Pages
See whether AI can reach, read, cite and send people to each page that matters to your brand, one URL at a time.
You have a page you care about. The useful question is not whether it has a good score. It is whether AI can open it, read the same words a person sees, match it to a real question, cite it, and send someone to it.
Pages at /pages answers that one URL at a time. It is the first step of a loop: find the page with the problem here, do the work in Actions, and read what it earned in Results.
A page record is the lasting record for one URL, whether you own it or it sits on another site. It holds the page's diagnosis, the work linked to it, the measured result of that work, and a dated history. The index helps you find the record that needs attention; the record is where you understand and act.
/pages sends you to the dashboard, this brand is not in the rollout yet, and nothing about the brand or the plan is wrong.How pages enter the list
Trakkr builds the list from evidence it already has. You should not have to paste your whole site into a tracker before it becomes useful.
Pages you own arrive through your sitemap, an Optimize scan, Search Console data, AI crawler activity, an article published through Trakkr, or Add pages when you want a URL tracked now. Crawler discovery has a floor: a URL needs at least three fetches before it joins, because bot traffic is full of probe URLs that are not real pages.
Each source points at the same record. Tracking parameters are stripped and common URL variants are joined, so one page does not split into several because a link carried a campaign tag.
External pages follow a stricter rule. Trakkr adds one when work or evidence gives it a reason to exist: an outreach source you want to be cited by, a discussion thread with reply work on it, a competitor page behind a competitor gap, or a page behind a perception or diagnosis finding. Threads and videos get their own identity, so every form of the same permalink lands on one record. The higher bar is deliberate: an AI answer can mention many URLs that will never matter again.
Owned and external pages do not show the same diagnosis. Trakkr can test the pages you own. It cannot claim to know the technical state of someone else's site, so an external record shows why the page matters, its citation evidence, linked work and history instead of a six-stage journey.
The six checks in a page journey
For a page you own, the journey reads left to right. Each stage asks a different question, and a later stage depends on the earlier ones.
| Stage | Question | What passes it | Reading |
|---|---|---|---|
| Available | Can a request reach it? | A 200 response and an indexable page. A redirect chain or canonical mismatch is Degraded; an error status or robots block is Stalled. | Observed now |
| Reached | Do AI bots fetch it? | Five or more fetches in 30 days, including search bots. Fetched but rarely or never by search bots is Degraded. Zero fetches while Available passes is Stalled. | Observed overnight |
| Understood | Does a bot get the text? | A non-rendering bot reads what a browser reads. Getting 30% to 60% of the browser text is Degraded; blocked or very thin machine output is Stalled. | Observed now |
| Relevant | Does it match what people ask? | Two or more citable queries detected on the page match prompts you track. A weak match lowers the state. | Estimated, never observed |
| Selected | Is it cited as evidence? | AI cited the page in the last 30 days. Cited historically with none in 30 days is Degraded. Readable and relevant but never cited is Stalled. | Observed from citation history |
| Visited | Do people arrive? | People reached the page from an AI answer or search. Impressions without clicks is Degraded. Cited with no arrivals is Stalled. | Observed overnight |
The header states how many of the six pass. Relevant needs care: Trakkr cannot watch a model privately decide a page is relevant, so it infers the stage from questions detected on the page and how they match your prompts. The interface marks it Estimated everywhere, and it can never be the site-wide Bottleneck.
Open a row and it explains three things: the rule that set the state, why that rule matters, and where to go next. A grey row is not a failed row. It means Trakkr does not yet have the source needed to measure that stage.
Reading the verdict
The verdict is one sentence at the top of the record. A fixed rule writes it. No language model grades the page or chooses the wording.
The rule scans from Available to Visited and names the first stage that is Degraded or Stalled. Start there, because the journey is a chain: there is little value in polishing Selected while Available is broken, and a model cannot cite a page it cannot open.
The sentence uses one of three registers:
- Working: "This page is working end to end. Protect the pattern."
- Degraded: it names the leftmost weak stage and the fact behind it, such as "This page is available but AI crawlers aren't fetching it."
- Too early: it gives the date measurement began and the date of the first reliable read, or falls back to "Too early to judge. We have not measured enough of this page yet."
A verdict stays blank when nothing has been measured yet, and the working sentence is withheld until at least three stages have been measured. "Working end to end" is a claim about breadth, and Trakkr does not make it on a third of the evidence.
Finding the right page in the index
The index is a full list, not a percentage report. The view menu names each one with its absolute count, because "203 pages" tells you the size of the work more clearly than a share of an unknown base.
| View | What it finds |
|---|---|
| All pages | Every page Trakkr has found for this brand |
| Needs fixes | Pages the latest site scan found issues on, with the issue count in place of the work column |
| Performing | Every measured stage through Selected passes |
| Not cited | AI crawlers fetch the page, but AI has never cited it |
| Not measured | The page is in your sitemap, with no crawler fetches, citations, search clicks or AI referral clicks |
Not measured names both a view here and a stage state on a record. The view lists pages with no signal of any kind; the state means one check is missing its source.
Two more views are supported but do not appear in the menu. Add ?view=ranks_not_cited for pages Search Console sees in the top 20 with no citation in 30 days, or ?view=cited_unhealthy for pages AI cites while Available or Understood is weak.
Use the Filter menu to switch between Your pages, Everywhere else and Both, or to pick Stalls at one stage and see every page whose first problem sits at the same point. One page can fit more than one view; these are practical slices, not exclusive labels.
The default columns are page, AI outcome, journey, work, cited, last cited and By. Bots and Clicks sit behind the column control, because many brands have not connected those sources and a column of en dashes teaches nothing.
What to do and History on a record
Every owned record has one work list, under What to do. There is no separate suggestions area: the status column says where each item stands, and the title says what the work is.
The status words are shared with Actions and Results. Suggested means work is ready for your decision, Needs you means it waits on an approval, In progress means someone is doing it, Measuring means it shipped and its window is open, and then comes the result: Earned, No movement, Couldn't measure, Reverted or Coincided with a drop. Shipping and earning stay separate, so shipped with No movement is a valid, visible answer.
Where a measured result has a chart, What the work earned draws the before-and-after series with the shipped date marked.
History keeps the dated record: when Trakkr found the page, when work shipped, when measurement finished, and when a stage slipped back. Turning off tracking changes whether the page is watched; it erases none of that.
Measure now
Measure now starts a fresh check for one page. The results do not all arrive at once, because the six stages use different sources.
While the check runs you may see Queued, Measuring or Scheduled for tonight. When the immediate checks finish the page can say Measured even though the overnight stages are still scheduled. If a check cannot finish, your previous results are unchanged. Starting several measurements in quick succession is rate limited.
The primary control on the record changes with what the page needs most: Review next action when work is waiting, Connect crawler or Connect traffic data when a stage is dark for want of a connection, and Measure now otherwise.
Add pages yourself
Use Add pages in the index when a page matters before Trakkr has found it elsewhere.
Paste one full URL per line, up to 25 at a time. Duplicates are removed and the field counts what it accepted. Manual pages must belong to the active brand's website; assets, invalid addresses and unrelated hosts are rejected.
The button says Add and measure, because the same step creates the record and starts its checks. For more than 25 URLs, add them in batches. A sitemap is still the better route for a whole site.
What Not measured means
Not measured is a first-class state. It does not mean zero, broken or bad.
A stage is grey because crawler tracking is not connected, traffic data is not connected, citation history is pending, or the on-demand check has not run yet. It stays visible so you know what is missing, and links to the connection that would fix it.
The rule is strict: Not measured never counts against you. It is excluded from the bottleneck, the verdict and every view count. Grey means present and non-blocking.
Common questions
Why is a stage grey?
Trakkr does not have the evidence for that check. Open the stage to see whether the next step is a page check, crawler tracking, traffic data or citation history. Grey is Not measured, not failure.
Why does Relevant say Estimated?
Relevance is inferred from questions detected on the page and your tracked prompts. Trakkr cannot observe a model's private selection, so the state is marked Estimated wherever it appears.
Why does an external page have no journey?
Trakkr can observe citations and linked work for an external URL, but it cannot claim the technical access or traffic evidence it has for your own site. The record shows why the page matters instead.
How does Pages relate to the Pages tab in Optimize?
They share page identity but answer different questions. Optimize shows pages from a site scan and the checks found there. Pages joins that evidence with crawler activity, citations, traffic, work and measured results over time.
Why is a page I published not here yet?
Trakkr normally finds it through the article record, sitemap, next scan, Search Console or crawler activity. If the URL is live and belongs to your site, use Add pages to create and measure it now.
What happens to history if I stop tracking a page?
The record stays. Its work, measured results and History are not deleted, and turning tracking back on continues from the same record.
Why is the list empty when my site has thousands of pages?
Read the sentence on the empty state: it separates a site not crawled yet from one that turned Trakkr away, timed out or published no readable sitemap. Bot protection that blocks Trakkr usually blocks AI crawlers too, so that answer is itself a finding.