The Analytics dashboard shows how your organization uses Ejento AI. It counts the questions people ask, the votes they give to answers, the tokens used, and the documents indexed, and lets you break each of these down by assistant, by person, or by team over any period you choose.Why Analytics matters#
See real adoption. Find out how many people use Ejento AI and how often, instead of guessing from licenses or anecdotes.
Know which assistants matter most. The busiest assistants are the ones to keep accurate and up to date first.
Catch weak answers early. Downvotes point you to assistants whose answers people don't trust.
Understand where AI usage goes. Token counts show which assistants, people, or teams use the most AI processing.
Follow your knowledge base. Document counts show how much content is being indexed.
How Analytics works#
1.
Open Organization settings > Analytics.
2.
Under Filter by, choose what to compare: Agents, Users, or Teams.
3.
Under Charts, choose what to count: Query Count, Votes Count, Token Count, or Document Count.
5.
Read the totals, the graph over time, and the distribution table that ranks each agent, user, or team.
Key terms#
Agent: The dashboard's name for an assistant. Filtering by Agents compares your assistants with each other.Query: One question or message someone sends to an assistant.Vote: An upvote or downvote a person gives to an assistant's answer. Votes are the quickest signal of whether people find the answers useful.Token: The small units of text that AI models read and write. Ejento AI estimates the cost of each query from its token counts and the model's pricing, so tokens are the closest measure of AI usage. For spend in US dollars, see Usage.Document: A file that has been indexed, so that assistants can search it and use it in their answers. See the Corpus Overview.Active agents, users, or teams: The number of agents, users, or teams that had any activity in the selected date range.Distribution: Each agent's, user's, or team's share of the period's total, shown as a percentage. For example, an assistant that received 158 of 375 queries has a distribution of 42.1%.What you can compare#
Filter by decides how the data is split. Agents compares your assistants, Users compares individual people, and Teams compares the teams in your organization. The totals, the graph lines, and the table rows all follow this choice; for example, the second total reads Active agents, Active users, or Active teams.What you can measure#
| Chart | What it counts | What it tells you |
|---|
| Query Count | Queries sent to assistants | How much Ejento AI is used, and where |
| Votes Count | Upvotes and downvotes on answers | How people rate the answers. You can show All Votes, Upvotes Only, or Downvotes Only. |
| Token Count | Tokens used | Where most AI processing, and therefore cost, goes |
| Document Count | Documents indexed | How much content is being added to your knowledge base |
Time periods#
Every chart covers the period in Date Range. Choose a preset (Today, Yesterday, Last 7 days, This month, or Last 3 months) or pick any start and end date in the calendar. This month covers the current calendar month so far. The dashboard opens on the last 30 days. Short periods help you check recent activity, for example after launching an assistant, while longer ones show trends.The graph and the distribution table#
Each chart shows the same data in two ways:The graph shows the metric over time, with one colored line per agent, user, or team. Use it to spot trends and busy days.
The distribution table ranks every agent, user, or team by its total for the period, with its share as a percentage and a bar. Click one or more rows to show only those items on the graph, and click Clear all to show everything again.
Use cases#
Prioritize assistant maintenance: Compare agents by query count over the last three months. The assistants at the top of the table are the ones most people depend on, so their instructions and corpora deserve attention first.Improve answer quality: Compare agents by votes and show downvotes only. For the assistants with the most downvotes, read the rated conversations in Chatlogs and improve their instructions or documents.Drive adoption: Compare teams by query count. Quiet teams may need training or an assistant built for their work, and busy teams can show others what works.Watch AI usage: Compare users or agents by token count to see where usage is concentrated. For costs and spend limits, use Usage.Who can access Analytics?#
The Analytics section can only be accessed by Global Admins. It's in Organization settings > Analytics.
