Auditing duplicate coach outreach

Published November 2, 2026

A repeatable method for finding duplicate recruiting messages across family, athlete, and support-staff channels without inflating the count or blaming a single sender. The protocol below keeps the observation window, the unit, and the evidence standard in one place. It is written for a recruiting team that may hand the record to another reviewer. Use it to review one routine case and one exception before changing the workflow.

Question and scope

This audit asks how often the same program receives more than one outreach message about the same athlete within a defined window, and where the duplicate originated. It looks at the message record and the sending channel. It does not measure whether a coach minded the duplicate, whether the athlete was a good fit, or which staff member is responsible. The purpose is to find a routing gap that can be fixed.

Set the window before counting. A duplicate sent within the same day may be a coordination error; a second message six months later may be a legitimate reconnection after new film or a changed roster need. The study should define the window that matches the outreach cadence it wants to review, and it should record the date of every message in the thread. A moving window will make the count impossible to reproduce.

Population and unit

The unit is one program-athlete pair within the observation window, not one message. A pair with three messages is one affected pair with two duplicate events. Record the first contact, every later contact, the channel, the sender role, and the response state. This structure prevents a single enthusiastic athlete from looking like a systematic duplicate problem across the whole program.

Decide whether different sports at the same school count as separate programs. In most cases they should, because the coaching staffs and recruiting calendars can be separate. Decide how to treat a general admissions address that forwards to several coaches. Record the routing rule and the observed destination rather than assuming every message reached the same person. Exclude messages that are clearly part of the same exchange, such as a reply to a coach's question or a requested follow-up document. A duplicate is an unsolicited repeat of the same outreach intent, not every message in a conversation. The coding rule should include examples so two reviewers can classify a thread the same way.

Duplicate definition

Define a duplicate as a second unsolicited outreach to the same program-athlete pair within the window that repeats the same core request without a material update. Core requests include asking the coach to review film, requesting a call, announcing a visit date, or sharing a profile. A new season of film, a corrected academic document, or a response to a coach's request is a material update and should not be counted as a duplicate. Record the severity of the repeat. A near-identical message sent from two addresses on the same day is different from a second introduction sent a month later because the first one was forgotten. A severity scale can be simple: same-day duplicate, short-window duplicate, and long-window repeat. The scale should guide the fix, not become a score for the sender. Keep a separate field for the cause: no shared log, multiple senders, a family forwarding a draft, a platform retry, or an address book that contains two contacts for the same coach. The cause often determines the remedy. Adding a duplicate check to the log will not fix a platform that sends the same campaign twice.

Data sources and linkage

Collect the sent folder, the recruiting-platform message history, the shared contact log, and any documented phone or text outreach. Use the school and program as the first join key, then the coach name or role, then the athlete identifier. Do not rely only on email addresses, because the same coach may use a program address and a personal address that the family has recorded separately. Record the message identifier, send time, sender account, subject or first line, and the destination address. If the message was sent through a platform that masks the coach address, record the platform thread identifier instead. Keep the raw message in the approved system and put only the minimum reference in the audit table.

Build the linkage before counting. A duplicate audit is only as good as its ability to recognize that two rows refer to the same program. If the team cannot link a general athletics address to a specific coach, mark the relationship as uncertain rather than forcing a match. That uncertainty belongs in the result.

Sampling and collection

Export the message frame for the window and freeze it. Sample program-athlete pairs, not individual messages, then review every message in the selected pair. This two-stage approach prevents a small number of very active families from dominating the result. Stratify by sender role or outreach batch if the workflow has distinct paths.

Use a random or systematic start and record the rule. If the population is small, review all pairs. A complete census is easier to explain than a sample when the team wants to fix a specific workflow. Include pairs with no response, because a duplicate can occur before any coach replies.

Have a second reviewer recode a small subset. Compare the duplicate classifications and the cause codes. If disagreements cluster around a particular kind of message, clarify the definition before continuing. The audit should produce a reliable count, not a debate about whether a reminder counts as a duplicate.

Analysis and reporting

Report the number of pairs reviewed, the number with at least one duplicate, the number of duplicate events, and the share by cause and channel. Show the count of same-day versus longer-window duplicates. A single rate without the event count can hide a few pairs with many repeats, so report both. Inspect the pairs with the most events before proposing a fix. A family that sent the same introduction from three addresses may need a clearer single-sender rule. A batch with the same message from two staff accounts may need a platform setting or an ownership change. The pattern should point to one workflow gap, not a general instruction to be more careful.

Do not rank individual senders in the report. The audit is about the system that allowed the duplicate. If the team needs individual coaching, use a separate private review with context. A public duplicate league table discourages people from recording their messages accurately, which makes the next audit worse.

Limitations

The audit can see only the channels in the frame. A coach may receive a duplicate through a club coach, a phone call, or a camp conversation that never appears in the export. The count is therefore a lower bound for the selected channels, not a complete measure of every contact. State the unobserved channels and the likely effect.

Message records may be incomplete after a platform migration, a mailbox cleanup, or a staff departure. A missing sent item can make a duplicate invisible or make one message look like two if it was forwarded. Preserve the original export and note the retention limit. Do not reconstruct missing messages from memory without labeling them as later recollections.

The definition of a material update involves judgment. A new highlight clip may be material to one coach and redundant to another. The study should provide examples and report coding disagreements. The result describes this team's outreach practice; it is not a universal standard for how often a family may contact a program.

Review and next action

Choose one fix tied to the dominant cause: a single sender per program, a duplicate check before each batch, a cleaner contact list, or a platform configuration change. Name the owner and the date the change begins. Decide how the team will know the fix worked, such as fewer same-day duplicates in the next four batches. Repeat the audit after the change with the same window and definitions. Compare the event count and the cause mix, not just the headline rate. If duplicates persist, inspect whether the new rule is being followed or whether the underlying data still contains two identities for one program. End with a decision about whether to keep, adjust, or stop the change.