Getting started with Manifest Signal
Set up a workspace, choose the questions to track, and read your first visibility baseline in about ten minutes.
Updated
Manifest Signal shows how AI answer engines such as ChatGPT, Perplexity, Gemini, and Claude respond when people ask questions about your category, and whether those answers mention, recommend, or cite your organization. This guide walks through the first session.
1. Create a workspace
A workspace represents one brand or organization. During onboarding you enter the brand name, the primary domain, any alternate names people use (for example an abbreviation), and a one-sentence description. Signal uses the names to detect mentions and the domain to detect citations of your own pages.
2. Add competitors
Add up to the number of competitors your plan allows. Competitors are used to calculate share of voice and to show which organizations engines name when they do not name you. Choose organizations a buyer would realistically compare you with, not only the largest companies in your industry.
3. Choose prompts
Prompts are the questions Signal asks each engine. Onboarding suggests a starting set based on your category; edit them so they reflect how your customers actually phrase questions. A good starting set has 10 to 25 prompts spread across discovery ("best X for Y"), comparison ("X vs Y"), evaluation ("is X worth it"), and brand questions ("what is X"). See Choosing and organizing prompts.
4. Read the baseline
New workspaces start with labeled sample data so you can explore every screen immediately. Sample answers are generated deterministically from your prompts and competitors; they are not observations of real engines and are never mixed into live metrics. When live engines are enabled for your workspace, switch the data source in Settings → General to begin collecting real observations. See Answer engines and data sources.
5. Run a readiness audit
The readiness audit fetches your site the way an AI crawler would and checks crawler access in robots.txt, indexability, structured data, server-rendered content, and response time. Failing checks become tasks automatically. The same check is available without an account at /tools/ai-readiness-checker.
6. Work the task list
Signal turns observations into prioritized tasks, and every task links to the evidence that created it: a failing check, a prompt where competitors appear and you do not, a source engines rely on, or a claim reviewed as inaccurate. Assign tasks, mark progress, and watch the related metrics over the following runs.