How to Get Employee Support When Introducing Telemetry
When a company tells employees "we're installing a system that can see what developers are working on" or "we're introducing a monitoring system," the reaction is almost always the same: this is surveillance. Even if your goals are good — less chaos in workload, less burnout, a clearer view of where processes bottleneck or are simply broken. Resistance here is normal. It appears because the framing does not inspire trust. And the further a manager is from the team, and the more past reasons there were to doubt sincerity, the less trust there is in the change.
Meaning first
The most common mistake is to install an agent or plugin and only then explain what it is for — or explain nothing at all. An employee who found an unfamiliar process on the system themselves will not listen to explanations.
Before install, we recommend answering the following questions out loud for everyone involved:
- Why? — do not say the install is done abstractly "for control." There should be a concrete goal, for example to understand where the team's focus goes, or to compare hypotheses about process effectiveness.
- What do we collect? — in our case these are telemetry signals: time in the IDE, application usage, linkage to tasks.
- What do we not collect? — PanDev does not collect keystrokes, mouse movement, the content of private messages, screenshots or screencasts, camera video, or after-hours activity.
- Who has access to the data? — access levels should be transparent. In our system, for example, there is a role model: the user themselves, their direct manager, and managers higher in the hierarchy have access. Data is shown in full, and the manager does not get any extended dashboards about their employee beyond that.
If you do not give clear answers to these questions, you can get lost along the way and never reach the intended goal.
Talk to managers first, then everyone
Managers stand on the business side, so goals and ways of using the data need to be discussed with them first. How well team leads and mid-level managers understand the goals determines rollout success, because they are the ones who will answer the team's questions. If they themselves learned about the system the day before install, anxiety will pass from them to the team.
Since any process should be designed in advance, the integration should be treated as such a process — with a clear beginning and end. Its lifecycle looks like this:
- A conversation with the people who will read the reports: why, and which decisions can and cannot be made based on this data.
- An all-hands Q&A or a letter to the team — with the same wording, without softening fluff.
- A written policy: what is collected, when, who has access, how long it is kept, and what the data is not used for.
- A pilot with a small group willing to give feedback. The ideal case is when that team's manager leads by example, shows how it works, and demonstrates it to the team.
- Setting changes after the pilot. Process feedback, adjust processes and training, and only then roll out to more people or everyone.
Refusal to participate or strong objections in the pilot are not yet a failure, but a strong signal that the wording or process settings are still raw.
Let people see what you see
A one-way mirror is almost always read as surveillance. When a person opens their own dashboard and sees the same numbers as their manager, it becomes clear that the user and the manager share the same picture. That is a way to reduce resistance — as is the manager showing their own data. It also usually improves data quality: people start noticing labeling errors, odd categories, wrong task links, and get involved in improving internal processes.
Answer objections directly
Typical lines and workable answers:
"You'll reduce all my work to one number"
One metric does not describe an engineer's work. Use a set of signals and context. Individual KPIs on activity hours or lines of code are a straight path to gaming the metric.
"This is for firings"
Write it down: system data alone is not grounds for disciplinary decisions. If you cannot promise that — people resist for good reason.
"Our work differs; comparison is unfair"
Compare a person to their own trend. Teams — with ones that have similar context. Do not build a public ranking of "who codes more."
How to talk about the goal
Wording that usually works:
- "We want to see where processes eat time, not who is 'working badly'"
- "Data for conversation and load balancing, not for a public ranking"
- "We start with teams and trends; individual cuts are only for 1:1s and support"
Wording that almost always hurts:
- "Now we'll have full transparency on everyone"
- "We'll see real productivity"
- "If you don't install — you must have something to hide"
Technical deploy is separate from trust
A healthy pairing looks like this: announcement and meeting are done → people know the solution will appear → IT rolls out the agent or plugin → for the first days there is a channel for questions ("why is my day empty," "is this personal," "how do I turn it off after hours").
Short checklist before you start
- There is one clear rollout goal for the next quarter
- What is collected and what is not is written down
- Roles with data access are named
- There are clear rules: for example, not for punishment and not for public individual rankings
- Managers were prepared before the general announcement
- The team got an explanation before install
- People have access to their own data
- A pilot and a place for feedback-driven changes are planned
Resistance appears when people lack an understanding of why the tools are introduced and who controls the meaning of the numbers. If the meaning was made transparent in advance, install stops being a surprise. If there is no meaning, any agent on the machine will read as surveillance, no matter how soft the presentation is.
On using metrics you already collect without a toxic culture: Engineering Metrics Without Toxicity.
