Ideas
The backlog of queries answered without you, each with the evidence behind it and a score for how winnable the spot looks.
On this page
Ideas is the first tab on Content. Every row is a query where AI engines answer without you, carrying the signal that found it, the brands being named instead, and a score for how much room there is.
Nothing here comes from a keyword tool. Each row traces back to something Trakkr observed in your own data, or is marked as inferred when it does not.
Where ideas come from
The Signal column names what found the row, and the filter menu groups them the same way. A filled dot means the signal was observed in your data. A hollow ring means it was inferred, with no observation behind it yet.
| Signal | What it means |
|---|---|
| Citation gaps | Queries where competitors are cited before your brand. |
| Prompt weakness | Tracked prompts where your position is missing or low. |
| Query gaps | Search queries AI engines ran while answering without you. |
| Reddit demand | High-signal threads drawing competitor attention. |
| Crawler opportunities | Hot or orphaned pages on your site worth refreshing. |
| AI referral expansion | Pages already getting AI traffic and worth expanding. |
| Recommendations | Inferred ideas without observed evidence yet. |
| Custom entries | Titles you added yourself in the composer. |
Reddit and crawler signals only appear when those sources are connected for the brand. Without them, the remaining signals still work, and the score adjusts rather than penalizing you for the missing data.
Reading the table
Five columns, plus the row action on the right.
| Column | What it shows |
|---|---|
| Idea | The query, prompt or topic the row targets. |
| Signal | What found it and where the signal was observed. |
| Answered by | Brands an AI engine names when it answers this question without you. |
| Potential | High, Medium or Low, on one scale across every row. |
| AI vol | Estimated monthly AI chat volume for the query. |
Rows group by campaign by default. Right-click a column header to regroup by signal or turn grouping off, and sort on Potential, AI volume, citations or recency. The header Export button writes whatever is on screen to CSV.
How Potential is scored
Potential is a band over one 0 to 100 score. Sixty and above reads High, thirty and above reads Medium, everything else reads Low. Hovering the cell opens the working.
| Component | Weight | What it measures |
|---|---|---|
| Citation coverage gap | 30% | How many engines cite a competitor for this query. |
| Query demand | 20% | Citation frequency over 30 days plus estimated AI search volume. |
| Competitor density | 20% | How many distinct rivals are named. |
| Ease | 15% | How extractable the recommended format is. FAQ and comparison shapes score highest. |
| Reddit signal | 10% | Thread demand, when Reddit is connected. |
| Crawler hot | 5% | Recent crawler hits on the matching page, when crawler tracking is on. |
When a component has no data, it drops out and the remaining weights are renormalized, so a brand without Reddit or crawler tracking is not capped below the rest. The breakdown labels which components were degraded.
Some rows are flagged as quick wins. That flag is deliberately rare: it means the brand almost certainly already has standing on the query, usually a direct comparison against a named competitor or its own core category, so a focused page may be worth investigating. It does not predict a win.
Opening an idea
Clicking a row opens the detail view, which is where the evidence lives:
| Section | What it holds |
|---|---|
| Recommended | The template and length Trakkr would use. |
| Potential breakdown | The six components and their contribution. |
| Evidence | Signal, engines, your position, AI volume, citations, competitors, top source domains. |
| Why this | The reasoning behind the row in plain language. |
| Source chain | Every signal and enrichment step that produced it, with timestamps. |
| The angle | What would make this piece the one an engine cites. |
| Write these sections | The buyer sub-questions the page should answer. |
Acting on a row
| Action | What it does |
|---|---|
| Write article | Opens the composer with the query, template and evidence already filled in. |
| Promote to action | Moves the idea into Actions as work to schedule. |
| Assign | Promotes it and hands it to a teammate. Solo accounts skip the picker. |
| Open in Citations, Prompts, Reddit or Crawler | Jumps to the surface the evidence came from, filtered to this query. |
| Dismiss | Hides the row. You are asked why: already covered, not our category, not this quarter, or the evidence looks wrong. |
Dismissals are not permanent. Toggle the dismissed view to see them and restore anything you hid. Selecting several rows at once lets you write them as a batch or move them into a new campaign.
When the list refreshes
Ideation runs on its own after every completed report, and again on a daily pass. Refresh ideas in the command palette forces a run, with a 30-minute cooldown between manual runs. The verdict line above the table says how recently the list changed.
A brand needs at least one active prompt and one completed report before ideation has anything to work from. New brands see an empty tab until their first report lands.
Common questions
Why is an idea here when we already have a page on the topic?
Because an engine is not citing that page. Coverage and citation are different things, and a page can exist, rank in classic search, and still never be quoted. Check the page in Pages to see which of its checks is failing before writing a second one.
What does "Answered by" mean exactly?
It lists the brands an AI engine names when it answers the question without you, taken from the citation evidence. It does not claim which engine named which rival for that specific question, because no per-prompt provider data exists. Where a rank is shown, the query is verbatim one of your tracked prompts.
Should I work top down through Potential?
Mostly. The score is deliberately conservative, so the top of the list is where headroom and winnability overlap. Break the order when a lower row sits closer to your actual expertise, because authority mismatch is something engines eventually detect.
Can I add my own idea?
Yes. Write article in the header takes any title you type and creates the row as a custom entry. It carries no observed evidence, so it scores from the template shape alone.
Do dismissed ideas come back?
Not on their own. Dismissal is stored per query, so the same gap will not resurface in the main list. Open the dismissed view to restore one.
Why did my idea count drop without me writing anything?
Ideas can change as the available evidence changes. A query where you started getting cited stops being a gap, and duplicate rows from different signals get merged into one. Both reduce the count without any work on your side.