See sentiment move and understand why it moved
SentiTrack's Time Graph plots 1–10 sentiment as smoothed parallel time-series, split by any attribute or group, and overlays a calendar of events so a dip lines up with the reorg, the layoff or the policy change that caused it. Your zoom and layout are saved per user.
Sentiment-over-time analysis tracks how the emotional tone of communication trends across days, weeks or months. SentiTrack's Time Graph plots 1–10 sentiment scores as smoothed parallel time-series, split by department, location or any attribute, and overlays calendar events so every rise or dip can be tied to a real cause.
A single number can't tell you what changed
A snapshot tells you sentiment is at 6.2 today, but not whether that's a recovery or a slide. Raw daily scores are too noisy to read, and when tone does drop you're left guessing which week — and which event — set it off. The story is in the trend, and the trend is hard to see.
Smoothed trend lines, split any way, anchored to real events
The Time Graph turns scored messages into clean trend lines. Rolling smoothing is applied server-side so the chart reads at a glance instead of jittering, you split one series into many by any attribute to compare slices side by side, and a calendar overlay marks the events that explain each move.
- Parallel time-series — explicit series or auto split-by any attribute
- Server-side rolling smoothing turns noisy daily scores into a readable trend
- Calendar-event overlay lines a dip up with the reorg, layoff or policy change
- Zoom and layout saved per user, so you return to exactly the view you left
From scores to a story in three steps
Every email, call and chat is scored 1–10 across your connected channels — the same engine that powers every SentiTrack view, with bodies discarded after scoring.
Scores are bucketed over time and a rolling average is computed server-side, producing clean parallel lines you can split by department, location or any attribute you track.
Drop a calendar of full-day events onto the chart — reorgs, launches, holidays — and read shifts in context, then zoom into any window and save the layout.
Server-side rolling smoothing flattens day-to-day jitter so a real slide or recovery is obvious at a glance.
Split one line into many with split-by — by department, location, role or any attribute — to see which group is diverging.
The calendar overlay aligns a dip with the event that caused it, so you correlate tone with reorgs, launches and policy changes.
A downward trend in a team's line surfaces friction weeks before it becomes a resignation or a complaint.
Zoom level and layout are saved per user, so the window you were analysing is waiting when you come back.
Save a view and embed a read-only timeline for leadership without giving them dashboard access.
A trend line built from scores, never content
Storing sentiment alongside identifiable employees and demographic attributes can be special-category data under GDPR Article 9. SentiTrack minimizes that exposure by never persisting content: every point on the Time Graph comes from a 1–10 score and non-content metadata, with bodies discarded after scoring.
- Zero message bodies retained — only scores and metadata are plotted
- Zero subjects stored in plaintext (hash only)
- Hard-delete endpoint plus a configurable retention window (default 365 days)
- DPIA, lawful basis and consent remain the customer's responsibility under the ToS
Sentiment over time — FAQ
How does SentiTrack track sentiment over time?
SentiTrack plots 1–10 sentiment scores as parallel time-series, bucketing them over your chosen range and applying server-side rolling smoothing so the trend reads cleanly. You split the chart by any attribute and overlay a calendar of events to tie each rise or dip to a real cause, with zoom and layout saved per user.
What is rolling smoothing and why is it done server-side?
Rolling smoothing replaces each point with a running average of nearby points, flattening day-to-day noise so the underlying trend is readable. SentiTrack computes it server-side so the chart loads pre-smoothed and consistent across users and embeds, rather than recalculating jittery raw scores in every browser.
Can I split the time graph by department or attribute?
Yes. You can define explicit series or use split-by to fan one line into many across any attribute you track — department, location, role, seniority or a custom field. Each slice becomes its own smoothed line, so you instantly see which group's sentiment is diverging from the rest.
How does the calendar-event overlay work?
You attach per-org calendars holding full-day events — reorgs, launches, holidays, policy changes — and the Time Graph draws them as markers behind the trend lines. A dip then visibly lines up with the event around it, turning correlation into a question you can investigate rather than a number with no context.
Does the Time Graph use email content?
No. Every line is built from 1–10 sentiment scores and non-content metadata only. Message bodies, subject text and attachments are never stored — bodies are scored in memory then discarded — so you see how tone trends over time without retaining what anyone actually wrote.
Are my zoom and layout saved between sessions?
Yes. Your zoom window and chart layout are persisted per user, so when you reopen the Time Graph you land on exactly the range and view you were analysing. Saved views can also be shared or embedded read-only for stakeholders without dashboard access.
Watch your sentiment trend take shape
Explore the live demo's Time Graph with sample data — no signup needed — or connect your own channels and overlay your calendar to see what moved the line.