Email Insights
Your inbox is full of unprompted feedback: complaints, questions and thank-yous that customers send without ever being asked for a survey. Email Insights reads that mail for you. Every email is scored for sentiment and sorted into topics, so you can see at a glance how much mail you're getting, how people feel when they write to you, and what they're writing about — then open the emails themselves to read the real words.
What's in the report
The report is split into five pages, from the big picture down to the individual email.
Overview
The whole period at a glance: how much mail arrived, how it felt and what it was about.
- Emails received per month — your mail volume, and whether it's rising or falling.
- Emails by source — how many emails came from each source, with each one's sentiment split and score.
- Overall sentiment gauge — the average tone of every email in the period.
- Sentiment segments — the share of emails that are negative, neutral and positive.
- Topic frequency — the topics people write about most.
Topic Structure & Trend
What the emails are about, how the topics nest inside each other and how the mix moves over time.
- Level 1 topic flow — the broad themes and how much mail each one carries.
- Topic flow across levels 1 → 2 → 3 — how each broad theme breaks down into narrower ones, and then into specific issues.
- Quarterly and monthly evolution — whether a topic is growing or fading.
Sentiment Analysis
Overall sentiment first, then the same split opened up per topic.
- Average score and segment shares — the headline tone of your mail.
- Segments and score by month — whether people are getting happier or angrier over time.
- Topic volume against score — each topic's number of emails set against its average score, so the subjects that upset people most and affect the most emails stand out.
- Score and segments by topic — the sentiment of every topic, with absolute numbers alongside the percentages.
Emails
A comfortable reading view of the emails themselves, newest first. It shows the body text only, so you can skim what people actually wrote without the surrounding columns getting in the way.
Raw Data Extract
The same emails with every reporting column — received, source, subject, body text, topics and sentiment — ready to export for further analysis.
Filters
Date range, Sentiment, Source, Subject, Topic level and Topic are available on every page, and apply to every chart and table on it.
How to read the figures
- Dates — every date and figure is based on when the email arrived in your inbox, not when it was later forwarded to us for analysis.
- Sentiment score — each email is scored on a 1 to 5 scale: 1 is very negative, 3 is neutral and 5 is very positive. The figures shown are the average of the individual email scores, so an average of 2.7 means your mail leans negative overall — not that 2.7 emails were negative.
- Topic levels — level 1 is the broad theme, level 2 narrows it and level 3 is the specific issue. The Topic level filter sets which level the charts use; the topic flow chart shows all three at once.
- Source — the address that forwarded the email to us, not the person who wrote it. Where several departments forward mail, each email is attributed to the one that sent it on.
Topic counts overlap
An email can be assigned more than one topic. That means topic counts don't add up to the total number of emails, the counts at one level don't add up to the level above it, and percentages across topics can add up to more than 100%. Sentiment segments do add up to 100%, because each email is scored once.
Personal data is masked before analysis
Names, phone numbers, email and postal addresses, and ID and tax numbers are detected automatically and replaced with placeholders such as [PERSON] or [PHONE] before the subject and text are analyzed. Everything else is left exactly as written.
Where to look
| If you want to know… | Go to |
|---|---|
| How much mail are we getting, and is it rising? | Overview → Emails received |
| What are people writing to us about? | Overview → Topic frequency |
| Which broad theme breaks down into which specific issues? | Topic Structure & Trend → Topic flow across levels |
| Is a topic growing or fading? | Topic Structure & Trend → Monthly or quarterly evolution |
| Are people getting happier or angrier over time? | Sentiment Analysis → Segments and score by month |
| Which subject upsets people most, and how many emails does it affect? | Sentiment Analysis → Topic volume against score |
| What did people actually write? | Emails |
| I need this in a spreadsheet. | Raw Data Extract — export the table |
Questions about a figure?
If a number doesn't look the way you expect, contact your account manager — they can walk you through how it was worked out.