Why you lost. Who AI named instead of you.
For each question AI left you out of, see who was named instead and a possible reason. The reason is a model's reading, and the screen says so.
Questions lost
18
A page could win back
11
Nothing a page can change
7
A comparison page that names them
4 questions| Question | Question type | Won by | Assistants lost |
|---|---|---|---|
| Which social media scheduler posts at the best time automatically? | Non-branded | Kestrel | 4 of 4 |
| Which social media scheduler is easiest for freelancers? | Non-branded | Northline | 3 of 4 |
| Postprism vs Northline for small agencies | Non-branded | NorthlinePublished | 2 of 4 |
Show 1 more
A review roundup the assistant trusts
3 questions| What is the best tool for scheduling social media posts? | Non-branded | Northline | 4 of 4 |
| What social media tool works for a small agency? | Non-branded | Nobody | 3 of 4 |
Show 1 more
A directory you cannot join
2 questions| Which social media schedulers are listed for small businesses? | Non-branded | HaldenNo page can win this | 2 of 4 |
Show 1 more
- Assistants that did not name you
- 4 of 4
- Won by
- Kestrel
- A page could win it
- Yes
- How it was won (our AI's judgement)
- A comparison page that names them
No page written yet
Write the pageIllustrative.
What a loss is
A loss is a question that no assistant named you on. If one assistant named you and another did not, it is not a loss.
Each loss gets one of four reasons, or "Not sorted yet" when the reason is not clear:
A loss also says whether a page of yours could change it. A directory you cannot join is marked as not winnable, because no page of yours can fix it.
A comparison page that names them.
Pricing an assistant can read.
A review roundup the assistant trusts.
A directory you cannot join.
How a loss is judged
Each loss is read from the stored answers, after the scan has finished. Nothing is asked again to judge it.
- Agentcites finds every question that no assistant named you on.
- A model reads the answers to that question, with the pages each answer linked to.
- It names the brand that won, where the answers point to one. Where they do not, no single winner is recorded.
- It picks one of the four reasons, or leaves the loss as "Not sorted yet".
- It marks the loss as winnable with a page of yours, or not.
The table below shows what three rows could hold.
| Question | Winner | Reason | Winnable |
|---|---|---|---|
| Which accounting app is best for freelancers? | A rival app | A review roundup the assistant trusts | Yes |
| Is App A cheaper than App B? | App B | Pricing an assistant can read | Yes |
| Where can I find vetted accountants near me? | A directory | A directory you cannot join | No |
What you see on the screen
The Lost questions screen, on its Who won them tab, groups your losses by reason, so the shape shows before the detail. Each row gives the question, whether it names you, who won it, and how many assistants left you out of all the assistants that answered.
Inside each reason, the rows are sorted by how easy the model thinks they are to win. That sense of ease orders the list, but it is never printed as a score.
Open a row and a panel opens beside the list. It holds the reason, a way through to the answers the loss was judged from, and the page drafted for it if there is one.
The Overview adds the view from above: how often each brand was named, topic by topic, on the questions that do not name you. A dark cell is a topic where that brand is named often.
The reason is a model’s judgement
The reason for a loss is decided by a model reading the answer and the list of pages it linked to. It is a judgement, not a measurement, and the screen says so beside every reason.
The answers it read are always one click away, so you can disagree with the reason by reading what the assistant actually wrote.
What it does not do
It does not check that the winner it names is a real company.
It does not give a score for how easy a loss is to win. It says winnable or not, and the reason.
A loss that a page cannot win is still shown, labelled not winnable, so you can see what you are up against. It is kept out of your to-do list, and no draft is written for it.
Answers from the assistants your buyers use
Each question goes to each assistant you choose, with web search on, the way a buyer would ask it.
- ChatGPT
- Perplexity
- Gemini
- DeepSeek
Questions
How is this different from a share of voice figure?
A share tells you that you lost. A loss tells you which question, who won it and how, which is what you need to decide what to write.
What happens after a loss is found?
For a loss a page could win, Agentcites can [draft that page](/features/drafts).
Can I see why the model chose a reason?
Yes. Every loss links to the answers it was judged from, with their links.
Read more about getting named in AI answers
See it for your own site
We ask 20 of your questions to four AI assistants, and you see every answer before you pay anything. You do not need a card.