Tracking Terms and Rivals

The signals overview explains when Tracking is the right tool: you already know what you care about, and you want to be told. This is how each half works.

One sidebar entry, two subjects behind it. Terms watch a word or phrase across everything you follow. Rivals watch specific accounts over time. Max plan.

Terms

A term watch is a query plus any aliases, checked against every article that arrives.

Aliases are the whole craft of it. One subject is rarely spelled one way — a company has a name, a ticker, a product and a nickname, and a watch that only knows one of them misses most of what it is for. Add them all to the same watch rather than making four:

Anthropic, Claude, ANTH

Case is preserved on purpose. This is what separates the ticker ON from the word "on", and IT the sector from "it" the pronoun. Write a ticker in capitals and it matches only capitals. Write a normal name in lower case and it matches however the article spells it.

That single rule prevents the most common failure here. A watch on ai in lower case behaves reasonably; a watch on short lower-case words that also appear inside other words does not, for the same substring reason muted topics have.

Sentiment

A term watch can show how coverage of it has moved over a seven-day window: an overall reading plus a point per day.

Read it as direction, not verdict. It measures the tone of how your sources wrote about something, which is a real signal about the coverage and a weak one about the thing itself.

Alerts

A watch can push to the chat bots you have connected as well as showing in the reader. Alerts follow the channels you pick per watch, so a watch you want to know about immediately can go to Telegram while a background one stays in-app.

Chat delivery needs a connected bot — see Telegram and Discord.

Rivals

A rival watchlist is a set of accounts you want measured rather than merely read. Over a window you choose — anywhere from 1 to 30 days, 7 by default — it reports:

Output. How much each account published in the window. The plainest number and often the most useful: someone who normally posts nine times a week and posted two is telling you something.

Cadence. The typical gap between posts. Distinguishes a steady daily account from one that publishes six things on Tuesday and nothing else, which read identically as a weekly total.

Posting times. When in the day they publish. Useful if you are trying to land in the same conversation rather than after it.

Cross-source pickup. For each of their posts, how many other sources in your feed covered the same thing. This is the one that answers "did this land?" — an account's own volume says what they attempted, and pickup says what the rest of the world did about it.

Hot items are ranked by pickup across the whole watchlist, so the top of that list is what your rivals published that everyone else then wrote about.

Which half to use

A term when the subject is a thing: a company, a ticker, a technology, a piece of regulation. It follows the subject wherever it is mentioned, including in sources that have never written about it before.

A rival watchlist when the subject is a person or publication. It follows them regardless of what they write about.

They answer different questions and both are cheap to keep. A term watch that fires twice a year is not waste — it is the two times you needed it.

If something looks wrong

My watch never fires — check the spelling and the case first, then check that your sources would plausibly cover it. A watch only sees articles from feeds you follow; it does not search the web.

It fires constantly — the term is too broad, or it is matching inside longer words. Prefer specific multi-word terms, and use aliases for the genuine variants rather than a shorter stem that catches them all by accident.

A ticker matches ordinary sentences — write it in capitals. Case is preserved precisely so ON and "on" stay different.

Pickup is always zero — pickup counts other sources you follow covering the same story. With a narrow feed there is nobody available to pick anything up; it is a measure relative to your own sources, not the whole internet.

Cadence looks wrong for an account that posts in bursts — it is a typical gap, so a burst publisher's average sits between the burst and the silence and describes neither. Read it alongside output rather than on its own.

I do not have Tracking — Max plan, along with Drafts. See plans.