Did an assistant cite you this week?
Citability tells you a page can be quoted. The Citation Tracker tells you whether it was, this week, on the questions your buyers ask. What it counts on Monday, how to choose the prompts, the only two sentences it adds to your reading, and what it refuses to be.
What you get on Monday
On Monday you know which questions stopped citing you.
Once a week, on Monday morning, the Citation Tracker asks an answer engine the questions you gave it. For each one it writes down two things: whether your brand was named, and which of your pages were cited. Then it puts this week next to last week.
What you read takes a minute. How many of your prompts named your brand. How many cited one of your pages. Which prompts stopped citing you since last week, and which started. Which of your pages were cited, and how often. Every number is a count of something that happened, and every page in it is yours.
Choosing the prompts a buyer types
You write the questions buyers ask. Monday does the rest.
The module does not invent questions. You write them, because you know what your buyers ask when they are close to choosing. Think of three kinds: the job they need done, the comparison they are weighing, and the question they ask just before they pay.
- Write each prompt the way a buyer would type it into an assistant, not the way you would write a keyword.
- Twenty to thirty prompts make a good set. The Agency plan allows up to 30 per site.
- Every prompt is one call to the engine every week, so each one should earn its place.
The prompt that stopped citing you
A prompt cited you last week and not this week. You see which, and the page it lost.
This is the first of only two sentences the tracker can add to your reading. A prompt that cited one of your pages last week does not cite it this week. The sentence names the prompt and the page, so you know where to look before you open anything.
Nothing here is a guess. A page counts as cited when its address is among the sources of the answer. Your brand counts as named when its name appears in the text as a whole word. Both are rules in code, not the opinion of a model, and only addresses on your own site are kept.
You can wait for Monday or ask for this week’s run from the screen. Either way a week gets one run, so a button pressed on Tuesday does not count the week twice.
The strong page nobody cites
Your best page never cited? See which one is cited instead.
The second sentence is about a page that should be quoted and is not. The tracker takes the strength of each page from your latest reading: an indexable page that passes every citability check starts strong, each failed check costs it points, and a page behind a closed gate counts as zero. When a page that strong is never cited while a clearly weaker page of yours is, the reading says so.
Open the weaker page. It is the one the engine found easy to quote, and the strong page can learn from it.
Those two sentences are all the reading gets. A count that did not change is not news, and a count that went up is not work, so neither reaches Monday. Gains, cited pages and totals stay on the module’s own screen. The sentence is not a card either: it does not join the week’s five and has nothing to mark as fixed.
Can be quoted, or was quoted
Citability says an assistant could quote you. This says whether it did.
Citability already answers the first question: can an answer engine quote this page, and how easily. Three gates come first: a snippet directive that forbids quoting, a crawler blocked by robots, and HTML that is nearly empty without JavaScript. One closed gate and nothing else matters.
What Citability cannot answer is whether you were quoted, this week, on the questions someone pays for. That cannot be read off your pages. It has to be asked. The tracker asks, counts, and compares with last week. The two modules meet in one place: the strength behind the strong page nobody cites comes from the Citability checks.
What it refuses to be
A citation count, not a rank tracker. No charts that only go up.
It is not a rank tracker. It never asks where you appear in an answer, never estimates a share of voice, never draws a trend line and never compares you with competitors. It asks one engine, no more than once a week, and it never writes prompts for you.
It does not keep the answers. Code judges what the engine wrote and throws it away, and only the verdict stays. No sentence the module says needs the text, and holding someone else’s answers for weeks is a liability with nothing in return.
It lives on the Agency plan, with a cap of 30 prompts per site, because a weekly run is a cost: every prompt is a call to an answer engine with web search. If no engine is set up, the run fails with the reason written down, so nobody mistakes an invented number for a measurement.