Sentiment in monitoring is a filter for mentions, not a score for news

When your company gets mentioned in Telegram every day, both volume and priority become a problem. Forty mentions in a day: thirty-five neutral reprints of a press release, three happy reviews, two critical write-ups. Two of them need a response, but you have to read all forty. Tone tagging solves exactly that. It doesn't tell you whether the news is good or bad; it tells you how you were written about.

Tone is a property of the mention, not of its importance

It's easy to confuse tone with significance, and that leads to the wrong expectations. Gista scores significance separately: a news-filter feed combines match with your interest, reach, influence and factuality; a tracking feed reports the percentage match with your description. Tone is not part of that formula and never moves the score. A negative post doesn't become more important just because it's negative, and a flattering one doesn't sink.

Instead, tone is tagged as its own field, relative to the subject of the post — the thing you're watching. «Company faces a lawsuit» is negative toward the company even if the channel wrote it in flat, factual language. «Service fixed a long-standing bug» is positive even if the wording is dry. It's not about the author's emotions, it's about the light your object of interest is cast in. The model tags this for every feed regardless of type, so the data accumulates even if you aren't using tone yet.

When tone becomes a filter

The feed page has a Filter by tone switch and three pills: Negative, Neutral, Positive. Turn the filter on, keep only Negative, and the feed delivers just the mentions where you were written about badly. That's the support-and-PR mode: instead of reading the whole stream, you get what needs an answer as early as possible.

One detail worth knowing up front: this filter is stricter than any other rule. Normally a match with the "always important" field pushes a post through regardless of the threshold — but if its tone isn't among the ones you selected, it still won't pass. The logic is simple: you explicitly said "I only want negative", and a bypass for neutral mentions would make the setting meaningless. So if your stream drops sharply after switching the filter on, check the selected tones first, not the interest description.

The reverse case works just as well. A marketing team leaves only Positive to collect reviews and mentions worth showcasing, without wading through complaints. An analyst keeps all three and watches the proportion: the share of negative mentions week over week is a metric in itself, visible even when nothing urgent is happening.

The marker inside the message

Tone doesn't have to be a filter — you can simply see it. The Show tone in the Telegram message switch adds a colored dot after the score of every item in the feed. Three display modes: All puts a marker on every post, No neutral marks only the colored ones, and Negative only shows a red dot where you need to step in and leaves the other lines clean.

This is purely visual: it doesn't affect selection and is independent of the filter. The combination worth assembling most often is filter off (the full stream of mentions still arrives) with display set to negative only. The feed stays complete, but the problem lines stand out at a glance.

How to turn it on

Tone works in any feed, but it earns its place in tracking feeds — the mode that measures the match with your description rather than the significance of an event. Describe what you're monitoring in the interest fields, set the focus, then decide what you need more: filtering the stream by tone, or seeing all of it with colored markers. Both are switches on the feed page, changeable at any time, with nothing rebuilt retroactively.

If you haven't tried it yet, start for free: a feed takes a few minutes to set up and arrives straight to your Telegram direct messages.