` markup in ways that conflict with schema.org embedding. Switching to a theme-friendly Yoast schema output is usually the fastest fix — contact support if you need specific guidance. ## Disconnect Open **Settings → Integrations → WordPress** and click **Disconnect**. Videntic stops pushing to the site and forgets the stored application password. For extra safety, also revoke the application password from your WordPress profile page. ## Related The full gap-to-published flow. The feature this integration powers. The alternative CMS path. The Content Creator specialist writes the drafts. # AI Visibility Score Source: https://docs.videntic.com/metrics/ai-visibility-score How often your brand is mentioned across the LLM providers you track. **AI Visibility Score** is the headline metric on your dashboard. It answers the core GEO question in one number: *when customers ask an LLM about your category, how often does your brand come up?* ## Overview For every analysis run, Videntic simulates customer queries against the major LLMs and checks each response for a text mention of your brand or domain. It calculates a mention rate for each provider, then averages providers equally. Higher is better. A score of **35%** means one in three AI answers about your topics mentions you — a meaningful position in most categories. ## How AI Visibility is calculated ``` Provider visibility = responses mentioning your brand / responses from that provider × 100 AI Visibility Score = equal-weight mean of provider visibility values ``` A response counts when the answer text contains your brand name, domain, or a known alias recognized by Videntic's mention detection. URL citations are measured separately and do not add weight to this score. The score is computed per run and per model, then aggregated: * **Overall score** — averaged across all models in your plan, weighted evenly. * **Per-model score** — the same metric broken out by ChatGPT, Perplexity, Google AI Overviews, and Gemini (where included in your plan). Providers that produced no responses are excluded from the overall mean rather than counted as 0. Per-model scores often tell you more than the average. It's common to be strong on one model and weak on another, because each one sources information differently. ## How to read the number Scores fall into four human-readable tiers: | Tier | Range | What it means | | ----------------------- | ------ | --------------------------------------------------------------------------------------------------------- | | **Building visibility** | 0–5% | You're invisible in most AI answers. Focus on content gaps and baseline GEO‑16 fixes. | | **Gaining traction** | 5–15% | You show up in a meaningful minority of answers. Work on quality (Citation Position) and coverage. | | **Well positioned** | 15–35% | You're a recurring answer in your category. Consolidate wins and defend against competitors. | | **Category leader** | 35%+ | LLMs consistently mention you. At this point, the lever is depth — richer content, more specific queries. | Benchmarks vary by category. In high-competition consumer categories, 35% is excellent. In niche B2B domains, 60%+ is achievable. ## Why AI Visibility matters ### Business lens: demand capture A rising share of research and purchase decisions is now happening inside AI assistants. Every prompt where you're *not* cited is a prospect who saw a competitor's answer instead. ### Marketing lens: channel measurement AI visibility is to GEO what keyword rankings were to SEO. If you can't measure it, you can't report on it, budget for it, or defend its value internally. ### Product lens: positioning Which queries you win tells you how the market *frames* your product. If you win "best CRM for startups" but lose "CRM with advanced workflows", the LLMs — trained on the open web — are telling you something real about your brand's current associations. ## How to improve it In rough order of leverage: 1. **Fix GEO‑16 pillars** flagged red on your audit, especially Structured Data, Evidence & Citations, and Metadata & Freshness. These are the baseline requirements for being quoted. 2. **Publish content targeting your top content gaps**. Videntic ranks gaps by impact; start from the top. 3. **Work on Citation Position**. Being cited 5th is worth less than being cited 1st. See [Citation Position](/metrics/citation-position) for why. 4. **Expand to under-served LLMs**. Your per-model breakdown will show you which model you're weakest on; often one content fix lifts all four. ## Related metrics Same data, but normalized against your tracked competitors. How prominent your mention is when you are cited. The on-site signals that drive citation likelihood. Prompts where you're missing. The biggest lever for AI Visibility. # Citation Position Source: https://docs.videntic.com/metrics/citation-position How early and prominently your brand appears in LLM responses that cite you. **Citation Position** measures the quality of the citations you do earn. Being mentioned first in an LLM's answer is very different from being mentioned fifth — users tend to skim, and the first brand frames the entire response. ## Overview For every response that cites you, Videntic records where in the answer you appeared — first mention, second, third, and so on — and averages those positions over time. Lower is better: an average position of **1.0** means you're consistently the first brand named. Citation Position only includes responses where you were cited. If your AI Visibility is 20%, Citation Position is computed over that 20% — not over every run. ## How Citation Position is calculated ``` Citation Position = average(position of first mention) across all responses where you were cited ``` For each qualifying response, Videntic finds where your brand appears in the cited sources or the narrative, records its ordinal position (1 = first, 2 = second, etc.), and averages the values per run. Like the other metrics, it's computed per model and overall. A strong overall score with a weak per-model score on Perplexity often means Perplexity is surfacing a competitor ahead of you — a targeted problem to solve. ## How to read the number Roughly: * **1.0–1.5** — You're consistently the first or second citation. Category leader behaviour. * **1.5–2.5** — You're in the top tier but often sharing the answer. Normal for competitive categories. * **2.5+** — You're regularly cited but buried. Work on authority and structured data. Users read answers top-down and rarely click every source. A meaningful share of real-world intent goes to whichever brand is named first. ## Why Citation Position matters ### Conversion lens: attention hierarchy The first brand mentioned gets the most narrative weight, the most perceived authority, and — where the LLM surfaces clickable sources — the most clicks. Position 1 and position 5 are not equivalent citations. ### Competitive lens: share within share You can have healthy AI Visibility and healthy Share of Voice and still lose the category narratively because a competitor is always cited *first*. Citation Position surfaces that. ### Content lens: canonical authority When your domain is cited *first*, it tells you the LLM treats you as the canonical source for that query. That's the GEO equivalent of owning a keyword — and it's defensible. ## How to improve it Citation Position tracks authority signals more than coverage. Improving it is slower than fixing AI Visibility but pays off longer. 1. **Invest in GEO‑16 Authority & Trust**. About, Team, Policies, citations of your own work — these are the pillars LLMs weight when deciding who's the primary source. 2. **Ship structured data**. Well-marked articles, products, and FAQs make you easier to quote cleanly, which models prefer. 3. **Earn real references**. Being cited by high-authority external pages (docs, research, reputable press) still matters — LLMs learned from the web, and that hierarchy carried over. 4. **Be the definitive answer, not a footnote**. Publish full explanations on canonical pages, not scattered across short posts. Depth wins position. Position gains lag content changes by one or two analysis cycles, because LLMs need to re-encounter updated pages before citations shift. ## Related metrics Whether you're cited. Position is the quality layer on top. Your slice of citations versus competitors. The on-site authority signals that drive Citation Position. See which competitors tend to be cited ahead of you, per model. # Core metrics Source: https://docs.videntic.com/metrics/overview The four numbers that summarize your AI visibility. Videntic reports four headline metrics. Together they tell you *whether* you're visible in AI search, *how well* you're positioned when you are, *who else is being cited*, and *whether your site is technically ready* to be cited more. ## The four metrics The share of tracked prompts where your brand is cited, across all LLMs. Your citations divided by total citations of tracked competitors. How early you appear when you *are* cited. First mention vs. fifth is a big deal. Technical readiness for citation, scored against 16 pillars. ## How they fit together Think of them as four lenses on the same question — *are LLMs citing you when they should?* * **AI Visibility Score** is your headline result: *do they cite you at all?* * **Share of Voice** normalizes that against your competitors: *how does that compare to the rest of the market?* * **Citation Position** measures quality: *when you're cited, are you the star or the footnote?* * **GEO‑16 Score** — from your [GEO audit](/features/geo-audit) — measures cause: *is your site set up to be quoted by models?* You can move AI Visibility and Share of Voice by publishing better content or fixing technical issues flagged by the GEO audit. Citation Position is slower to shift and usually tracks alongside authority and structured-data improvements. None of these metrics are "one shot". They're tracked over time, per LLM, per website. The trend is what matters — a single report is a snapshot, not a verdict. ## Reading a report Every report shows the four metrics at the top, with: * **Current value** — what the most recent run measured. * **Trend** — change since the previous run, coloured up or down. * **Per-model breakdown** — the same metric computed separately for ChatGPT, Perplexity, Google AI Overviews, and Gemini, so you can spot model-specific gaps. Below the headline, the report surfaces your **top content gaps** and your **GEO‑16 pillar breakdown** — the actionable part. See [Dashboard](/features/dashboard) for a tour. ## What to look at first If you're reading your first report, start here: 1. **AI Visibility Score** — establishes your baseline. 2. **Citation Position** — tells you whether the citations you *do* get are meaningful. 3. **[GEO audit](/features/geo-audit)** — points at what to fix on-site. 4. **Share of Voice** — contextualizes everything against competitors. Then dig into individual content gaps. That's where action happens. ## Related Where these metrics live in the app. Every term used in a metric, defined. # Share of Voice Source: https://docs.videntic.com/metrics/share-of-voice Your brand mentions as a share of the competitive mention pool. **Share of Voice** (SOV) is the competitive version of AI Visibility. Instead of asking *are you cited?*, it asks *of all the brands being cited in your space, how much of the airtime is yours?* ## Overview For every provider, Videntic records brand mentions in the LLM response — yours and every tracked competitor's. SOV is your slice of that provider's competitive mention pool. SOV is a **relative** metric. If every brand in your category disappears from AI search, yours included, your SOV can stay flat while AI Visibility collapses. Always read it alongside AI Visibility. ## How Share of Voice is calculated ``` Provider SOV = your mentions / (your mentions + tracked-competitor mentions) × 100 Overall SOV = equal-weight mean of non-empty provider SOV values ``` The denominator includes your brand plus every domain on your **Tracked Competitors** list — both the ones you add manually and the ones Videntic auto-detects from past runs. Uncategorized sources (UGC, news, reference sites) are excluded; SOV is a brand-vs-brand measure. SOV is computed per run and per provider. Providers with an empty competitive pool carry no signal and are excluded from the overall mean rather than counted as 0. Provider market weights do not affect canonical SOV. ## How to read the number Good benchmarks depend on how many competitors you track and how concentrated your category is. Rough guidance: | You track | Healthy SOV | | ---------------- | ----------- | | 1–2 competitors | 40%+ | | 3–5 competitors | 25%+ | | 6–10 competitors | 15%+ | More important than the absolute number is the **trend** — is your slice of the pie growing or shrinking, and at whose expense? The SOV chart on the dashboard stacks every tracked brand so you can see the handoff: if you gained 5 points and your nearest competitor lost 5, you know exactly where the gain came from. ## Why Share of Voice matters ### Business lens: competitive positioning AI Visibility can rise on a growing tide — the whole category becoming more visible in LLMs. SOV strips out the tide and shows whether you're winning the category or just benefiting from it. ### Strategy lens: content attack plan SOV points at *specific* competitors to take market share from. If "Competitor X" dominates prompts about "onboarding" and you're absent, you have a targeted content brief. ### Exec lens: share narrative Board and leadership audiences understand market share intuitively. SOV translates AI visibility into a story they already know how to interpret. ## How to improve it 1. **Audit the competitors outgunning you**. Click a dominant competitor in the SOV chart to see which prompts they're winning. That list *is* your content roadmap. 2. **Refine your competitor list**. Auto-detection may include UGC or media sites that shouldn't count as competitive. Promote real rivals, demote the rest. See [Manage competitors](/how-to/manage-competitors). 3. **Win citation-quality, not just citation-count**. A first-position mention is worth more narrative weight than a fifth-position one. Pair SOV work with Citation Position work. If your SOV is very high but AI Visibility is low, you're the biggest fish in a small pond — your category simply isn't being discussed much in AI search yet. That's a content-creation opportunity, not a competitive threat. ## Related metrics The absolute version of SOV. How the citations you earn are weighted. How to add, classify, and review your competitor set. Prompts your competitors are winning that you aren't. # Common issues Source: https://docs.videntic.com/troubleshooting/common-issues The handful of issues that account for most support tickets, and how to fix them. The top problems customers hit during setup and normal operation, each with the fix. If yours isn't here, check the [FAQ](/troubleshooting/faq) or email [support@videntic.com](mailto:support@videntic.com). ## Website verification ### "Verification failed" when I know the tag is there Almost always one of four causes: * **The page is cached by a CDN.** Purge the CDN cache (or wait for TTL) and retry. Cloudflare, Fastly, and Akamai all cache HTML aggressively by default. * **The tag is in `` instead of ``.** Move it into `` — many CMS block editors default to body injection. * **Your homepage redirects.** If `example.com` redirects to `www.example.com` (or the other way), add the canonical version as the website URL in Videntic. * **Your site is gating the page behind an interstitial** (cookie banner, geo wall, bot protection). Add Videntic's verification path or IP range to your bot protection's allowlist. ### "File upload" method returns 404 on check The verification file needs to be reachable at `https://yourdomain.com/` — at the root. Placing it in `/public/` locally works if your framework serves `public/` at the root (Next.js, Nuxt, most static generators). If your build puts it elsewhere, adjust the deploy path and re-verify. ### Verification worked but crawl didn't start Give it a minute. If the crawl hasn't started within ten minutes, try adding the site again — most of the time it's a one-off transient and a retry gets you moving. ## Running an analysis ### Analysis is stuck at "running" for more than 15 minutes Our pipeline times out per-step, so a run should never hang forever. If yours has: 1. Refresh the page — the progress indicator occasionally goes stale even though the run completed. 2. Check **Settings → Notifications** for any error messages attached to the run. 3. If nothing has resolved after 20 minutes, email support with the website name and approximate start time. ### "Prompt limit reached" error Your **prompt library** is full — you've hit the number of active prompts your plan allows (25 on Basic, 100 on Pro, 200 on Max). This is a cap on how many prompts you can keep active at once, not a monthly run allowance. Options: * **Deactivate a prompt** you no longer need — this frees a slot while keeping its history, and you can reactivate it later. * **Delete a prompt** to remove it permanently. * **Upgrade to a higher plan** for more slots — takes effect immediately. See [Plans & billing](/billing/plans) for the prompt library size on each plan. ### Analysis results look wildly different between runs Some variance is normal — LLMs aren't deterministic, and Perplexity especially can be noisy run-to-run. That said, a 20%+ swing in AI Visibility between adjacent runs usually means: * One of the LLMs had downtime or degraded quality during the run (rare, but happens). * Your site changed significantly (e.g. migration, CDN issue) between runs. * Your prompt set changed — did you edit keywords or target audience? Compare the per-model breakdown: if one model swung hard and the others held steady, it's probably that one model's volatility rather than a real visibility change. ## Reports and email ### A recipient isn't getting reports 1. Confirm the address is listed in **Settings → Reporting → Recipients**. 2. Ask the recipient to check spam/junk. 3. Check **Settings → Notifications** for a bounce against that address. 4. If there's a bounce, the fix is usually on the recipient's mail server — add `reports@videntic.com` to their allowlist or remove the forwarding rule that's dropping the mail. ### The PDF is stripped from my emails Some corporate mail gateways strip attachments from external senders. The email body contains an inline summary plus a link to view the report online. If you need the PDF, ask IT to allow PDF attachments from `@videntic.com`. ### Reports are empty or missing sections Usually means the analysis hasn't completed yet. Check the analysis run status on the website dashboard — if it's still running, the report generated against incomplete data. Re-run the analysis and regenerate the report. ## Sira ### Sira is writing generic or off-brand content Upload at least one **brand document** (guidelines, tone-of-voice, a best-in-class past article) and ask Sira to try again. The drafts Sira produces without any brand context are necessarily generic — it has nothing to anchor to. See [Work with Sira](/how-to/work-with-sira#brand-documents-change-everything). ### Sira is answering about the wrong website Check the website switcher in the top-left of the chat. Sira's context is scoped to the currently selected site. Switch, then rephrase the question. ### A Sira task has been running for an hour Long tasks — full-site audits, multi-article content plans — can take that long. Check the task card in the conversation for progress. If it's clearly stuck (no progress messages for 15+ minutes), email support with the task ID. ## Integrations ### Shopify install fails with "Permission denied" Your Shopify role needs **Edit articles**. Either ask an admin to grant it, or install the integration under an admin account. ### WordPress says "Authentication failed" Two common causes: * You entered your *email address* where the username field is — WordPress needs the username. * A security plugin (Wordfence, iThemes) is blocking REST API access. Allowlist Videntic, or temporarily disable the plugin to test. ### Drafts appear in the wrong blog/category Set the default blog (Shopify) or default category (WordPress) in the integration settings. Videntic pushes there unless you override per-draft. ## Billing ### Downgrade didn't take effect Downgrades take effect at the end of the current billing cycle, not immediately. Until then, you keep your higher-tier features. This is intentional so you don't lose what you paid for. ### I was charged after cancelling Check the charge date — Stripe sometimes processes a final renewal charge if the cancel happened after the billing cutover. Email [billing@videntic.com](mailto:billing@videntic.com) with the invoice number and we'll review. ### "Payment failed" notification but my card works fine Usually a 3D Secure / SCA challenge your bank issued that wasn't completed. Go to **Settings → Billing → Manage billing in Stripe** and retry the latest invoice — the portal will walk you through authentication. ## Still not working? If the fix here doesn't solve your problem — or the issue isn't listed — email [support@videntic.com](mailto:support@videntic.com) with: * The website name / URL affected. * The approximate time it happened. * A screenshot if the error is visual. * What you'd already tried. We usually reply within a business day; Pro and Max customers get priority support. # FAQ Source: https://docs.videntic.com/troubleshooting/faq Common questions about Videntic — product, data, billing, and privacy. Short answers to the questions that come up most often. For anything that's not here, email [support@videntic.com](mailto:support@videntic.com). ## Product ### How is Videntic different from a regular SEO tool? Traditional SEO tools track where you rank in Google's organic search results. Videntic tracks whether LLMs — ChatGPT, Perplexity, Gemini, Claude — cite you in their answers. The techniques that move Google rankings don't automatically move AI visibility; Videntic measures the new channel, audits your site for it, and tells you how to improve. ### How often does Videntic re-run my analyses? Depends on your plan's cadence. Daily scheduled analysis is available on Pro, Max, and Enterprise; Basic runs twice weekly (Monday & Thursday). You can also run an analysis on demand at any time. Report delivery cadence is configured per website in Settings → Reporting. ### Can I run an analysis on demand? Yes. Click **Run analysis** from the dashboard any time. An on-demand run executes your active prompt library — it doesn't draw down a separate monthly budget. ### Which LLMs does Videntic test against? ChatGPT, Perplexity, Google AI Overviews, and Gemini. The list evolves as AI search evolves; we prioritize the models real customers ask our customers' categories about. ### How long before I see results after publishing a new blog post? Usually one to two analysis cycles. LLMs need to re-encounter the new page in their crawl or index before citations shift, so expect 1–3 weeks depending on the model. Perplexity tends to pick up new content fastest; ChatGPT is slower. ### Does Videntic publish content on my behalf? No. Sira can draft content and push it to Shopify or WordPress as an **unpublished draft**. The final publish action is always yours. ## Data and privacy ### Does Videntic train AI models on my data? No. Your content, your analyses, your drafts, and your uploaded brand documents are used only to serve your workspace. They're not used to train shared models or any other customer's data. ### Where is my data stored? In the EU. Our infrastructure runs in European data centers and we use EU-hosted subprocessors where available (see our [privacy page](https://videntic.com) for the current list). ### Can I delete my data? Yes. Cancelling your subscription keeps your data accessible so you can pick up later. If you want permanent deletion, email [support@videntic.com](mailto:support@videntic.com) and we'll wipe your account — permanent action, no undo. ### Who can see my data inside Videntic? Your workspace is private to the accounts invited to it. Videntic employees can only access customer data with explicit permission for support purposes and under audit. A small number of senior engineers have standing access to production for incident response. ## Billing ### How does the free trial work? Every new account gets a **7-day free trial** of a paid plan. You pick Basic, Pro, or Max at signup and get full access to that plan — including its prompt library size, competitor slots, and features — for the length of the trial. You won't be charged until the trial ends, and you can switch plans or cancel any time before then. See [Plans](/billing/plans). ### Can I change plans later? Yes, any time. Upgrades are prorated and take effect immediately. Downgrades take effect at the end of the current billing cycle. See [Managing your subscription](/billing/managing-your-subscription). ### Do you offer annual billing? Yes, for Max and Enterprise plans. Email [billing@videntic.com](mailto:billing@videntic.com). ### What happens if my payment fails? Stripe retries on its standard schedule, and Videntic gives you a \~7-day grace period with full access after the last retry fails. You won't suddenly lose access on a failed charge. ### Does Videntic charge VAT? Yes, where applicable. Invoices include VAT for EU customers. Add your VAT ID in the Stripe Customer Portal (Settings → Billing → Manage billing in Stripe) to get it reflected on invoices. ## Team and access ### Can I invite my teammates? Team accounts are rolling out gradually. For now, every Videntic account is single-user. Reports can be emailed to an unlimited number of recipients (on Pro and Max), so stakeholders can follow progress without a login. ### Can my agency manage multiple clients from one account? Today, each client would need their own Videntic account. Multi-client agency features are on the roadmap — [contact us](https://videntic.com/contact) if you want to talk about your use case. ## Integrations ### Which CMSes do you support? Shopify and WordPress. See [Shopify](/integrations/shopify) and [WordPress](/integrations/wordpress) for setup. ### Do you integrate with Google Analytics? Not directly yet. We're evaluating a GA integration that would let Videntic correlate AI visibility with downstream traffic. If that would be valuable to you, let support know. ### Can I export my data to Looker / BigQuery / a warehouse? Not yet. Reports export as PDF; programmatic data export is on the roadmap. For urgent needs, email support. ## Still stuck? [Email support](mailto:support@videntic.com). We typically respond within a business day; Pro and Max customers get priority support with a faster SLA.