Dig into traffic and conversion drops right in the chat

The agent starts from the list of counters your account can reach and then works through the Metrica API: it builds reports on visits, sources, pages and devices, counts goal completions and conversion, and when the aggregates are not enough it drops down to visit logs for the details of an individual session. It always compares a period against the previous one, so a drop shows up as a change rather than as a single number.

The agent keeps two Metrica specifics in mind. First, data ripens: the freshest tail of the day is not fully processed, so «yesterday versus the day before» easily shows a false dip, and for comparisons the agent takes closed periods. Second, conversion exists exactly as far as goals are configured in the counter. If a goal is not set up it will not appear in the report — and the agent says the goal is missing rather than that conversion is zero.

The agent never writes to Metrica: it does not create or rename goals and does not touch filters, segments or counter settings. All the work is reading and analysis: rolling up traffic by channel, finding the sources and pages where visits and goals slipped, spikes and dips in traffic tied to sources and devices, and a short digest of the key numbers instead of opening the cabinet.

The agent builds reports through the reporting API: it picks the metrics and dimensions the question needs rather than pulling the whole set, and splits a long period into windows. When the aggregates are not enough it goes to the visit logs — those do not answer at once but prepare an export, so per-visit detail arrives after the summary. The agent holds the request rate itself, and if there is no data for the period or the counter is unavailable it says so directly instead of showing an empty table.