Your homepage publishes a typed contact channel AI agents can route customers to
Have you labeled your phone or email so AI knows whether it is sales, support, or billing?
What this signal tests
We check that your homepage's Organization data includes at least one ContactPoint with a clear type label (customer support, sales, billing, etc.) and a working telephone number or email. Optional but valuable extras include the area served, available languages, and hours of availability. This structure lets AI agents pick the right channel when routing a user.
Why it matters for your visibility in AI
AI shopping assistants and customer-service agents increasingly handle the first step of customer queries before a human gets involved. When a user asks "how do I reach Acme support in Spanish on Saturday?", the assistant needs to know which of your contact numbers handles support, which speaks Spanish, and which hours apply. Without a structured ContactPoint, the assistant has to guess from your page text, which often goes wrong: support calls land on a sales line, English-only queries are routed to a Spanish-only team, or the user is told you are open when you are closed. A well-specified ContactPoint with contactType, telephone, areaServed, availableLanguage, and hoursAvailable turns these mistakes into reliable routing and reduces friction for both you and your customers.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| At least one fully-specified ContactPoint. |
How we test it
We parse the Organization JSON-LD on your homepage and look for a contactPoint property. The value must be one or more objects with @type set to ContactPoint, a contactType string (such as 'customer support' or 'sales'), and either a telephone or email field. We check that the telephone is in E.164 international format and the email is syntactically valid. If no ContactPoint exists, or it lacks a type, the signal fails.
Show technical detection method
Parse Organization.contactPoint; require @type ContactPoint with contactType and (telephone OR email).
If your site fails: how to fix it
- Identify the contact channels you actually want AI to route users to: typically customer support, sales, and billing as separate contactType values.
- For each channel, add a ContactPoint object inside the contactPoint array of your Organization JSON-LD with @type 'ContactPoint', a contactType string, and a telephone in E.164 format (+1-555-555-1234) or an email address.
- Add areaServed (an ISO country code or a region name) so agents know which geography each line covers.
- Add availableLanguage (an array of BCP 47 language tags like 'en' or 'es') so multilingual users get routed correctly.
- Add hoursAvailable using the OpeningHoursSpecification format if your hours vary by day.
- Validate the result with Schema.org Validator and re-run the scan.
Quick facts
| Maturity | ESTABLISHED |
|---|---|
| Weight | medium |
| Category | Entity |
Primary sources
Related signals
Frequently asked questions
What is a 'contactType' and what values should I use?
contactType is a free-text label that names the purpose of the channel. Common values include 'customer support', 'sales', 'billing', 'technical support', 'press relations', and 'emergency'. Use lowercase, descriptive labels that match how a user would describe the request.
Do I need both a phone and an email?
No, at least one is required per ContactPoint. Many businesses provide a phone for sales and an email for support, or vice versa. You can also include multiple ContactPoint objects, each with a different mix of channels.
Should I expose a personal mobile number?
Generally no. Use a routed business line or a shared inbox. The ContactPoint is meant to handle external traffic that AI agents may surface, and you want the volume to land on a channel built for it.
What if my hours are complex (e.g., closed midday)?
Use multiple OpeningHoursSpecification objects in the hoursAvailable array. Each can cover a range of days and a start and end time. AI systems are getting better at parsing these and will surface accurate hours when a user asks.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.