How Privora Works: A Complete Guide to Tracking and Improving Your AI Visibility

If you've ever wondered why your competitors keep showing up in ChatGPT answers while your brand goes unmentioned — or if you just signed up and aren't sure where to start — this guide walks through every part of Privora from first login to your first content improvement.

The core idea: measure first, then act

Most brands approach AI visibility backwards. They assume they're visible, or they publish a few FAQs and hope for the best. Privora inverts that: you measure your actual visibility score first, understand exactly where and why you're missing, and then get precise instructions for what to create.

The platform is built around four stages: Track → Score → Diagnose → Recommend.

Step 1: Create your brand profile

When you first log in, you create a Brand. This is the entity Privora tracks across ChatGPT, Claude, and Perplexity. A complete brand profile includes:

Brand name and aliases — all the ways your brand is referred to (e.g. "Acme", "Acme Inc", "Acme Software"). Privora checks for all of them in AI responses.
Website URL — used to provide context for content suggestions and schema markup recommendations.
Description — a 2–4 sentence summary of what your product does and who it's for. The more specific this is, the more targeted your content recommendations will be.
Industry — your product's category (e.g. "Project Management", "E-commerce", "HR Tech"). This shapes which query templates are suggested and how suggestions are framed.
Product type — whether you're a SaaS platform, physical product, service business, mobile app, or content creator. This matters because ChatGPT, Claude, and Perplexity each recommend different types of businesses differently.
Target market — your geography and audience (e.g. "United States — B2B SMBs" or "Global — DTC fitness consumers"). A brand targeting European enterprises needs different content than one targeting US consumers.
Tracked keywords — the terms you want to appear for ("best invoicing software", "affordable CRM for freelancers", etc.)

Why this matters: Privora's AI-powered content suggestions are generated with your full brand profile in context — not generic AEO advice. The more detail you provide, the more specific and actionable your recommendations will be.

Step 2: Add your competitors

In the Brands → Competitors section, add the brands you compete with for AI recommendations. Privora assigns each competitor an AEO visibility score as you run queries, so you can see side-by-side how your brand stacks up.

When you generate content suggestions, Privora shows the LLM your competitive landscape — so instead of saying "write an FAQ about your product," it says "write an FAQ that positions you against Competitor X, who currently has a higher AEO score for this query."

Step 3: Build your query list

Queries are the actual questions your potential customers type into AI assistants. They are the unit of measurement in Privora.

Go to the Brands → Queries section and add the questions you want to monitor. There are five query types:

Type	Example
Best	"best CRM for small businesses"
Comparison	"Acme vs Competitor comparison"
Category	"project management tools for remote teams"
How-to	"how to automate invoicing for freelancers"
Brand	"what is Acme and how does it work"

Query templates by industry are available to help you get started quickly. For SaaS products, common high-value queries include "best [category] tools", "affordable alternative to [big competitor]", and "how to [solve core problem]".

Start with 5–10 queries across different types. Avoid queries so narrow that only your brand would be mentioned — the most valuable queries are the ones where several competing brands could all plausibly appear.

Step 4: Run queries and read your visibility score

Once your queries are set up, run them from the Test Query page. Privora fires the query simultaneously to ChatGPT (GPT-4o), Claude (Sonnet), and Perplexity (Sonar) — with no system prompts or brand-nudging, so you're measuring the same experience your customers get.

For each query run, Privora returns:

Platform responses

The full text of what ChatGPT, Claude, and Perplexity actually said in response to your query. You can read each response directly in the platform.

Brand mention detection

Did your brand appear? Privora checks the response text for all your brand name variations and flags whether you were mentioned.

AEO visibility score (0–100)

Your score for each query run is calculated based on:

Mentioned or not — the base signal (50 points)
Position of first mention — appearing in the first 20% of the response adds up to 25 points (early mentions = stronger recommendation)
Mention frequency — being referenced multiple times in a single response adds up to 25 more points

Scores are aggregated across platforms to give you an overall visibility score per query, and across queries to give you a brand-level score.

What a good score looks like

0–25: Essentially invisible. Your brand is not appearing in AI answers for these queries.
26–50: Occasional mentions. You appear in some responses but not reliably, and typically not prominently.
51–75: Moderate visibility. Appearing in most responses, sometimes early.
76–100: Strong presence. Appearing prominently and consistently across platforms.

Step 5: Read the platform responses to understand why

The raw AI responses tell you exactly what the platforms are saying and who they're recommending instead of you. Look for:

Which competitors are mentioned — these are the brands whose content strategy is working for this query
How your brand is framed — are you described accurately? Positively? With outdated information?
What claims are made about the winner — what does their content or reputation have that triggers the AI to mention them?

This diagnosis step is what separates AEO from guessing. You can see the exact gap.

Step 6: Generate content suggestions

On the Optimize page, select a completed query run and click Generate Suggestions. Privora sends the full context — your brand profile, your competitive landscape, the AI responses, and your current score — to GPT-4o and asks it to prescribe specific content improvements.

The suggestions you receive depend on your plan:

Free tier

Three cross-platform suggestions, ranked by estimated visibility impact. You'll see the content type, title, and implementation guide.

Starter and Agency plans

Up to 12 platform-specific suggestions per run — three each for ChatGPT, Claude, Perplexity, and cross-platform. Each suggestion includes:

Diagnosis: Why you're losing visibility for this query on this platform
What to create: The exact content type (FAQ section, how-to guide, comparison table, JSON-LD schema, citation strategy, etc.)
Concrete example: A draft of the actual content tailored to your brand and audience — not a template
Estimated score improvement: A projected visibility score if you implement the suggestion

The seven content types Privora recommends:

Type	What it is	Best for
FAQ	Q&A sections answering exact queries	ChatGPT, Claude
How-to	Step-by-step guides for your use case	All platforms
Comparison table	Side-by-side feature or pricing tables	ChatGPT, Claude
Definition	Clear, authoritative definitions of industry terms	Claude
Schema.org markup	Structured data added to your web pages	All platforms
JSON-LD	Machine-readable entity definitions	All platforms
Citation strategy	Getting cited by authoritative sources	Perplexity

Step 7: Implement and re-measure

Once you've published the suggested content, run the same query again. Privora tracks score changes over time so you can see exactly which content improvements moved the needle.

This is the AEO improvement loop:

Query → Score → Diagnose → Publish → Re-measure → Repeat

Most brands see measurable improvement within 2–4 weeks of implementing content suggestions, though Perplexity typically responds faster (because it pulls live web results) while ChatGPT and Claude may take longer as their training data updates.

The Run History page

Every query run is stored in Run History. You can filter by brand, query, platform, and date range to see how your visibility has changed over time. This is your AEO analytics dashboard — the longitudinal view of whether your strategy is working.

The Integrations page

If you run an e-commerce store, connect your Shopify, WooCommerce, or Etsy account to import products directly into Privora. Product data is used to generate more specific, product-level suggestions and queries (e.g., "best leather wallets under $50" for a specific SKU rather than just your brand).

API keys for OpenAI, Anthropic, and Perplexity can also be configured here. If you bring your own keys, queries run against your own API quota.

Frequently asked questions about Privora

How often should I run queries? Weekly monitoring is the default for most queries. For fast-moving competitive categories, daily monitoring is available on Agency plans.

How accurate are the visibility scores? Scores reflect real API responses from each platform at the time the query was run. They are as accurate as the AI platforms' responses themselves, which can vary with model updates.

Why does my score differ across platforms? Each AI platform has different training data, algorithms, and recommendation patterns. A brand strong on Perplexity (web citations) may be weaker on Claude (documentation quality) and vice versa. This is by design — the per-platform breakdown tells you where to focus.

Can I track multiple brands? Yes. Starter plans support up to 1 brand; Agency plans support up to 5 brands in a single workspace.

Is Privora free to start? Yes. The free tier gives you 1 brand, up to 10 queries, and 25 query runs per month — enough to get a clear picture of your current AI visibility before committing to a paid plan.

Start tracking your AI visibility

The brands winning in AI search aren't lucky — they're publishing the right content in the right format for the way AI models process information. Privora tells you exactly what that content is.

Create your free account →