Measuring probation conversion evidence across teams

This study asks whether probation conversion decisions are supported by the criteria, evidence, calibration, and approval that the employer's procedure requires. It compares records across teams to see whether the same standard is being applied. The result is intended to improve consistency and documentation, not to rate individual workers or managers. The scope is one employer or one defined group of teams during a fixed period. The study may compare teams, but it should not present a local consistency rate as an industry benchmark. A small sample can reveal a process gap even when most decisions are well documented.

The population should be a frozen list of probation periods that reached a conversion decision during the window, with the team, the role, and the decision recorded. Include extensions and separations as well as conversions, because the standard applies to every outcome. Exclude workers who left before the decision date for an unrelated reason, and record the reason. The unit is one probation decision. If a worker had two probation periods, treat each as a separate unit and link them. Report the number of workers and the number of decisions separately.

Define the required evidence before reviewing any decision. The record should state the criteria given to the worker, the check-ins held, the work samples reviewed, the calibration conversation, and the approver. A conversion should show that the criteria were met, not only that no problem was recorded. An extension should state the specific gap and the support provided. A separation should follow the employer's procedure and record the approver. Distinguish a missing document from a missing decision and from a decision that is documented but inconsistent with the criteria.

Draw the sample from the frozen decision list using a documented method. Stratify by team, role family, and outcome so the study does not overrepresent conversions or one team. Include decisions that were appealed or revisited, and record the outcome. Record the selection method and any replacement. If the sample is small, report every decision rather than a rate. A sample of only smooth conversions will overstate consistency, while a sample of only disputed cases will describe failure modes without showing the base rate.

Two reviewers should classify a calibration set before the main review. Compare their judgments on each element and discuss the cases where they disagree. A disagreement about whether a check-in produced usable evidence is common, and it should be resolved in the standard. Revise the definitions and repeat the calibration. During the main review, preserve the initial ratings and the adjudication. If the reviewers cannot agree, the problem may be the evidence standard rather than the manager's record.

Collect the probation plan, the check-in notes, the work samples, the calibration note, and the final decision record for each sampled case. Record the version and the date of each document. If a conversation was not documented, record the calendar entry and any follow-up message that shows a decision. Do not reconstruct the content of a meeting from memory. If the manager kept notes in a personal file, ask for the approved copy rather than copying the file into the study. The reviewer should be able to trace each scored element to a dated source.

Report the number of decisions reviewed and the number with each element present, partial, or missing. Show the result by team and by outcome. Include the number of conversions with no calibration note and the number of extensions with no specific gap. Use counts and rates together, and state the denominator. A single consistency score can hide the difference between a missing document and a decision that conflicts with the criteria. Report the disputed cases separately and describe the resolution.

Compare element rates across teams only when the subgroup is large enough. Inspect the records behind the largest gaps before assigning a cause. A cluster of missing calibration notes in one team may reflect a manager who reviews informally, while a cluster across teams may reflect a procedure that does not require a note. A cluster of conversions with vague criteria may reflect a template that was never updated. The remedy should match the stage where the gap entered, not a general call for more training.

The study reviews written records and cannot observe the quality of the work or the fairness of the decision. A complete record can still reflect a biased judgment, and a sparse record can sit beside a fair process. The study also cannot determine whether the employer's criteria are job-related or consistent with applicable law. It measures documentary support against a written standard. System changes, template updates, and manager changes can affect the result, so note the relevant dates.

Use the minimum worker information needed for the study. Replace names with study keys, restrict access to the crosswalk, and follow the employer's retention rule. Do not copy performance details or protected characteristics into the analysis file when a reference to the approved record is enough. The operations coordinator may prepare the sample and score administrative elements. The HR owner or authorized manager decides whether a conversion, extension, or separation is appropriate. The study should not be used to make an employment decision about an individual.

Choose one change based on the most common or most consequential gap. It might be a required criteria section, a calibration note, a specific extension plan, or an approval step before the decision is communicated. Name the owner, the due date, and the measure that will show whether the change helped. Recheck the same measure with the next cohort. Keep a short decision log with the finding, the change, the owner, and the result.

Two teams convert a worker after probation. One record states the criteria, three check-ins, two work samples, and a calibration note. The other record states that the worker performed well and includes no criteria or samples. Both decisions may be reasonable, but only the first can be reviewed against a standard. The study marks the second record as partial and routes the gap to the team lead. The example shows why consistency is a documentation question before it is a fairness question, and why the standard has to be written before the decision.

Review a sample after each hiring cohort and whenever a team, role, or criteria template changes. A quarterly review may be enough for a stable operation, while a monthly review is more useful during rapid growth or a reorganization. Record the cadence in the probation procedure. Between reviews, keep a short log of decisions that were appealed or revisited. The log helps the team see whether the process is drifting before the next formal review.

A short table can show the result by team and by outcome without exposing worker identities. Use one row per decision and columns for the team, the role family, the outcome, the criteria status, the check-in status, the calibration status, and the approval status. Add a notes column for a brief reason when an element is partial or missing. Keep the definitions beside the table so a reader can interpret the codes. Report the frame size, the sample size, and the number of excluded decisions. End with the one change the team will test and the date of the next sample. A reader should be able to move from the summary to a specific decision and then to the source record.

The report should not say that a conversion decision was fair or that a manager applied the policy correctly. It should say whether the required criteria, evidence, calibration, and approval were present against the employer's standard. That distinction matters because a complete record can still support a wrong decision, and an incomplete record can be corrected before the decision is final. The report also should not compare teams as if their work and hiring volumes were identical. Keep the comparison inside the defined standard.