Understanding your monthly conversation insights email

Every month, Tribe sends you a report summarising the conversations prospective students had with your ambassadors: what they asked about, what they responded well to, and what worried them.

This guide explains how the report is put together and what each section shows.

The report starts with two numbers: how many conversations took place in the reporting window, and how many subjects were discussed within them.

A subject is a stretch of conversation about one theme. One conversation usually contains several: a prospect might ask about the application process, then fees, then housing in a single chat, and each of those counts as its own subject. Greetings, thank-yous and small talk are excluded from both numbers.

Three to five short observations that connect the report's numbers, for example a rising topic paired with a related concern. They are written from the aggregated figures shown in the report and nothing else, and they describe patterns rather than giving recommendations.

This section is the core of the report. It lists every topic prospects raised during the period, ranked by how often it came up, with a bar showing its relative weight. The most common topics are expanded further: they show which subtopics made up the discussion and include example questions, while the remaining topics appear as a ranked list only.

The percentage next to each topic is the share of conversations that touched it. Because one conversation can cover several topics, the percentages add up to more than 100%.

The arrows (▲▼) show the change in percentage points compared with the previous reporting period. A first report has no arrows because there is no earlier period to compare against.

The expanded topics break down into their most common subtopics. Those percentages are the share within that topic. For example: of the money-related discussions, half touched working while studying.

The "prospects are asking" lines are example questions attached to the most common subtopics. Each one summarises a question pattern that appeared repeatedly across different conversations. They are written summaries, never a real prospect's message quoted back, and they only appear where a pattern was frequent enough to summarise reliably.

Every subject a prospect raised is read for tone and tagged as positive, neutral or concerned. The percentages show how all subjects in the period were distributed across the three. Because tone is tagged per subject rather than per conversation, one chat can contribute to more than one category: the same prospect can be worried about funding and excited about the course a few messages later.

Positive means enthusiasm about the subject itself:

Neutral means factual questions with no emotional weight:

Concerned means the prospect is seeking reassurance about something:

Two boundaries keep the measure consistent. It is not a satisfaction score: prospects haven't experienced the institution yet, so the tone reflects how they feel about the subject they raised. And politeness doesn't count: a cheery "thanks so much!" is never tagged as positive, because gratitude towards the ambassador says nothing about the subject discussed. Only the prospect's side of the conversation is measured, so an enthusiastic ambassador cannot influence the score.

The concerns and enthusiasm lists further down in the report are built from these same tone tags, so the two sections always agree with each other.

Two ranked lists: the subjects where prospects sounded most worried, and the ones they responded to most warmly. Both come from the same tone reading as the general sentiment section. A subject only appears in the concerns list if it was tagged as concerned, and only in the enthusiasm list if it was tagged as positive. The counts next to each item show the number of conversations in which that subject appeared with that tone. They are rounded, and items that appeared too rarely to be meaningful are left out. Because the lists show only the most common themes, their counts add up to less than the sentiment percentages imply; the remainder is spread across many smaller subjects. The top concerns carry an example question, in the same summarised form as elsewhere in the report.

The analysis reads each conversation, splits it into subjects, and tags every subject against a fixed list of topics. The list is kept identical from month to month, so counts from different periods are directly comparable: if a topic rises, it is because prospects raised it more often, not because the categories changed. Subjects that don't fit the list are collected separately and reviewed quarterly. If a new pattern keeps appearing there, it becomes a category of its own in a future edition.

Reply to the report email. It goes to a person, and your feedback directly sharpens the next edition.