For months, the delivery cooperative used a simple log so that orders would not disappear between the warehouse and the routes. Each person recorded when they collected a parcel, what incident they encountered, and when delivery would be possible. The log helped someone call a customer when an address was incomplete, redistribute an area when it rained, and alert the team that a van had broken down. Nobody understood it as an individual assessment. It was a coordination tool, made to stop everyday work from falling into disorder.
One Monday, management presented a new slide. Using those same data, it had produced a productivity ranking. The person with the fewest deliveries appeared in red; the person with the most appeared in green. The slide did not explain that some routes had broken elevators, that certain parcels required age verification, or that some workers spent part of their shift training new colleagues. Nor had anyone been told that the log would change purpose. What began as a shared aid had become, without discussion or agreement, a disciplinary instrument.
The harm does not depend only on whether the ranking is technically imperfect. It changes the employment relationship. When data created to coordinate are reused to measure or punish, each entry stops being a signal for solving a shared problem and becomes possible evidence against the person who wrote it. Silence is the predictable result: incidents go unrecorded, help goes undeclared, difficult routes are avoided, and colleagues compete for a number that cannot describe the work itself. A tool can keep functioning while weakening the cooperation it was built to support.
One purpose does not travel automatically to another#
Accepting an application to organize a task does not mean accepting every future use of its data. The distinction can seem formal until it is viewed from the workplace. Recording that a delivery was delayed can help adjust a route. Using that same entry to decide a bonus, open an investigation, or label someone “slow” requires a different justification, different information, and different safeguards. Practical agreement to the first tool is not automatically inherited by the second purpose.
Accepting an application to organize a task does not mean accepting every future use of its data.
This separation also protects an organization from poor decisions. An operational log usually captures what is easy to count, not everything that explains an outcome. It may record departure and arrival times, but not time spent handling a safety incident, supporting an apprentice, or preventing a vulnerable customer from losing a service. Turning that incomplete part of reality into an individual score does not add objectivity. It can turn a design omission into an appearance of certainty.
That is why the right question is not only, “Can we extract this data?” It is, “Why were they collected, and what changes if we use them differently now?” The second question requires the group to describe the decision it wants to make, the possible harm, and less intrusive alternatives. If the aim is better planning, reviewing areas, schedules, or team resources may be enough. If the issue is a repeated incident, it may call for a conversation, training, or redesign—not retrospective surveillance of every movement.
A ranking is not an innocent photograph#
A ranking looks simple because it reduces a complex reality to an ordered list. That simplification is exactly its power and its risk. A list makes it visible who is at the top and bottom, invites comparison, and can associate a position with merit or blame even where the metric does not prove either. In a workplace, that effect grows: a number can influence shifts, opportunities, reputation, and economic security before the person knows how it was calculated.
The cooperative may find, for example, that one courier appears consistently below average because she serves an area with buildings that lack elevators and customers with narrow delivery windows. Another colleague may rank highly because he covers a short, stable route. If nobody includes or allows discussion of those conditions, the ranking rewards favorable context and penalizes less visible work. Still, the slide tells a tempting story: the list reveals who works best. The problem is not solved by adding a small footnote. The group has to ask whether the list should exist at all.
Individual comparison can also pull attention away from problems that can actually be repaired. Instead of fixing an unequal assignment system, it invites people to run faster inside it. Instead of adding staff for complex incidents, it treats the person who handles them as less productive. A measure that turns collective failures into individual defects may be convenient for someone reading a dashboard, but it is costly for the people who sustain the service.
Worker representation belongs at the beginning#
A decision to reuse data should not be announced after it has already been built. Worker representatives, delegates, or people chosen by the team need to participate before the metric is defined, access is configured, or a trial is run. Their role is not merely to check whether a document uses the right words. They can identify what part of the work is missing, which uses would become undue pressure, and what real consequences a score would have for shifts, pay, or continued employment.
The discussion needs understandable materials: a sample of the data, the exact purpose, who can see them, which decisions might rely on them, how long they would be retained, and what will not be allowed. “We will analyze performance to improve” is not enough. It does not say whether the team or the individual will be observed, whether sanctions are possible, whether sources will be linked, or whether an inference can be corrected. Ambiguity deprives workers of the ability to debate a change that may affect their working lives.
Representation can also propose a necessity test. Before designing a ranking, the cooperative can state the problem in a testable sentence: “There are delays in three areas during two specific time windows.” It can then try solutions that do not individualize: rebalance loads, review routes, repair the address system, and add support at peak times. Only if those measures fail to meet a demonstrated need would it make sense to discuss additional data. Even then, the question is not how to score people, but what minimum information can repair the process.
Make the limit a verifiable prohibition#
Saying that data “will not be misused” is not real protection. A useful prohibition can be checked. It should state that coordination records will not feed individual rankings, disciplinary files, pay decisions, or behavioral profiles without a newly agreed basis and prior review. It should identify which fields are separated, which systems cannot be connected, who has permissions, and how an exceptional request is recorded.
Technical design matters because a policy without barriers relies too heavily on goodwill. If a platform administrator can export any history in two clicks, the prohibition is fragile. The cooperative can limit exports, separate databases, block identifying columns in operational analysis, and require dual authorization for exceptional access. It can also keep an access log that worker representation may inspect. The purpose is not to surveil the person who administers the system. It is to make a boundary detectable before harm becomes normal.
Verification should include a periodic review with concrete questions: Has any individual dashboard been created? Have records been linked to human-resources data? Has there been an access request outside the original purpose? Was information deleted once it was no longer necessary? Answers need a named owner and a date. If a breach appears, the protocol must stop the use, inform those affected, withdraw the result, and repair its consequences. Without that route, a written limit becomes a promise with no remedy.
Restore the tool's purpose#
Separating purposes does not mean that teams must give up learning from their records. It means restoring the purpose that justified those records in the first place. The cooperative can analyze, in aggregate, which areas create most incidents, which schedules concentrate delays, or what kind of order needs clearer instructions. It can share that finding with the team and agree changes instead of turning every timestamp into a personal mark.
The next step is concrete. The cooperative can review its current log with worker representation and write down, field by field, its permitted purpose and prohibited uses. It can remove the ranking, explain why, and publish an audit mechanism with named owners and dates. Then, when a service problem appears, the first question should be what is failing in the system before asking who falls on a list. Coordination makes common work possible; surveillance cannot be presented as its inevitable consequence.
