Communication Analysis
Analysing who talks to whom, how often and in what tone. The newest capability and the one with the least settled ground.
Newer products analyse messages, email and meeting patterns: volume, timing, sentiment, network position. It is presented as organisational insight and it is a substantial extension of what is being observed.
The practical lesson in “Communication Analysis” is to connect every record to a clear operational question without presenting visibility as certainty. Teams exploring does microsoft teams track your activity can review how Microsoft Teams records activity as one source of time and project context, provided the purpose is disclosed and the configuration is reviewed with the people affected.
What is produced
Message and email volume by person and by pair.
For an independent reference relevant to “Communication Analysis”, consult the European Commission data-protection resources; it provides a useful external check on scope, terminology, governance and the claims made during procurement or review.
Response times.
After-hours activity.
Meeting load and fragmentation.
Network maps showing who is central and who is peripheral.
And increasingly, sentiment or tone, derived by a model.
What is genuinely useful
Aggregate meeting load, which identifies a real and fixable problem.
After-hours volume as an organisational signal — a team consistently working late is a finding about workload, not about dedication.
Fragmentation: how much uninterrupted time people actually have.
These are legitimate and they are all aggregate, which is the pattern.
Where it goes wrong
Network position read as importance: the person at the centre of the map may be a bottleneck rather than a leader.
Peripheral position read as disengagement, when it may be deep focused work or a role that does not require coordination.
Response time read as responsiveness, which rewards interruption.
Each of these is a plausible reading that is frequently backwards.
Sentiment analysis specifically
Deriving mood or attitude from employees' written messages is a meaningful step beyond measuring activity.
Accuracy on workplace text is poor, particularly across cultures, first languages and neurotypes.
And the use — identifying disgruntled employees — is one that changes what the workplace is, which should be decided explicitly rather than enabled as a feature.
The content question
Does the analysis read message content, or only metadata?
Metadata analysis is a different proposition from content analysis and the products do not always distinguish clearly.
Ask, and get the answer in writing, because this is the capability most likely to have expanded since you bought.
The aggregate-only line
Everything useful here is available at team level.
Nothing useful requires naming individuals.
Hold that line explicitly in configuration, because the individual view exists in most products and will be requested.
What to decide before enabling
What question this answers that you cannot answer otherwise.
Whether content is read.
Whether anybody will see individual data, and under what process.
If the honest answers are vague, the feature should stay off.
What to check
Does your deployment analyse communication, and does it read content?
Is any of it reported at individual level?
Is sentiment analysis enabled, and did anybody decide that?
And what aggregate question is it answering?