Method
How we judge.
We're asking you to trust a verdict about your brand, so here is exactly how it's reached. This page is generated from the scanner's own settings: if we change a question or a weight, it changes here too.
What we ask
We ask ChatGPT, Perplexity and Gemini the questions buyers ask, with web search switched on so they answer the way they would for a real buyer. Every scan shows the exact questions and when they ran.
- “Is [your brand] worth it?”
- “Any problems with [your brand]? What do real customers say?”
- “What are the best alternatives to [your brand]?”
Answers change from run to run. A scan is a snapshot of one moment, and we say so on every result.
What we search
Assistants repeat what ranks, so we check what ranks. These searches run on Google UK, top ten results each:
When Google shows an AI Overview on any of these searches, we read it like an assistant's answer: the same word-for-word rule for claims, and the pages it cites count as quoted.
We also pull monthly Google search volume for 6 related searches. That number feeds the cost estimate, and nothing else.
How a claim gets flagged
A judging model reads every answer and pulls out each sentence that says something about your quality, service, price or trustworthiness, and whether a buyer would read it for or against you. Then two rules apply that we don't bend:
- Word for word or not at all. A flagged claim must appear in the assistant's answer exactly. Paraphrases are thrown out before you see them.
- When unsure, neutral. The judge is told to be conservative. We'd rather miss a borderline claim than invent one.
How a page earns a verdict
It reads against you, and an assistant is quoting it or it ranks in the top 3 on Google.
It reads against you but nobody is repeating it yet, or it's neutral but quoted from Reddit, a forum or a review site.
Everything else. Most mentions land here, and replying to them would do more harm than good.
Pages are then ranked by how much damage they can do: where they come from, how high they rank, how negative they are, and whether an assistant quotes them. Being quoted multiplies the weight by 1.5.
How we put a number on it
Every scan ends with one range: the monthly demand that meets a bad page or answer before it reaches your site.
searches × (Google click share + AI answer share) × share who walk away × conversion × average order
- Google click share
- 28% at position 1, 15% at 2, 10% at 3, falling to 1.8% at 10
- AI answer share
- 20% of branded searches, scaled by how many assistants said something against you
- Share who walk away
- 5% to 20%, which is why it's a range
- Conversion and order value
- 6% and €100 unless you tell us yours
Plainly: these are our working assumptions, not measured results. They'll be replaced with figures from real client outcomes as we get them, and this page will say when. If search volume isn't available, we show no number rather than one built on a guess.
What a scan can't do
- It's a snapshot. Assistants answer differently from one run to the next.
- It sees what Google ranks and what assistants cite. Posts inside TikTok, Instagram, Facebook groups, Discord or behind a login only show up if they rank.
- Gemini shares which sites it used, not the exact pages.
- If any source didn't answer, the result lists it, and an incomplete scan is never shown as quiet.
- It's automated. The Full Verdict is where a person checks every source.
What we never do
- Post anything for you. There is no posting code in VerdictDeck, and we hold no posting permissions.
- Create accounts, write reviews, seed threads or bury complaints. We turn that work down.
- Draft from facts you haven't approved.
- Sell or share your details. See the privacy page for every service that touches them.
See it on your brand.
Free, and nothing is posted.