A case-file ticket is a task that starts from one thing an AI engine said about your brand, quoted word for word, and ends with the fix wording to publish and a line that says when you are done.
A score tells you that an engine does not name you, or names you with a doubt. A case file tells you what to do about one specific answer. It is built from the audit's own evidence, so it reads differently for every brand.
A case file holds these parts:
This is an illustrative sample from a fictional brand called Vantagely. It is not a real customer. The result strip at the bottom shows the check that follows once the ticket is marked done.
“Vantagely offers custom domains only on its top plan.”
Asked: “Which client reporting tools let agencies use their own domain?”
The roundup Gemini cites repeats an older line that custom domains need the top plan. Your own site never says which plans include them, so the engine passes the page's version on.
agencystack-weekly.com
Review guide · cited by Gemini for this answer
“Vantagely: custom domains are limited to the top plan, according to the vendor's older docs.”
Engines combine several pages, so this is the most likely source, not proof.
Now: no page today
Custom domains are included on every plan. Add yours in Settings, point a CNAME at reports.vantagely.io, and your clients see your name on every report.
Engines quote the clearest statement they can retrieve. A plain page that answers the exact question gives them something current to quote instead of the older line.
Done when: An engine cites /white-label when asked about custom domains.
You should see a change in: 4–12 weeks (typical, not guaranteed)
A check shows what happened. It does not prove the ticket caused it.
The ticket writer is asked for 5 or 6 case files, and for fewer when the evidence cannot support them honestly. Each case file rests on a different thing an engine said, and no more than two rest on the same page.
Case files lead the plan. They are followed by the standard kinds of work: digital PR, counter-positioning, authority listings, community response and technical hygiene. The plan shows the first 5; the rest sit behind one link.
The level belongs to the kind of work, not to your brand. It says how well that kind of work is known to move answers from engines that search the web. It is not a forecast.
A written ticket is checked by code before you see it, and it is thrown out if it fails. A ticket is dropped when:
Sometimes the engine says something true, such as a feature you do not have yet. Contested marks the case file as one where the engine is right, and the fix answers it honestly: it states what exists, what does not, and what to say until then. It does not argue with the engine.
A second sample from Vantagely shows that case, and a third shows a disclosed reply that a person on the team reads, edits and posts.
The engine was rightSo the fix answers it honestly.
Perplexity“Vantagely does not connect to TikTok Ads.”
The fix: New wording for /integrations. TikTok Ads: not available yet, planned for next quarter.
When you mark a ticket done, Contested keeps a copy of it. After each later audit it runs one yes-or-no check for that kind of work and shows the result on the ticket: Happened, Not yet or Can't tell.
A check shows what happened. It does not prove the ticket caused a change in the score or in an answer. Engines vary from one run to the next, and many things change at once.
When a check cannot be made, for example a site blocks automated visits, the ticket named no page, or the audit had answers missing, the result is Can't tell. A Can't tell is never turned into a No. Not yet shows how long that kind of work usually takes.
Each brand has one running list. An audit adds only tickets that are new. A tick belongs to the brand, not to one audit, so rewording or renumbering a ticket does not undo it.
Every ticket can be exported as CSV, Markdown or a pre-filled Jira link. On paid plans you can also send tickets to Trello, Asana or monday.com as cards, once you have connected the tool. The paste-ready fix wording stays in the Contested ticket and in the CSV and Markdown export; it is not in the card.
A case-file ticket is a task that starts from one thing an AI engine said about your brand, quoted word for word. It names the page that probably fed the answer, gives the fix wording to publish, explains why it changes future answers, lists the steps and says when you are done.
No. After an audit Contested checks the tickets you finished and reports Happened, Not yet or Can't tell. That shows what happened, not what caused it.
The ticket writer is asked for 5 or 6 case files, fewer when the evidence cannot support them. The plan shows the first 5, and the rest sit behind one link.
It drafts the fix wording. Your team reads, checks and edits it, and publishes it. Contested does not post for you, and a reply to a community thread is always posted by a person on your team.
The level describes how well that kind of work is known to move answers. Structured data and llms.txt are rated speculative because no major engine has confirmed that it reads llms.txt, and structured data helps Google only indirectly.