When a model answers a question about your business, it works with whatever it can find — an outdated price list, a moved address, hours from two winters ago. The Verified Business Card is where you write those facts down yourself and put a date on each one. Services, prices, location, hours, licenses and documents: you enter them, you verify them, and only what you currently stand behind is published.
Five steps, and none of them ask you to trust a number you did not produce. You fill the card in, you date it, and the product enforces the dates.
The card is organised into ten field groups: services, price ranges, licenses, experts, social links, YouTube, ratings, contact, working hours and locations. Every group has a strict shape — a price range wants a service name, a minimum, a maximum and a currency; a location wants an address, a city and a country. Anything that does not fit the shape is rejected before it is saved, so the card cannot quietly fill up with half-entered facts.
Filling a group in is not the same as vouching for it. Each group has its own Verify action, and the timestamp is written by the server — a timestamp sent from the browser is always ignored, so a card cannot refresh itself. An empty group cannot be verified at all: you have to enter the data before you can stand behind it.
A verification is fresh under 60 days, ageing between 60 and 90 days, and stale once it passes 90 days. Fresh and ageing groups go into the published output; stale groups and groups you never verified are dropped from it entirely. If you edit a group's value, its stamp falls back to unverified and the clock starts again — changed facts have to be vouched for again.
Separately from the individual fields, you can stamp a declaration that the business is still operating. While that declaration is 90 days old or newer, the published output carries a confirmed-active line and a dateModified value. Once it is older than that, it drops out of the output rather than lingering as an unbacked claim.
The card generates three artefacts from the verified groups: an llms.txt file in markdown, a schema.org JSON-LD graph, and a ready-to-paste embed snippet. You put them on your own site. Because the output is built from the same freshness rules as the badges, what you publish and what the dashboard shows never drift apart.
Each one is verified, dated and published separately from the others — a stale price list does not drag your address out of the output with it.
A single validated record, a score that tells you how much of it is actually backed, and output formats you can publish and query.
Services, price ranges, licenses, experts, social links, YouTube, ratings, contact, working hours and locations are validated against a single shared schema, so the dashboard, the API and the published output all agree on what a valid fact looks like. Every save bumps the card's version and stores a full snapshot, which means you can see what the card claimed at any earlier point instead of guessing.
The card carries a trust score between 0 and 100: the weight of the groups you filled in and verified, over the weight of every group you filled in. Contact, services and locations count for the most, because those are the facts an AI answer is most likely to get wrong; prices, licenses, experts and hours sit in the middle; social links, YouTube and ratings are supporting signals. Filling a group in without verifying it lowers the score rather than raising it. The score is recalculated when it is read, so it can never freeze at a value the badges no longer support.
Beyond the three publishable artefacts, the card is exposed through the get_business_card tool on SOV Tracker's MCP server. It returns, group by group, whether the group is filled, how fresh its verification is and whether it currently makes it into the output, along with the attestation state, the publish flag and the card's trust score — so an assistant connected to your workspace can tell you what has gone stale without you opening the dashboard.
It is a structured record of the facts about your business that you want machines to read: what you do, what it costs, where you are, when you are open, who your experts are and which licenses you hold. You enter those facts yourself and you verify them yourself, group by group. SOV Tracker does not invent or scrape them — the card's value comes from the fact that the business owner stood behind each group on a specific date.
The server stamps that group with the current time. From then on the group counts as verified while it is under 60 days old, and it stays in the published output up to 90 days. Past 90 days it is treated as stale and removed from the output. If you change the group's value, the stamp is cleared immediately and the group leaves the output until you verify it again — the product would rather publish nothing than publish something you have not confirmed.
The card produces an llms.txt file, a schema.org JSON-LD graph and an embed snippet, all built only from the groups that are currently verified and not yet stale. You publish those on your own site. What no tool can promise is that any given AI engine will fetch, read or repeat them: SOV Tracker makes your facts machine-readable, consistent and dated — it does not control what a model does with them.
The Business Card lives at /dashboard/business-card. On the Free plan the page shows an upgrade wall instead of the editor, and the card's API routes answer with an upgrade-required error, so it is available from the Starter plan upwards. Within an organisation, saving, verifying and attesting are limited to owners and admins; members and viewers get the same editor in read-only mode rather than buttons that silently fail.
Start with a free scan to see how your brand is described today, then build the card that says it properly.