Cross-group sentiment recovered from 4.6 to 7.2 over two quarters, closing about 94% of an inclusion gap the engagement survey had averaged away.
Vantage Retail Group opted in to aggregate, anonymized climate monitoring across demographic dimensions, governed entirely by its own consent and DPIA process. The annual engagement survey read broadly positive and no formal complaint had been escalated, yet tone between different employee groups had quietly cooled from 6.7 to 4.6 over five months while tone within each group held near 7.2. SentiTrack surfaced that widening gap as an early inter-group climate signal that the blended company average could not show. Leadership responded with targeted inclusion training, structured mixed-group team-building, and deliberately cross-group project staffing, and the same measure tracked cross-group sentiment back to 7.2.
A positive survey that hid a widening divide
Vantage employs 1,200 people across Manchester and EU sites. Its annual engagement survey scored well overall, and HR had received no formal complaint, so by every conventional measure the workplace climate looked healthy.
The weakness was structural, not procedural. A single annual average blends warm within-group interactions with cooling cross-group ones, so a problem confined to how distinct employee groups treated each other could grow for months without moving the headline number.
Vantage wanted to know whether its inclusion posture held in day-to-day working relationships, not just in a once-a-year self-report. It enabled aggregate, anonymized monitoring across demographic dimensions under its own consent and DPIA process, with GDPR Article 9 obligations in mind.
What SentiTrack saw
From January, cross-group sentiment fell month over month: 6.7, 6.4, 6.0, 5.5, 5.0, to a low of 4.6 in June, a drop of 2.1 points. Over the same period within-group sentiment barely moved, holding near 7.2 within a narrow 7.1 to 7.5 band, while the company-wide average dipped only modestly, from 7.1 to 6.3.
The signal was the gap, not the level. At the June low point, tone between groups trailed tone within groups by roughly 2.5 points, a divergence an aggregate survey cannot reveal because it reports the blended company number, which stayed in the comfortable low-to-mid 6s.
SentiTrack flagged the widening cross-group gap in May, before it reached its floor. The system reported only anonymized, aggregate scores between groups; it never named a group, never scored an individual, and never attributed tone to any protected attribute.
- Cross-group tone fell from 6.7 to 4.6 over five months (-2.1 pts)
- Within-group tone held near 7.2 throughout (7.1 to 7.5 band)
- ~2.5 pt gap between within- and cross-group tone at the June low
- Flagged in May, ahead of the floor
Three sentiment series, January to the following February. Cross-group tone falls from 6.7 to a low of 4.6 in June, then climbs to 7.2, diverging sharply from the steady within-group line near 7.2; the company average holds in the low 7s and dips only into the mid-6s, masking the gap. Markers show when SentiTrack flagged the cross-group gap (May), when inclusion training and team-building began (July), and when mixed-team projects rolled out (October).
Before-and-after group-to-group sentiment matrix for four anonymized groups (A-D, never named or individually identified). In June the off-diagonal cross-group cells are cold (4.1-5.2); by February they have warmed almost uniformly to 6.5-6.9. The within-group diagonal stays steady at roughly 7.3-7.5 throughout.
Treating it as a climate problem, not an incident
Leadership read the divergence as an early indicator of inter-group tension and bias in the workplace climate, not a single incident to investigate. Because the data was aggregate and anonymized, the response targeted the environment rather than any individual or group.
Two intervention points are marked on the timeline. In July, Vantage ran targeted inclusion and anti-bias training alongside structured mixed-group team-building. In October, it rolled out deliberately cross-group project staffing so that distinct groups worked together on shared goals by default.
Each step was sequenced so its effect could be read against the same cross-group series, making the programme something Vantage could measure rather than assume.
A measured recovery, not a hopeful one
Cross-group sentiment turned in June and climbed through the second half of the year: 4.7, 4.8, 5.6, 6.2, 6.7, 7.0, 7.1, reaching 7.2 by the following February. That is a 2.6-point recovery from the low, closing about 94% of the gap to the steady within-group baseline near 7.2.
The before-and-after group-to-group matrix made the change concrete. In June the off-diagonal cells, representing tone between different groups, sat cold between 4.1 and 5.2; by February they had warmed almost uniformly to 6.5 to 6.9, while the within-group diagonal held around 7.3 to 7.5.
The value was twofold. Vantage surfaced a climate risk months before its annual survey could have, and it watched the recovery land in the same measure as the inclusion programme took effect, rather than inferring the outcome from a later survey.
- +2.6 pts cross-group recovery (4.6 to 7.2)
- ~94% of the climate gap closed over two quarters
- Off-diagonal matrix cells warmed from 4.1-5.2 to 6.5-6.9
Our engagement survey told us we were fine. The tone between our teams told us we had two quarters to act. We could see the inclusion work landing, month by month, in the same number that warned us.
- A blended company average can stay comfortable while tone between groups cools. The gap between within-group and cross-group sentiment is the signal that matters.
- SentiTrack flagged a roughly 2.5-point inter-group divergence in May, months before the annual engagement survey could surface it.
- The same aggregate, anonymized measure that detected the climate risk also tracked the recovery as the inclusion programme took effect, with cross-group tone returning from 4.6 to 7.2 and about 94% of the gap closed.
See your own org's tone, live
Explore the full product with sample data — no signup — or connect your mail in minutes.