Your musician website and AI: how ChatGPT and Perplexity read you
More and more people ask an assistant instead of searching. What they read from your site, what confuses them, and how to leave them clear data.
By Joan Subirats, founder of The Jazz Passport7 min read
In this article
A venue programmer is planning the season and wants to know more about a trio someone recommended. A journalist needs to check which city you play in on Friday. A listener wonders whether there is a new record. More and more of them skip the search and the five open tabs and type the question to an assistant instead: ChatGPT, Claude, Perplexity, or the AI answers Google now puts above its results.
The answer is built from what is published about you, and most of that is out of your hands. Your site is the one source you fully control.
You cannot decide what an assistant will say. You can make the right facts easy to read and hard to mix up.
How an assistant reaches your site
There are two broad routes. Some assistants answer from what they learned in training, which is a snapshot of the web taken a while ago. Others look things up when the question arrives. They run a search, open the pages it returns and summarise what they read. In that case your site gets a visit from a crawler, a program that downloads the page much as a search engine's does.
What the crawler takes away is mostly text. It doesn't see your design or hear your music, and it can't reliably read words set inside a picture. It reads words, links and a few hidden facts written for machines.
What they read: text, not posters
The most common mistake on musicians' sites is keeping the important information inside images. The tour poster holds all the dates, the bio is a scanned PDF, and the band's name appears only in the logo. A person reads all of it without a second thought. For a program, that information barely exists.
What should be written as text on your site
- Your name, or the band's, exactly as it appears on posters
- What you play and what kind of music, in one sentence
- Where you are from or where you work from
- Who is in the band, with their instrument
- Every date: day, city, venue and a ticket link
- Your records, with year and label if there is one
- How to book you
Write in whole sentences. “Bassist and composer, based in Valencia, leading her quartet since 2019” makes sense on its own to anyone who reads it. A loose string of keywords helps nobody, person or machine.
What confuses them
When an assistant reads several sources and finds different facts, it has to pick one, and sometimes it picks wrong. The confusion nearly always starts with something you can fix.
What clears things up
- The same name everywhere, spelled the same way
- One official site, linked from your profiles
- Past dates kept apart from upcoming ones
- Instrument, city and style next to your name
- A short bio that says the essentials in two lines
What confuses
- Ruiz Trio in one place and The Ruiz Trio in another
- An old site still online with a different bio
- A tour from two years ago listed as upcoming
- A name another artist shares, with nothing to tell you apart
- A home page that only says welcome to my website
If another musician shares your name, you don't need to change it. Pair it every time with what sets you apart: instrument, city, style. Your bios work the same way. If you have a short one, a long one and one for the press, they should say the same thing at three lengths. The bio guide explains how to write them.
Structured data: your card for machines
Besides the text visitors read, a page can carry a small block of data they never see. It's called structured data. It uses a shared vocabulary, schema.org, that search engines understand, and so does any other program that cares to read it. The usual format is JSON-LD, a list of fields and their values.
A musician has two things worth saying this way. The first is who you are: a person (Person) or a group (MusicGroup), with name, site, photo, description and links to your profiles. The second is when you play: each show as a music event (MusicEvent), with date, venue, city and tickets. Trimmed down, a band with one date looks like this:
{
"@context": "https://schema.org",
"@type": "MusicGroup",
"name": "Your Trio",
"url": "https://yourtrio.example",
"description": "Jazz trio from Valencia: piano, double bass and drums.",
"sameAs": [
"https://open.spotify.com/artist/...",
"https://yourtrio.bandcamp.com"
],
"event": [
{
"@type": "MusicEvent",
"name": "Your Trio live",
"startDate": "2026-11-14T21:00",
"location": {
"@type": "Place",
"name": "Jazz club",
"address": {
"@type": "PostalAddress",
"addressLocality": "Valencia",
"addressCountry": "Spain"
}
}
}
]
}
Look closely at sameAs. It lists your other profiles and tells whoever reads it that this Spotify, this Bandcamp and this site belong to the same artist. Nothing protects you more directly from being mixed up with someone else.
Structured data won't guarantee that anyone cites you, and it doesn't change what your site says. It removes doubt: the date is a date, the city is a city, and the name is yours. To see how Google uses it for shows, read concerts in Google search.
Consistency beyond your site
Your site is one source among several. An assistant may also read your Spotify profile, the venue's page, an interview or your Instagram. When they all agree, there is little room for error. When they don't, the answer is anyone's guess.
Once a year, go through the places where you appear. You want the same name, the same recent photo, the same short bio and a link to your site on every one. If a venue still has a page about you with old details, write to them. They are usually glad to update it.
Let them read you
Every site has a file called robots.txt that tells crawlers what they may read. Some hosting services offer a switch that blocks AI crawlers, and some sites have it turned on without their owners knowing. The choice is yours, but know the cost before you make it. If you block them, the assistants that search when the question arrives can't read your site, and they'll answer with whatever they find elsewhere.
A test you can run today
- Ask two or three assistants who you are, what you play and when your next show is.
- Write down every fact that is wrong or out of date.
- Work out where each one might come from: your site, an old profile, a venue's page.
- Fix it at the source, starting with your own site.
- Check that your dates and your name are written as text, and not only in images.
- Run the test again in a few weeks, because changes take time to show.
In Pages
Every Pages site publishes its structured data, and you don't write a line of it. Your card goes out as Person or MusicGroup, depending on whether it presents you or a band, with your links in sameAs. The dates you add to the Calendar go out as MusicEvent, with venue, city and ticket link. Anything you hide on your card, such as your town, stays out of that data too. Pages doesn't block AI crawlers either, so assistants can read your site like any other public page.
Your site, without the website work. Ask for your invitation.
Pages is the site and the Studio for musicians by The Jazz Passport. The technical side is done; you bring the music.