Philippines staffing guide
Review Assistant Work With a Small Quality Sample
Pick a few completed items, check the evidence, and fix one cause. A small sample beats a vague impression.
Published · 9 minute read
On this page
Quality review for an assistant does not need a large program. A small, regular sample of completed work will reveal more than a general feeling that things are going well. The sample should include routine work, exceptions, and at least one item the assistant found difficult.
The operating problem
Quality review for an assistant does not need a large program. A small, regular sample of completed work will reveal more than a general feeling that things are going well. The sample should include routine work, exceptions, and at least one item the assistant found difficult.
Define what good looks like for the task. For a report, it might be the correct source, the right audience, a clear recommendation, and a checked number. For an inbox, it might be the right label, the next action, and the absence of unsupported promises. Write the criteria before reviewing.
Pick the sample from the work record, not from the assistant's favorite examples. Include items from different requesters and different days. If the assistant chooses the sample, the review will show what they are proud of, not what the owner needs to see.
Review the evidence, not only the output. Open the source and compare it with the deliverable. Check the note, the version, and the destination. A polished report with a stale source is a quality failure. A plain note with the right facts may be a success.
What to control
Record one correction and one thing that worked. The correction should be specific and tied to the criterion. The positive note should be equally specific. This keeps the review balanced and gives the assistant a clear next step.
Look for patterns across the sample. If the same issue appears in three items, it is probably a process problem. The brief may be unclear, the source may be missing, or the tool may be hard to use. Fix the cause rather than coaching the same behavior repeatedly.
Review the sample with the assistant, not about them. Ask what they saw and what they would change. The assistant often knows where the process is weak. A review that feels like a test will produce defensiveness; a review that feels like a joint check will produce better evidence.
Keep the record short. The date, the items, the criteria, the findings, the correction, and the next review date are enough. A quality program that takes more time than the work will not survive. The goal is a reliable queue, not a perfect file.Choose the sample before the review, not after seeing the work. A random or stratified sample is more credible than a hand-picked set. If the assistant selects the sample, the review becomes a showcase.
A workable routine
Use the same rubric for every item in the sample. If the criteria change mid-review, the results are not comparable. Write the rubric in advance and note any exception. A simple three-level scale is easier to apply than a ten-point scale.
Share the finding with the assistant in private. A public comparison creates defensiveness and does not improve the work. The conversation should focus on the evidence and the next step, not on the person's value.
Review the rubric itself every quarter. A criterion that no longer matters should be removed. A new risk should be added with an example. A static rubric becomes irrelevant as the role changes.
Use a sample size that the reviewer can actually complete. Ten items reviewed carefully are more useful than fifty skimmed. If the queue is large, stratify by task type and pick a few from each. The goal is a credible signal, not a statistically perfect audit.
Review and evidence
Close the loop with the person who requested the work. A deliverable can meet the rubric and still miss the requester's need. Ask one requester what worked and what was missing. That feedback improves the brief and the criteria.