# You just read a customer story. Here's how to know it's real. > Wait, is any of this actually real? Good — you're right to ask. Most customer stories are written by the vendor, lightly approved, and handed to you on faith. The one you just read works differently. Here's exactly how it was built, plus five questions you can put to any story yourself. ## Q1 — Authorship: Did the AI just make this up? No. Every word traces back to a recorded interview. Each claim is pinned to the exact moment in the transcript where the customer said it, so any line can be played back to its source. How: claims aren't written, they're extracted. The system scores how tightly each one matches what was actually said, from a word-for-word quote to a loose paraphrase, and keeps that grounding score on the claim. A paraphrase never gets dressed up as a direct quote. ## Q2 — Consent: Did the customer actually agree to this? Yes, and before recording started, not after a draft looked good. The interviewee verifies their email, then consents, in plain language, to being recorded and used in marketing. No one is recorded who didn't knowingly opt in. How: consent is the entry condition for the interview, and the agreement is stored with the date and time it was given. That's a dated record, not a secondhand "they were fine with it." ## Q3 — Revocation: What if they change their mind? It comes down, and everything built on it comes down too. Pull a claim, or mark it unverified, and every featured quote and headline metric derived from it is suspended automatically. How: each published asset stays linked to the exact claims it was built from. Withdrawing one claim travels those links and pulls down whatever rests on it. No one has to remember every page a quote landed on. Most systems can show you how opting in works. The harder thing to build, and the more honest thing to publish, is the way back out. ## Q4 — Permanence: Could someone quietly rewrite the record? No. Every verification, edit, and deletion is logged with who did it and when, and the log outlives the claim itself. Deletions aren't even instant: a removed claim is recoverable for 30 days before it's truly gone. How: the history is append-only. Entries are added, never overwritten, and they survive a permanent deletion of the thing they describe. The fact that something existed, and who changed it, can't disappear. ## Q5 — Identity: Was the person even real? Yes. Before a word was recorded, the interviewee proved the email was theirs with a one-time code sent to that address. No Shine story rests on an anonymous testimonial. How: the code is time-limited and rate-limited against guessing, so a quote traces back to a verified email, not a name someone typed into a form. ## Loose ends **Can I check the source behind a specific claim?** Yes — with the customer in the loop. Every published claim is pinned to the moment in the interview where it was said, so the source is on record. But that recording belongs to the customer, not to us to hand out. If you think a claim doesn't hold up, tell us at hello@shine.studio and we'll confirm it together with the customer, with their consent — never by handing their interview to whoever asks. **Does Shine hold its own stories to this?** Yes. The same five questions run on our customer stories before they run on anyone else's. "Verified by Shine" means the same thing on our site as it does on a customer's — there's no softer standard for our own marketing. Full statement: