AEO Orchestra — Complete Manual

Stai cercando la guida definitiva per AEO Orchestra? Sei nel posto giusto! Questo è l’unico orchestra complete manual di cui avrai bisogno per padroneggiare ogni aspetto del plugin. Qui troverai una copertura esaustiva di ogni funzionalità: cosa fa, i vantaggi che ti offre, il costo in crediti e come configurarlo al meglio in ogni scenario (sito classico, page builder

1. How AEO Orchestra works: SEO + AEO, specialists and credits

1.1 SEO and AEO: why you need both

Traditional SEO optimizes pages for classic search engines: title, meta description, schema, sitemap, redirects. AEO (Answer Engine Optimization) optimizes those same pages for AI answer engines — ChatGPT, Claude, Gemini, Perplexity, Grok, Copilot, Meta AI, DuckDuckGo — which answer questions directly, citing 2-3 sources. If your pages aren’t structured to be citable (direct answers, FAQs, schema, llms.txt), you simply don’t exist for AI engines, even if you rank well on Google.

AEO Orchestra does both from a single panel, with an architecture of AI specialists (21 tools) coordinated by an Orchestrator that analyzes the site and turns every problem into a concrete action.

1.2 The architecture in brief

The plugin lives in your WordPress and directly manages the SEO output (head, sitemap, llms.txt, schema, redirects). The artificial-intelligence operations (analyses, generations, rewrites) are performed by the external AEO Orchestra service: the plugin sends the necessary context, the service returns the proposal. Nothing is applied to the site without your explicit approval.

1.3 Credits: the golden rule

Credits are the currency of AI operations. Three principles hold everywhere in the plugin:

• Previews are free. Clicking “Run” on an action generates the proposal and shows it to you at zero cost.
• You pay only when you apply. You are charged when you click “Apply changes”, never before. “Cancel” = no charge.
• Everything is tracked and reversible. Every operation has a row in the log (Credit Usage page) and every applied change can be restored for 7 days.

The cost of each operation is always stated on the button before you click (e.g. “Generate Meta Tags — 2 credits”). You’ll find the complete cost table in chapter 6.

2. Installation, license and privacy

2.1 Installation

1. From WordPress: Plugins → Add New, search for “AEO Orchestra” (free version) or upload the ZIP you received with your license (Plugins → Add New → Upload Plugin).
2. Activate the plugin: the AEO Orchestra entry appears in the sidebar menu.
3. Go to AEO Orchestra → Settings → AEO Orchestra Service, paste your Service API Key and click Verify Service, then Test Connection.

💡 The trial offers 7 days and 50 credits, no card required. Licenses are lifetime (Starter 1 site, Professional multi-site, Team up to 10 users/sites) with rechargeable credits.

2.2 Consent and privacy (GDPR)

When the license is activated, the plugin asks for explicit consent: only after consent does the plugin make network calls to the service (which receives the site’s domain and the Business Profile data required by the AI services). Without consent, no call is made. Consent is recorded with date and time and can be reviewed in Settings. Business Profile data has three visibility levels (public / context / private internal use) described in chapter 5.3.

2.3 Interface language

The panel is available in Italian, English, Spanish, French and German: Settings → Interface language, save and reload the page. The interface language is independent of the language of generated content (which follows the language of your site and your requests).

3. Compatibility: editors, page builders and headless sites

This is the most important section for understanding which AI actions the plugin can apply automatically on YOUR site. WordPress exists in many different configurations, and Orchestra adapts accordingly.

3.1 Automatic detection

In Settings → Site Compatibility the plugin shows what it has detected: the page builder in use (with a confidence percentage), whether the site is headless, and the resulting effective environment. From this comes the capability map: for each type of AI action, the plugin states whether it will be applied in automatic, automatic-with-verification, or manual mode.

3.2 Supported WordPress configurations

3.3 The per-action capability map

Real example of a map on a headless REST site:

💡 Why do actions sometimes end up in “Do manually”? It’s not a bug: it’s honesty. If the plugin detects that a surface of your site (typically the body on Git-based headless frontends or on certain builders) cannot be modified reliably from WordPress, it puts those actions in the manual group instead of pretending to apply them. Head actions (meta, schema, FAQ via head) remain automatic.

3.4 Manual override

Automatic detection isn’t infallible. In Settings → Manual override you can force: the page builder (Classic, Gutenberg, Elementor, Divi, WPBakery, Beaver Builder, Bricks, Oxygen, other), the headless state (Yes/No/Auto) and the headless mode (REST API / WPGraphQL / SSG-ISR). After Save override the per-action modes recalculate on their own. There’s also Re-scan site to run detection again from scratch.

4. Configuration by scenario

Five typical situations, with the right settings for each.

4.1 Classic site (blog, brochure site, business)

1. Complete the Guided Setup (7 steps): the first analysis of your homepage is free.
2. Fill in the Business Profile to 100% (or generate it from your site with one click, 30cr): it feeds all the AI specialists.
3. Enable the native stack in SEO+AEO Output: head, sitemap, llms.txt, schema. If you have no other SEO plugins active, turn everything ON.
4. Create your Brand Voice (25cr) from your best articles.
5. Launch the full analysis from the Orchestrator and work through the action plan.

4.2 Migrating from Yoast / RankMath / AIOSEO

Use SEO Migration (6-step wizard): 1) Scan (read-only: detects the active plugin, counts meta and redirects); 2) Automatic JSON backup; 3) Migrate meta — the per-post metadata import brings in title/description/canonical/robots/OG/Twitter for every single article (“without this, you lose 5 years of metadata”); 4) Import redirects (including those from Yoast Premium and the Redirection plugin); 5) Enable the native stack; 6) Uninstall the old plugin. Every step is reversible.

⚠️ Don’t keep two SEO plugins active with simultaneous head output: you’d get duplicate meta tags. Complete the migration or disable the other plugin’s output. To check: CTRL+U on a page and look for duplicate <meta name="description"> tags.

4.3 E-commerce and Custom Post Types

Orchestra automatically discovers public post types (products, services, custom). Check Settings → Page discovery for the right source: wp_posts (default), Frontend sitemap, Manual URL list or Auto-Crawl. The native sitemap indexes all CPTs. For the AI article Layout you can choose which post type generated content is created in (post, page, product, or custom CPT).

4.4 Headless site (Next.js/Astro/Gatsby on Vercel/Netlify)

The most powerful scenario and the one that requires the most attention. In Settings → Headless Mode:

1. Enable Headless mode: the analyzer will download the HTML from the frontend instead of reading the raw post_content (so the analysis judges what users and engines actually see).
2. Frontend base URL (HTTPS only) + Check reachability. The fetch cache lasts 1 hour.
3. Deploy Hook URL: after every applied change, Orchestra calls this hook to trigger a frontend rebuild (Vercel: api.vercel.com/v1/integrations/deploy/… · Netlify: api.netlify.com/build_hooks/… · or a custom endpoint). The URL contains a token: treat it like a password.
4. Anti-burst window (1-60 min): at most 1 rebuild per window — consecutive applies are batched together, with a queued trigger at the end of the window. For large batches 5-10 min is recommended.
5. Page discovery: for headless sites without a sitemap, Auto-Crawl is recommended; if the frontend exposes a sitemap.xml, use that.

How changes propagate: you apply in WordPress → the change enters the manifest (the endpoint the frontend consumes at build time) → the deploy hook restarts the rebuild → the public served site updates. In the Orchestrator a banner shows the changes “awaiting propagation” with the Propagate now button to force an immediate rebuild.

💡 The reachability probe is “deferred-aware”: it knows that on static frontends changes aren’t visible instantly, and it verifies the real chain (manifest, hook, evidence on the served page) instead of reporting false negatives.

4.5 Agencies and multi-brand

You can create multiple Brand Voice profiles (one per client or editorial line) and activate the one you need; only the active profile guides generations. The Designated author in the Business Profile defines who is cited in llms.txt. The score widget branding depends on your plan (Starter shows “Powered by AEO Orchestra”; Professional/Team can hide it). The Team license covers up to 10 users/sites with centralized credits.

5. The features, one by one

For each tool: what it does, what you get, how much it costs, how to use it, and the typical mistakes to avoid.

5.1 Dashboard

What it does: your at-a-glance view of everything: license status, active native stack, Brand Voice, available specialists, and access to every section, grouped by phase (Foundation & Profiles → Analysis & Verification → Creation & Generation → Operations → Account). What you get: immediate orientation and the Guided Setup bar with your progress. Shortcut: the Command Palette (CTRL+K) opens any tool from the keyboard.

5.2 Guided Setup

What it does: adapts Orchestra to your site in 7 steps, resumable whenever you want; each step can be marked “completed” or “to redo”. It includes the free first analysis of your homepage (SEO+AEO score, detected issues, action plan with “Fix with AI” buttons). What you get: a correctly configured plugin in ~15 minutes, without having to know every setting. Typical mistake: skipping it and launching massive analyses right away — an empty Business Profile degrades the quality of ALL generations.

5.3 Business Profile

What it does: it’s the identity card the AI uses in every operation. 19 fields across three visibility levels:

• Exposed to AI engines (goes into llms.txt, JSON-LD schema, Open Graph): company name, industry, description, public about, value proposition, target audience, founding year, public offering (plans/pricing as structured text — never internal strategy), designated author, verifiable facts (max 10), differentiators (max 7), products/services (max 15), use cases (max 5), partners (max 5), service area (max 8).
• Site context (injected as an authoritative block into every analysis): what the site does, detailed value proposition, detailed target audience, and the ambiguous terms to disambiguate (max 12) — essential so that AI engines don’t confuse your brand with namesakes.
• Internal use (stays private, NEVER exposed): direct competitors (max 10, for comparative analyses), strategic notes, pricing strategy.

What you get: more accurate generations, brand disambiguation with AI engines, and a rich llms.txt. Shortcuts: “Generate complete content from site” (30cr) populates every field by analyzing your homepage + about + services pages; “Generate automatically from site” (3cr) does the same for the Context only. The AI Context Preview shows the exact block the engines will receive, in full or public-only scope. Typical mistake: writing your offering with unverifiable claims — use only facts your site supports.

5.4 Brand Voice (25 credits per profile)

What it does: the AI analyzes ~15 of your recent pieces of content (blog posts, products, pages, or “Front pages” via sitemap scrape for page-builder/headless sites) and produces a style profile in ~30 seconds: tone, distinctive vocabulary, typical structure, anti-clichés. What you get: every future generation (articles, meta, AEO content) sounds like you, not “like AI”. Multi-profile: create as many as you want (per brand/client), only one active at a time; each profile can be edited by hand after creation. Typical mistake: creating the voice with fewer than 5 substantial pieces of content (min 200 characters each) — the profile comes out generic.

5.5 Orchestrator — the operational core

What it does: scans the pages you choose and produces: the site’s Health Score (0-100 with trend), SEO and AEO scores per page, “Pages needing attention”, and the Checks — every single check becomes an opportunity classified into three groups: Do with Orchestra (the AI does it for you), Do manually (needs your hand), Passed checks.

How to use it:

1. New analysis: select the pages (All / None / Pages only / Posts only / manual selection); the credit estimate is shown up front (10 credits/page: 5 SEO + 5 AEO). The homepage in the Guided Setup is free.
2. During the analysis: page-by-page progress bar; on large sites automatic checkpoints are saved every 5 pages — if the run is interrupted (tab closed, network), when you return you’ll find the “partial X/Y” state with the Resume button, which restarts from the exact point without re-charging for pages already analyzed. “Cancel” mid-run saves the partial results with the actual credits spent.
3. Working through the worklist: each action tile has “Run” (free preview) → modal with the proposal, “what will change on your site” and the estimated score impact → Cancel (0 credits) / Regenerate (new proposal) / Apply changes (charged only now). “I don’t want to make this change” archives the suggestion.
4. Analysis history: every run is saved and can be reopened; multi-page runs persist and survive reloads. Recent changes: every apply is listed with a diff and Restore (7 days).
5. On headless sites: the propagation banner shows the changes awaiting a rebuild, with “Propagate now”.

Typical mistake: judging the results from the frontend right after an apply on a headless site — wait for the rebuild (or force it). Also: the action count legitimately changes from one analysis to the next, because the site has changed.

5.6 SEO Analysis (5 credits/page)

What it does: single-page analysis (or a manual URL, even external) with a primary keyword: score 0-100, detected issues, secondary LSI keywords, actionable suggestions. When to use it: to dig into a specific page without re-analyzing the whole site; the history keeps previous runs.

5.7 AEO Analysis (5 credits/page)

What it does: checks how the page is perceived by AI engines (Google AI Overviews, ChatGPT, Claude, Perplexity, Bing Copilot…): AEO score, citability gaps, GEO suggestions. When to use it: on strategic pages (services, flagship products) with the target keyword you want to own in AI answers.

5.8 Verify-Live

What it does: the truth test — it queries the 8 AI engines in ~30 seconds and measures your Citation Score (0-100): is your site being cited? With 8-15 specific suggestions to improve visibility where it isn’t. What you get: the real, not theoretical, measure of your presence in AI answers — no other plugin offers it. Tip: run a Verify-Live before and after an optimization cycle to measure progress.

5.9 AI Performance

What it does: tracks the activity of 22 AI bots on your site with GDPR-compliant data: top 10 pages visited by bots, 28-day trend, robots.txt compliance. What you get: proof that AI crawlers are reading your content (or a warning if they’re not showing up — see troubleshooting 7.8).

5.10 SEO Cannibalization (free scan)

What it does: finds pages competing for the same focus keyword (when 2+ pages target the same query, Google struggles to choose and penalizes both). The AI proposes the primary page, alternative keywords for the supporting pages, and internal links — applicable in one click, reversible. When to use it: on sites with lots of similar content (old blogs, local pages, catalogs).

5.11 Keyword Research Autopilot (15 credits/research)

What it does: from a generic niche to 20/30/50 strategic keywords organized into semantic clusters, with intent, difficulty and article idea, in 5 languages. What you get: work that would take hours on SEMrush/Ahrefs, ready to be “pulled” into the other tools: wherever you see “Pull from Keyword Research” you can reuse past research. Tip: the more specific the niche description, the better the clusters.

5.12 Analytics

What it does: KPIs no other SEO plugin measures, with the Google Search Console integration: AI articles that made it into GSC’s TOP 10, Brand Voice impact, Auto-Pilot ROI, “saved” redirects, Meta Freshness. What you get: the full circle closed — how much what Orchestra generates actually pays off.

5.13 AI Content (10 credits)

What it does: generates complete SEO articles with H2/H3, optional FAQ, summary and CTA, in three sizes (~500 / ~1000 / ~2000 words), with your active Brand Voice. What you get: a publishable draft (or directly a new post) + keyword suggestions for your industry. The history keeps your generations.

5.14 AEO Content (12 credits)

What it does: the answer-first variant: articles with a question-answer structure, Schema.org markup and a structured FAQ section, with target engine selection (Google AI Overviews/Gemini, ChatGPT, Claude, Perplexity, Grok, Copilot, Meta AI, DuckDuckGo). What you get: content built to be cited; the topic suggester points to the topics with the highest citation potential in your industry. When to choose AEO vs AI: AEO to own questions (“how much does X cost”, “how to do Y”), AI for classic editorial content.

5.15 Meta Tags (2 credits per page)

What it does: generates optimized title, description and keywords for a page (with an optional focus keyword) or in bulk generation across multiple selected pages. What you get: consistent, fresh heads across the whole site at minimal cost. SERP preview included before applying.

5.16 SEO Images

What it does: a full audit of the media library: which images are missing alt, title, caption, description (score 0-4 per image), with single or bulk AI fix, filters, sorting and CSV export. What you get: image metadata in order — estimated traffic recovery of +10-25% from Google Images. Safety: here too, preview before applying in bulk.

5.17 Local SEO (4 credits per page)

What it does: generates pages optimized for local search (service + city, even multiple cities separated by commas) in three templates: Service Page, Landing Page, Informational Article. Includes LocalBusiness schema and Google Business Profile (NAP) sync. What you get: coverage of “service + city” searches without writing dozens of pages by hand. Tip: always customize the generated pages with real local details (references, job sites, reviews) — cookie-cutter local pages perform worse.

5.18 Article Planning (Content calendar)

What it does: an editorial calendar with slots: for each slot you define when to generate the draft (same day / 1-7 days before), the publication time, whether to auto-publish or leave as a draft, and the category. Defaults are set in Settings → Content calendar. What you get: an editorial plan that runs itself, with human review wherever you want it.

5.19 Auto-Pilot Scheduler (~25 credits/article, +15 with AI image)

What it does: recurring jobs that generate complete articles (text + image + meta) with your Brand Voice: you choose the keyword set (from Keyword Research), the frequency (daily → monthly), the length, the status (draft for review or direct publication), category and author. A WP cron checks the active jobs every hour and, when it’s time, pulls the next keyword from the set and creates the post. What you get: a blog that grows on its own, with the “awaiting review” queue for human control and the execution log. Prudent tip: start with “Draft (manual review)”; switch to auto-publish only when the quality consistently satisfies you. Technical note: WP cron only fires when the site receives visits — on very low-traffic sites, set up a system cron.

5.20 Redirect Manager

What it does: 301/302/307/308 redirect management from the WP database, with regex and backreferences, import from Yoast Premium or the Redirection plugin, a 404 log of hits received by the site, and an AI suggester that proposes the right redirect for each 404. What you get: it replaces Yoast Premium Redirects / Redirection / the RewriteRule entries in .htaccess, and recovers traffic from broken links. It toggles on/off from its own page (off by default until you need it).

5.21 SEO Migration

See scenario 4.2: a reversible 6-step wizard with JSON backup and per-post metadata import.

5.22 AI Crawlers

What it does: dedicated control of AI crawlers, in three tabs: Allowlist with toggles for 22+ bots grouped by provider (OpenAI: GPTBot, OAI-SearchBot, ChatGPT-User · Anthropic: ClaudeBot, Claude-SearchBot, Claude-User · Perplexity, Google-Extended, GoogleOther, Bingbot, Meta, Grok/xAI, DeepSeek, Applebot-Extended, CCBot, Cohere, Mistral, YouBot, Amazonbot, Bytespider, Diffbot…); Crawler Log with real visits from the last 30 days (IP logging can be disabled for GDPR privacy mode); Robots.txt generated automatically from your choices. What you get: you decide who can use your content for training and who for live citations — an important distinction: blocking the training bots (GPTBot, ClaudeBot) does NOT block the citation bots (OAI-SearchBot, Claude-SearchBot), and Bingbot serves both SEO and Copilot: blocking it also removes Bing search. Note: blocking is a robots.txt directive — a signal that well-behaved bots follow, not a server-side wall.

5.23 SEO + AEO Output (native stack)

What it does: the output engine that replaces Yoast/RankMath/AIOSEO, in 4 individually activatable modules:

• <head> output: title with site suffix, meta description (from focus keyword or auto-excerpt), canonical, full OpenGraph, Twitter Cards, robots (automatic noindex on search/404), published/modified time for articles. To verify: CTRL+U and look for <!-- AEO Orchestra · Native Output -->.
• Sitemap.xml: exposed at /seo-aeo-sitemap.xml, indexes all CPTs and public taxonomies, 6h cache auto-invalidated on every change, automatically added to robots.txt. THIS is the URL to submit to Google Search Console.
• llms.txt + llms-full.txt: the markdown maps for Large Language Models (llmstxt.org spec, adopted by Anthropic, Stripe, Vercel, Cloudflare): /llms.txt = curated index (titles + descriptions), /llms-full.txt = full markdown content of the top 30 pages. 6h cache, invalidated on every save. GPTBot, ClaudeBot, PerplexityBot and Gemini find them and use them for answers.
• Dynamic Schema.org: JSON-LD @graph for every page (Homepage: Organization + WebSite with SearchAction; Articles: Article + WebPage + BreadcrumbList + ImageObject + Person; Pages: WebPage + BreadcrumbList; CPTs/taxonomies: CollectionPage; author archives: ProfilePage). Verify with the Google Rich Results Test.

The Verification & languages tab helps you check that the output is clean and not duplicated by other plugins.

5.24 Score Widget

What it does: a SEO+AEO badge at the bottom right of the site showing the current page’s score. Configurable visibility: Admins only (private monitoring) or All visitors (public transparency). Branding depends on your plan (see 4.5).

5.25 Credit Usage

What it does: the accounting ledger: total credits used, total operations, most-used agent, credits today; breakdown by agent; daily usage (14 days); the Operations Log row by row (date, agent, credits, page, detail, user) with an agent filter. In Settings → Credit history you’ll also find the last 50 wallet transactions (top-ups, usage, bonuses). Rule: when in doubt about costs, the log is the truth.

5.26 AI Copy Coach and Regenerate entire page

AI Copy Coach: a rewriting assistant that improves existing text while keeping your Brand Voice. Regenerate entire page: a complete rebuild of a weak page starting from its content and your business context — useful for the “nearly empty” pages flagged by the analysis.

5.27 Command Palette

CTRL+K (CMD+K on Mac) anywhere in the admin: search and open any Orchestra tool, launch frequent actions. The fastest way to move around the plugin.

6. Credits: costs, transparency, refunds and reversibility

6.1 Cost table

Costs are always stated on the button before you click. Indicative values shown in the interface (they may vary with pricing updates):

6.2 The guarantees

• No charge without Apply. Cancelling a preview costs nothing; a failed operation is not charged.
• Refund window: 5 minutes after an Apply to request a refund (max 3 refunds/day, anti-abuse). “You only pay for what you approve.”
• 7-day reversibility: every applied change is in the Recent changes panel with Details (diff) and Restore.
• Complete log: every credit has a row in Credit Usage; the wallet and the transaction history are always available.

7. Troubleshooting

The most common real-world situations, with cause and solution. In order of frequency.

7.1 “I applied a change but I don’t see it on the site”

Headless site: it’s almost always propagation. Changes travel with the frontend rebuild: check the banner in the Orchestrator (“N changes awaiting propagation”) and use Propagate now, or wait for the anti-burst window. The Vercel/Netlify rebuild takes a few minutes. Traditional site: clear the cache (caching plugin, CDN, browser cache: CTRL+F5). Objective check: CTRL+U on the public page and search for the applied content (title, FAQ block, JSON-LD script) — if it’s in the source, it’s just visual caching.

7.2 “The analysis stopped halfway” (tab closed, connection lost)

No work is lost: the incremental checkpoint saves every 5 pages. Reopen the Orchestrator: you’ll find the “partial X/Y” state with Resume, which restarts from the exact point without re-charging for pages already analyzed. Still, keep the tab open during analyses: the run is driven by the browser.

7.3 “All the actions ended up in Do manually”

Check Settings → Site Compatibility → Per-action capabilities. Typical causes: (a) on your frontend the body cannot be modified reliably from WordPress (Git-based headless, certain builders) — body actions stay honest and manual, while head actions (meta, schema, FAQ) must be automatic; (b) detection got the environment wrong → use the Manual override and save; (c) on headless, the frontend reachability check temporarily failed → check the Frontend URL and deploy hook, then Re-scan site. If the “Do with Orchestra” group is still empty after these checks, contact support with a screenshot of the capability map.

7.4 “Spinner keeps spinning for a long time after Run”

AI generation can take from a few seconds to 1-3 minutes at peak times (FAQs, which generate the most text, are the slowest). Don’t click again and don’t reload: no credits are charged until you press Apply. If nothing appears after a few minutes, close the modal and try again: the regenerated preview is still free.

7.5 “Error or rejection after the page was left open for a long time”

WordPress sessions expire (nonce): the plugin has an automatic self-heal that renews the token and retries. If an operation still fails, reload the page and try again — the worklist state is persisted server-side and isn’t lost on reload.

7.6 “The worklist counts changed after a reload” / “Tiles I just ran don’t show the badge”

The displayed worklist comes from the persisted snapshot: after many consecutive applies the page can redraw with momentarily misaligned counters. The truth is always server-side: do a hard refresh (CTRL+SHIFT+R). If your doubt is about charges, the log in Credit Usage is authoritative row by row.

7.7 “llms.txt / sitemap aren’t updating”

Both have a 6-hour cache, invalidated automatically whenever content is saved. If you don’t see the update after a change: save any piece of content again (this forces invalidation) and check again. On headless sites remember that the public domain’s /llms.txt must be served by WordPress (a rewrite on the frontend) — if the frontend serves its own static file, you need to configure a proxy/rewrite to WP.

7.8 “AI bots aren’t visiting the site”

Check in order: (1) AI Crawlers → Robots.txt: any unintentional blocks? (2) AI Crawlers → Crawler Log: the real visits from the last 30 days — crawlers come on their own schedule, from hours to days; (3) llms.txt active and publicly reachable; (4) sitemap submitted to GSC (/seo-aeo-sitemap.xml); (5) patience: after optimizations, visibility in AI answers builds up over days/weeks, not minutes. Measure with Verify-Live after 1-2 weeks.

7.9 “Duplicate meta descriptions / two titles on the page”

You have two SEO plugins with head output active at the same time. Complete the SEO Migration (4.2) or disable the other plugin’s output. Verify with CTRL+U: there should be the <!-- AEO Orchestra · Native Output --> block and ONE single meta description.

7.10 “The displayed cost doesn’t match the charge”

The operations log (and the wallet history) is authoritative. If you find a real discrepancy between the label and the charge, report it to support: pricing labels are aligned with the backend. Remember the 5-minute post-Apply refund window.

7.11 “Health Score dropped after the changes”

The score only recalculates with a new analysis, and a new analysis can find new things (the site changes, the criteria are updated). Look at the details: if passed checks are growing and the remaining opportunities are in manual/content groups, you’re improving even if the raw number fluctuates. Always compare the same scope (the same pages analyzed).

7.12 “Headless site: Check reachability fails but the site is up”

The probe verifies the complete chain (frontend reachable, manifest consumable, deploy hook healthy). Typical causes: Frontend URL with a redirect or without HTTPS; deploy hook expired/rotated (regenerate it on Vercel/Netlify and paste it again); temporary rate-limiting on the frontend (the probe distinguishes “endpoint alive but rate-limited” from “dead” — retry in a few minutes). The frontend fetch has a 1h cache: after a fix, run the check again.

7.13 “Auto-Pilot isn’t generating at the scheduled times”

WordPress cron only fires when the site receives traffic. On low-traffic sites set up a system cron that calls wp-cron.php every 15 minutes, or an external ping service. Also check the Latest runs log on the Auto-Pilot page and that the job is “active” with keywords still available in the set.

7.14 “I applied something I shouldn’t have”

Orchestrator → Recent changes → Restore (within 7 days). Each entry shows the exact diff (+N bytes, what changed). On headless, the restore follows the same path as the apply: it lands with the next rebuild.

7.15 “The score widget doesn’t appear on the site”

Check Settings → Score Widget: if visibility is “Admins only (logged in)”, visitors don’t see it (and you only see it while logged in). Remember the mandatory branding on the Starter plan.

7.16 “Can I use the plugin without sending data to the service?”

The AI features require the external service (that’s where the intelligence runs). Without consent, the plugin makes no network calls at all: the local native features remain usable (head output, sitemap, basic schema, redirects). Consent is revocable and the data exposure levels can be controlled field by field in the Business Profile.

8. Glossary

AEO (Answer Engine Optimization)

Optimization for AI answer engines: making content citable in the answers of ChatGPT, Claude, Perplexity, Gemini, etc.

llms.txt / llms-full.txt

Standard markdown files (llmstxt.org spec) that map the site for Large Language Models: a curated index the first, full content of the main pages the second.

Schema / JSON-LD

Structured markup that describes the page to Google and AI engines (Organization, Article, FAQPage…). Orchestra generates it as a dynamic @graph.

Health Score

The site’s 0-100 score calculated by the Orchestrator analysis (SEO + AEO combined).

Citation Score

Verify-Live’s measure: how much your site is cited by the 8 AI engines.

Worklist / Checks

The list of opportunities produced by the analysis, divided into “Do with Orchestra”, “Do manually”, “Passed checks”.

Manifest (headless)

The endpoint that exposes all applied optimizations (meta, schema, FAQ) to the frontend; the frontend consumes it at build time.

Deploy hook

A secret URL that triggers the headless frontend’s rebuild; Orchestra calls it after applies, with an anti-burst window.

ISR / SSG

Rendering modes of modern frontends (incremental regeneration / static generation): changes become visible after a rebuild.

Brand Voice

A style profile extracted from your content and applied to every AI generation.

Credits

The currency of AI operations: free previews, charged only on Apply, complete log.

Incremental checkpoint

Automatic save of the analysis every 5 pages: interrupted runs resume without re-charges.

Manual v1 — updated for AEO Orchestra 3.74.x. For support: /guide · documentation · support@aeo-orchestra.com · WhatsApp at the bottom right.