding the difference behind one platform count. Google AI Overviews and AI Mode are measured in the search-results lane.
Last updated · 2026-07-17
Start with the decisions and questions that matter to your customers, then expand each topic into realistic phrasings, intents, and follow-ups. That set is your prompt universe. Keep branded prompts separate from non-branded discovery prompts, preserve the set over time for comparability, and add new prompts as a new cohort rather than silently rewriting history.
Last updated · 2026-07-17
AI answers vary between runs, so one response is evidence of one response — not a stable rate. Repeats let the platform estimate stability, effective sample size, and a Wilson confidence interval. Correlated repeats are down-weighted, and a result stays inconclusive when there is not enough defensible evidence.
Last updated · 2026-07-17
It can connect retained citation evidence with first-party analytics and reporting signals when those sources are available, but attribution is not automatically proof of cause and effect. A cited answer, a later visit, and a conversion can support a traceable evidence chain; they do not prove the citation caused the conversion. Missing analytics data is shown as unavailable rather than estimated into a customer result.
Last updated · 2026-07-17
Data & integrations
5Where the data comes from, how sources and provider connections are handled, and what each data label means.
Search and keyword data come from third-party SERP/keyword providers (such as DataForSEO); your own performance data comes from Google Search Console and GA4 once you connect them; technical findings come from a live crawl of your pages; and AI-visibility data comes from sampled prompts run against AI engines. Each surface tells you which of these it is using.
Last updated · 2026-06-30
These four labels tell you exactly how real a number is. "Sample" is illustrative data shown before you connect a source. "Works keyless" runs live on your own input without any connection. "Connect to run" means the feature needs a connected source (e.g. GSC) before it returns results. "Measured" is your real connected data. We never show sample data without the Sample label.
Last updated · 2026-06-30
No. NimbleSEO does not claim an Ahrefs- or Semrush-scale proprietary web index. Keyword, SERP, backlink, and local-search coverage uses configured third-party providers; your first-party performance comes from connected sources such as GSC and GA4; and technical findings come from the product crawler. Provider-backed metrics inherit the provider’s coverage and refresh limits.
Last updated · 2026-07-17
Freshness depends on the source and the run you schedule. Live crawls and explicit checks record their run time; connected analytics follow the source’s reporting delay; provider metrics follow that provider’s refresh cadence; and AI-answer observations are tied to a prompt, engine, capture mode, locale, and timestamp. The interface should show the observation window rather than implying everything is real time.
Last updated · 2026-07-17
Reports and supported result tables can be exported for analysis and client delivery. API, webhook, and warehouse automation are available only on eligible plans or configured workspaces; their presence in the codebase does not make them part of every plan. Check the current plan contract or contact sales for the exact endpoint, retention, and volume requirements you need.
Last updated · 2026-07-17
Accuracy & honesty
8What is measured vs sampled vs modelled — and the rule that we never invent a number.
No. NimbleSEO does not cloak pages, fabricate citations, create fake reviews, or recommend spam. It helps teams improve inspectable content, technical accessibility, source quality, and entity clarity; recommendations remain human-reviewed and outcomes are never guaranteed.
Last updated · 2026-07-17
As accurate as their source, and we label that source so you can judge it. Measured numbers come from your connected GSC/GA4 or a live crawl. Sampled numbers (like AI visibility) come from a defined prompt sample with a confidence band. Modelled numbers (like a bill projection or a forecast) are estimates from a documented model. We do not present any of these as more certain than they are.
Last updated · 2026-06-30
Measured is real data read from a source you connected. Sampled is computed from a representative subset (for example, AI visibility samples prompts across engines) and carries a confidence band. Modelled is an estimate produced by a documented formula (for example, the bill projection from your usage). Each surface states which kind it is showing rather than blending them into one figure.
Last updated · 2026-06-30
No. If a ranking has not been fetched yet, the keyword shows as pending — not a placeholder position. If an AI citation cannot be backed by inspectable evidence, it is not counted. Our first rule is that sample, heuristic, or computed output is never presented as measured or live; where we cannot show the real thing, we say so.
Last updated · 2026-06-30
Because that surface needs your data to produce a real result and you have not connected it yet. "Sample" lets you preview the layout with illustrative figures; "Connect to run" means the feature stays inert until you wire up the source it depends on. Connecting GSC, GA4, or a crawl replaces the sample with your measured data.
Last updated · 2026-06-30
As matched-control lift, never as proof of cause. When we report a change after an action, we compare against a matched control set and label the result that way. A matched-control comparison is evidence of association, not proof that the action caused the change — so we report the association and stop there.
Last updated · 2026-06-30
Not by default. NimbleSEO can prioritize actions, draft changes, and dispatch supported workflows, but consequential changes should pass through the visible preview and approval path. The AI companion is advisory, and a generated recommendation is not treated as an applied, published, or verified result.
Last updated · 2026-07-17
AI visibility is a sampled observation, not a census of every personalized answer; third-party search and link metrics depend on provider coverage; browser capture can be policy- and spend-gated; and not every additional tool in the wider application is part of the default launch workflow. NimbleSEO does not own a web-scale index, does not prove causation from an observational lift, and does not promise rankings or citations.
Last updated · 2026-07-17
AI assistant
6What the in-product AI can answer, what it costs, which sources it uses, and where it must stop.
The signed-in companion can explain product concepts, the current surface, measurement labels, supported workflows, billing rules, and practical next steps. Its product answers are grounded in the same FAQ, page-help, glossary, facts, methodology, and plan sources used by the interface. Workspace-specific analysis still depends on the context and evidence the current surface provides.
Last updated · 2026-07-17
FAQ and help searches are deterministic retrieval and cost 0 credits. Local navigation and supported visible page actions in the companion also cost 0 companion credits. A free-form model answer costs 1 managed credit, or 0 managed credits when an eligible workspace uses its connected Anthropic key. The interface discloses the charge before the answer path is used.
Last updated · 2026-07-17
The assistant retrieves a small set of approved product sources before it calls a model, treats page and customer content as untrusted context, and is instructed to say what is missing when the sources are insufficient. It must not invent routes, buttons, plan rules, actions, or evidence. Important billing, legal, security, and measurement answers link back to their canonical source so you can verify them.
Last updated · 2026-07-17
It can cover questions documented by the approved product knowledge base, but no responsible assistant can promise every possible answer. When the evidence is missing, account-specific, legal, or security-sensitive, it should say so and route the user to the correct support, privacy, security, or sales channel. An honest escalation is a successful answer; a plausible invention is not.
Last updated · 2026-07-17
A free-form AI answer is advisory and cannot silently mutate or publish work. The companion can navigate or dispatch only supported visible actions, and higher-risk changes still use the product’s explicit preview, approval, and evidence flow. It must not claim an action completed unless the product returns that result.
Last updated · 2026-07-17
Yes, a model-backed answer can follow the current interface language when that language is supported, while product names, routes, and technical terms may remain unchanged. Retrieval still comes from the approved canonical sources, so changing the answer language must not change the underlying plan, security, or methodology facts.
Last updated · 2026-07-17
Billing & credits
8Plans, credit costs, hard limits, top-up packs, annual vs monthly, upgrades, and cancellation.
Managed plans run Starter at €29 or $29/month (300 credits), Growth at €79 or $79/month (1,000 credits), Pro at €139 or $139/month (3,000 credits), Agency at €299 or $299/month (5,000 credits), and Agency+ at €499 or $499/month (10,000 credits). EUR applies to euro-area billing and USD elsewhere. Enterprise is custom. There is no free tier after the evaluation; eligible new customers can use the 7-day, 50-credit Starter trial once before choosing a paid plan.
Last updated · 2026-07-02
Credits are spent per AI action. Generating an article costs 10 credits, a content brief 10, a comparison or listicle 5, a landing page or email sequence 3, an image 2, humanizing 3–6 depending on the mode selected before the run, a manual brand-monitor run 15, a delivered scheduled brand-monitor run 10, an FAQ-schema or press-release generation 2, and a rewrite 1. Scoring, auditing, and most other features are free. These are the same costs shown in the in-app Help reference.
Last updated · 2026-07-14
Managed-credit actions hard-block when your available balance reaches zero; we do not create surprise overage invoices. You can buy a visible one-time pack (100 credits for €5 or $5, 500 for €20 or $20, or 2,000 for €60 or $60), upgrade the plan, or use your own provider key where supported. EUR applies to euro-area billing and USD elsewhere. Prompt-check and crawl-page allowances also hard-block at the published plan limit.
Last updated · 2026-07-13
Yes. Annual billing is discounted on every paid plan — for example Growth is the equivalent of €63 or $63/month billed annually instead of €79 or $79/month, saving €192 or $192 a year. EUR applies to euro-area billing and USD elsewhere. Annual subscriptions receive all 12 published monthly credit allowances up front after payment; monthly subscriptions receive one allowance each successful billing cycle. You can switch intervals from billing.
Last updated · 2026-07-13
Yes. Cancel anytime from your billing settings — no contracts, no cancellation fees. Your access continues until the end of your current billing period.
Last updated · 2026-06-30
Yes. The Starter trial lasts 7 days and includes 50 non-renewing managed credits — enough to run a multi-engine visibility scan and experience the core content workflow. A payment card is required to prevent repeat trials, but it is not charged automatically and the trial does not renew. One trial is available per verified email address and payment card; an email address or card that has already claimed one is sent to the paid plans.
Last updated · 2026-07-18
Change plans anytime from your billing settings. Stripe-confirmed upgrades update plan features immediately; the next applicable successful paid-invoice allocation supplies the new plan credits. Purchased credit packs remain in the ledger. Downgrades apply from your next billing period. Billing shows a recommendation based on actual credit, prompt, and crawl consumption.
Last updated · 2026-06-30
Yes. Growth plans and above include team workspaces where you can invite writers, editors, and managers. Assign roles, leave comments, and manage content workflows together.
Last updated · 2026-06-30
Security, privacy & compliance
11Data handling, access controls, sub-processors, GDPR rights, security evidence, and compliance status.
Your content and account data are stored in our managed database, and processing can involve service providers for database hosting, AI model calls, search-data access, and payments. The Privacy Policy describes the data categories and lawful bases; the Trust centre shows the public sub-processor registry status. Contact privacy@nimbleseo.ai for current processing and transfer details.
Last updated · 2026-09-02
No active, counsel-approved DPA, SLA, or BAA is currently published for self-service. The DPA and SLA pages show their preparation status. Contact privacy@nimbleseo.ai or legal@nimbleseo.ai for current data-processing and procurement terms. When a signed document is issued, its verification page can confirm the accepted version by document ID.
Last updated · 2026-09-02
NimbleSEO uses service providers for database hosting, AI model calls, search-data access, and payments. The Trust centre shows the public sub-processor registry status and says when no records are listed; an empty public registry must not be read as meaning no providers are used. Contact privacy@nimbleseo.ai for the current processing details needed for a vendor review.
Last updated · 2026-09-02
You have the right to access, export, correct, and delete your personal data. From Settings → Privacy & data you can request a full export (JSON/ZIP) or the deletion of your account and data; every request is tracked against the GDPR 30-day clock. The Privacy Policy describes the process and the contact path if you prefer email.
Last updated · 2026-07-10
We do not train our own foundation models on your content. Your content is processed by third-party AI providers to generate the output you ask for, under their data terms. The Privacy Policy and Trust centre document which providers are involved and the applicable data-handling terms.
Last updated · 2026-06-30
When a legal acceptance record is issued, its unique verification URL or document ID can be used to confirm the recorded version and available integrity evidence. The Trust centre shows only documents currently approved for public display; an unpublished agreement or a private customer record will not be advertised there as a downloadable document.
Last updated · 2026-06-30
Not currently. Internal readiness work maps controls to the AICPA Trust Service Criteria, but NimbleSEO is not SOC 2 attested and no Type II report is available. The Trust centre will list a compliance artifact only after it exists, has been reviewed, and is approved for release.
Last updated · 2026-07-17
Application traffic is configured for HTTPS, while database and storage encryption at rest are managed by the hosting providers. Exact cipher, protocol, key-management, backup-retention, recovery-objective, and point-in-time-recovery evidence is supplied during review when available; it is not turned into a public guarantee unless an approved SLA or signed order says so.
Last updated · 2026-07-17
Workspace data is separated with tenant-scoped application checks and row-level database policies, while roles control what members can access. SAML/OIDC, MFA, and SCIM are available only for configured workspaces and contractual plans; they should not be assumed from a lower-tier account. The Trust centre shows the currently published evidence and provides security and compliance contact paths for anything not listed.
Last updated · 2026-07-17
The public product does not make a blanket data-residency or backup-retention promise. Enterprise agreements can define custom retention and data-control requirements, while the deployed project region and any contractual isolation requirement must be confirmed during procurement. Deletion and return of customer data remain subject to the published Privacy Policy, any agreement actually signed with the customer, and legal-retention obligations.
Last updated · 2026-07-17
No. SEO and GEO outcomes depend on factors outside any tool’s control, so rankings and citations are never guaranteed — our numbers are measurements, samples, and models, not promises. Always review AI-generated drafts before publishing. The Disclaimer sets this out in full.
Last updated · 2026-06-30
Still stuck? Our support team can help.