Define what good outreach looks like
Candidate outreach is easy to count and harder to judge. A sent message can be timely, accurate and still feel impersonal. A short sample of real messages shows whether the assistant is representing the role and the employer well. The review needs a written standard before it can score anything. Write that standard before reading a live sample so the review has a fixed reference.
Describe the elements of a good first message in plain language. It should name the role, explain why the candidate was contacted and give one clear next step. It should use the candidate's name correctly and avoid claims the employer cannot support. Add the tone the team wants, such as warm, direct or formal. Keep the checklist short enough that a reviewer can apply it during a busy week.
Turn the description into a short checklist with four or five items. Each item should be observable in the message without guessing about intent. For example, "names the role" is observable while "sounds enthusiastic" is not. A checklist that two reviewers can apply the same way is more useful than a long rubric. Test the checklist on old messages before using it for coaching.
Choose a sample size that matches volume
A sample of one message proves nothing, and a review of every message is too slow to sustain. Choose a sample size that gives the reviewer a reasonable view of the week without turning coaching into surveillance. Ten messages per assistant per week is a workable starting point for a moderate queue. State the selection rule in the routine so the sample is not chosen after the fact.
Stratify the sample by message type when the queue has distinct stages. Include first contact, follow-up, scheduling and rejection messages. A sample drawn only from easy confirmations will make the quality score look better than the candidate experience. The mix should reflect the work the assistant actually does. Record which stages were included and which were left out.
Increase the sample when a new template or a new role launches. Decrease it after several clean periods, but keep a minimum. The sample size should be written in the routine and reviewed when volume changes. A fixed number is easier to defend than an informal sense of how much to check. Explain the reason for the change to the assistant before the larger sample starts.
Read messages against the role brief
The role brief is the reference point for accuracy. Keep the current brief beside the sample and check each message against the duties, schedule, pay range and location. A message that promises a benefit the brief does not mention creates a problem the recruiter must repair later. Add the approved benefits language to the brief so the assistant does not improvise.
Check the candidate's name, the role title and the next step first. These are the details that cause immediate confusion when they are wrong. Then read the message for tone and clarity. A polite message with the wrong interview time is still a failed message. Mark the next step clearly when the message is sent through a plain-text channel.
Look for missing context as well as incorrect claims. A message may be technically true but leave the candidate unsure what happens next. The review should note when the next step is vague or the timeline is absent. Those gaps produce replies that the assistant then has to handle. Note the missing context in the same row as the accuracy check.
Check timing and follow-up behavior
Quality includes when the message was sent and what happened after it. Compare the send time with the response standard and the candidate's original message. A good reply sent two days late still fails the candidate experience. Record the delay and its cause beside the quality note. Compare the send time with the time the candidate replied, not only with the deadline.
Review the follow-up behavior for candidates who did not respond. Check that the assistant waited the approved interval and used the correct next message. Repeated follow-ups too close together feel like pressure. No follow-up at all leaves a willing candidate waiting. Check the suppression list against the applicant tracking record.
Check that the assistant stopped when the candidate asked to stop. A suppression request that is missed is a serious defect, not a minor quality issue. The sample should include at least one suppression case when one exists. If none exists, review the written suppression rule with the assistant instead. Review the follow-up sequence as one case even when the messages are separate.
Record defects by cause
Group defects by cause rather than by person. Common causes include a missing role detail, an outdated template, an unclear next step and a timing error. The cause points to the fix. A person-based count invites defensiveness and hides the process problem that produced the mistake. Name the cause in plain language so the assistant can act on it.
Use a small set of cause codes and define each one. Add an "unknown" code for cases the reviewer cannot classify. A large unknown group is a finding about the review process, not a reason to guess. Review the codes every month and retire the ones that no longer apply. Review the codes at the end of the quarter and merge any that overlap.
Keep the defect count beside the sample size. A count without a denominator is not a rate, and a rate without the sample definition is not comparable. Report both so the reader can see the evidence behind the number. The same rule applies when the team compares one month with the next. Show the denominator beside the count in every summary.
Coach with examples, not scores
The coaching conversation should start with a real message and a specific observation. Name what worked and what the candidate would have experienced. Then show the corrected version and ask the assistant to explain the difference. This approach teaches the standard without turning the review into a ranking exercise. Ask the assistant which example clarified the standard most.
Choose one or two changes per review. More than that overwhelms the assistant and dilutes the follow-up. Write the changes in the response library so the next sample can test them. A coaching note that stays in a private document does not improve the queue. Write the change in the response library before the next sample.
Use the response-time standards and the thirty-day review plan as the operating frame around the sample. Guidance on employment records from the Equal Employment Opportunity Commission supports keeping review notes consistent and retrievable. A routine sample, read against a written standard, gives the assistant useful direction and gives the recruiting team a defensible view of outreach quality. The routine should be simple enough to survive a busy week.