Contested reads what ChatGPT, Gemini, Perplexity, Claude and Google say about your brand, then turns it into tickets your team can work this week. Send them to Trello, Asana or monday.com.
Harbourlight leads. See how every competitor compares, answer by answer.
Every audit ends in a short list of case files: what an AI engine said, where it probably came from, and the fix to make. Your team works the list, then the next audit checks what changed.
Each case file quotes the engine word for word, points to the page it most likely came from, and gives you the new wording to publish. Page wording, an FAQ entry with structured data, or a note to a page's editor.
“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.
Then the fix answers it honestly instead of arguing with it.
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.
Every ticket says how strong its evidence is: strong, moderate or speculative. We never promise an uplift.
Connect Trello, Asana or monday.com, choose the board or list, and send. Each card carries the engine's quote, the steps, why it works, what done looks like and the likely source page. Sending twice never creates duplicates. For Jira, export as CSV, Markdown or a pre-filled link.
Gemini: “Vantagely offers custom domains only on its top plan.”
Why it works: 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.
Where it probably came from: https://agencystack-weekly.com/best-client-reporting-tools
Sent from Contested.
The full fix wording stays in your Contested ticket and in the CSV and Markdown export.
Jira: export as CSV, Markdown or a pre-filled link.
Sending to these tools is on the paid plans.
We show Reddit and LinkedIn pages only when an engine cited them and we confirmed they mention your brand. Each card gives the page, the citing engine and the source excerpt. No general search results or inferred objections.
No Reddit or LinkedIn citation in this illustrative sample passed the page check. General search results are not used as brand evidence.
Your biggest competitor, where you win and lose by kind of buyer question, head to head with honest calls, and the pages AI cites when they win.
Metricnest
EvenThe gap moved 13 points in your favour since the previous audit.
You lead on Perplexity and Claude. Metricnest leads on ChatGPT and Gemini.
| Kind of question | Vantagely | Best competitor | Verdict |
|---|---|---|---|
| Beginner Fit | 3 of 5 | Metricnest5 of 5 | Behind |
| Purchase Decision | 3 of 5 | Metricnest4 of 5 | Close |
| Features & Reliability | 3 of 5 | Metricnest4 of 5 | Close |
| Pricing & Value | 5 of 5 | Metricnest5 of 5 | Level |
| Setup & Integration | 3 of 5 | Dashparrot3 of 5 | Level |
| Category Discovery | 5 of 5 | Dashparrot4 of 5 | Ahead |
| Community Consensus | 5 of 5 | Metricnest4 of 5 | Ahead |
| Instead of a Competitor | 5 of 5 | Dashparrot4 of 5 | Ahead |
Each row rests on a handful of answers, so one answer of difference reads “Close”, not “Behind”.
Metricnest
Too close to callYou 23 · Them 17
You came first in 23 of the 40 answers that named either of you.
Dashparrot
Too close to callYou 23 · Them 16
You came first in 23 of the 39 answers that named either of you.
Clientorbit
Too close to callYou 18 · Them 15
You came first in 18 of the 33 answers that named either of you.
reddit.com · Community · 2 pages
youtube.com · Reference · 2 pages
agencystack-weekly.com · Comparison article · 1 page
dashparrot.com · Competitor's own site · 1 page
Your site, what you sell, your competitors. Once.
The questions your buyers ask, reviewed by you first, put to each engine.
How often each engine recommends you, and how you do head-to-head against each competitor.
Case-file tickets your team can start on, sent to Trello, Asana or monday.com, and an 8-page PDF brief. After each audit we check what you finished.
Every brand the engines recommended, ranked by how often, with a plain label for how each one is described. Your brand is always highlighted.
Each engine answers differently. You see how often every one recommends you, and how many answers that rests on.
See which sites sit behind the answers you lose, and whether you are listed on them.
From the cover score to the action plan, in eight pages.
You always see whether a figure comes from 40 answers or from 8, so you know how far to trust it.
When there are only a few answers we say so, and show counts like "3 of 3" instead of a bold 100%.
When two brands are too close to call, we say so instead of naming a winner.
After each audit we check the tickets you finished and say Happened, Not yet or Can't tell. We never claim a ticket moved your score, and we never promise a result.
These are illustrative reports for fictional brands: invented data, built the same way as a real audit. Open one and click around: every tab, every ticket.
How often each AI engine recommends your brand when buyers ask category questions, who the engines pick instead of you, and what to do about it.
A ticket that starts from something an engine actually said about you. It quotes the engine word for word, points to the page that probably fed it, gives you the new wording to publish, and says when you are done. Each ticket is labelled strong, moderate or speculative evidence, and we never promise an uplift.
Many tools stop at the score. Contested continues: the quote, the page it probably came from, the fix, a ticket your team can start on, and a check after your next audit.
No. After each audit we check one narrow thing for every ticket you finished, such as whether the page is live, names you, is now cited, or whether the worry is gone, and we tell you Happened, Not yet or Can't tell. A score moves for many reasons, so we do not claim a ticket caused it.
The Community tab shows Reddit and LinkedIn pages cited by an AI engine only when we can confirm the page mentions your brand. It shows the citation excerpt and which engine cited it. We don't search for general community discussions or infer objections and sentiment from them.
No. Contested does not post to Reddit or LinkedIn for you. The Community tab shows only Reddit and LinkedIn pages cited by an AI engine and checked to mention your brand.
No. Contested reads what the engines actually answer when buyers ask, and it does not see Google's internal ranking or AI data (Google says no third-party tool can). For real impressions, use the Generative AI performance report in Google Search Console alongside Contested.
ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews.
Send tickets to Trello, Asana or monday.com from your Action Plan: connect once, choose a board or list, and send. Each card carries the engine's quote, the steps, why it works, what done looks like and the likely source page, and sending twice never creates duplicates. For Jira, export as CSV, Markdown or a pre-filled link. Sending to these tools is on the paid plans.
Brands, marketing and growth teams, and agencies that need to know how AI engines recommend them, and what to do to be recommended more.
A few minutes. There are five short steps, and most of it is filled in from your website. Create an account, set up your brand, pick a plan, and your first audit runs in a few minutes.
Yes. You describe your market in your own words, name your competitors with their websites, and review, edit, add or switch off the questions before anything runs. An audit scores about 12 buyer questions plus 4 about your brand, and you can add your own up to 18 in total.
A visibility score, how you compare with each competitor, the sources the engines relied on, confirmed Reddit and LinkedIn citations that mention your brand, an action plan of case-file tickets, and an 8-page PDF brief.
Usually a few minutes. You can watch its progress, and we email you when the report is ready.
It tells you how many answers a number rests on: 20 or more is high confidence, 11 to 19 is fair, and 10 or fewer is an early read. The more answers behind a number, the less likely it is a fluke.
Only your workspace. We don't publish or share your brand's audit.
More answers on the FAQ page.