Research note
What share of onboarding delays trace to missing documents?
A bounded method for attributing onboarding delay to missing documents without overstating the cause.
Published

Table of contents
Question and scope
This study asks what share of onboarding delay is attributable to a missing or unusable document. The scope is one employer or one hiring group, one onboarding process, and a fixed period. The result describes where time was lost in that setting. It is not an estimate of delay across an industry or a judgment about any candidate.
The unit is one new hire whose onboarding started during the period. Delay is the difference between the planned ready date and the date the packet was complete enough for the first scheduled workday. Record the planned date from the approved offer, not from a later recollection. Keep the offer date and the actual start date in separate fields.
Separate a missing document from a late document, an invalid document, and a document that was never requested. These categories lead to different fixes. A missing item may be a candidate communication problem. An invalid item may be a form or instruction problem. A never-requested item is a process design problem. Keep a short case note for each delayed hire. The note should name the stage, the missing item, the request date, and the next action. Do not include unnecessary personal detail. The note is a working aid for the review and a source for the final category.
Definitions and evidence standard
A document is missing when the packet requires it and no acceptable version exists at the review point. A document is late when it arrives after the agreed request date. A document is invalid when it is present but does not meet the stated requirement, such as an expired identity document or an unsigned form. Record the reviewer and the date for each classification.
Use the onboarding checklist as the requirement list, but verify that the checklist matches the current process. A stale checklist can create a false missing item. Record the version of the checklist used for the review and note any item that the team handles outside the checklist. If the request was sent through a portal, check the delivery status as well as the timestamp. A message that never reached the candidate is not the same as a candidate who ignored it. Capture the channel and the address used. The distinction changes the repair.
The evidence standard is a dated request, a dated receipt, and a recorded review outcome. If the request was verbal, record the date and the person who made it. If the candidate submitted through a portal, use the portal timestamp. A delay attributed to a missing document without a request date is only a guess.
Sampling and measurement
Sample new hires from a frozen start list. Include hires who started on time and hires who were delayed. Do not sample only the delayed cases, because the base rate matters. Stratify by role, location, and whether the hire was remote or on site. Record the source list and the date it was frozen.
For each hire, build a short timeline: offer accepted, packet opened, each document requested, each document received, each review completed, and first day worked. Mark the first point at which the packet could not move without an external input. That point is the delay start for the attribution. Use the same timeline template for every case so the comparison is consistent.
Classify the delay into non-exclusive categories: missing document, invalid document, late approval, system outage, scheduling conflict, and candidate unresponsive. A single delay may have more than one cause. Record the primary cause and any secondary cause, and report both. Keep the categories visible in the working file so the reviewer applies the same set to every case.
Have a second reviewer check a sample of attributions. The reviewer should see the timeline and the checklist, not the original classifier's conclusion. Compare the two results and resolve differences with a written rule. Keep the rule with the dataset so a later review can repeat it. If the sample is small, report the count rather than only a percentage. Five delayed hires out of twelve is easier to interpret than a rounded rate. Add the number of delay days so a single long case does not disappear. The reader should be able to see both the frequency and the burden.
Analysis and interpretation
Report the share of hires with any delay, the median delay days, and the share of delay days attributed to each category. A share of hires is different from a share of days. A small number of hires with long delays can dominate the total days even when most hires are on time. Show the count of hires and the count of delay days in the same table.
Show the result by document type. A cluster around identity verification may point to a communication or scheduling problem. A cluster around tax forms may point to a confusing portal. The category is a starting point for review, not a proven cause. Report the count behind each cluster so a small group is not overread.
Inspect the cases where the candidate was marked unresponsive. Check whether the request reached the right address, whether the instructions were clear, and whether the candidate had a way to ask a question. A label that blames the candidate can hide a broken request process. Check the request log before accepting the unresponsive label.
Do not compare this result with another employer unless the checklist, the start-date rule, and the attribution method are the same. Onboarding definitions vary widely, and a difference in the definition can look like a difference in performance. Describe the local definitions in the methods note so a reader can see why the result is not comparable. Keep that note with the dataset.
A useful second view groups delays by the owner who was waiting at the start of the delay. That view shows whether the wait sat with the candidate, the manager, the payroll team, or the onboarding coordinator. It is not a blame chart. It is a map of where the next action was blocked.
Limitations and governance
The study cannot observe everything that happened outside the recorded process. A candidate may have had a document ready but sent it to the wrong contact. The review may classify that as missing because the packet did not receive it. The limitation is about evidence in the packet, not about the candidate's effort.
The attribution rule is a judgment. Two reasonable reviewers may disagree about whether a delay started when the document was requested or when the first reminder was sent. Calibration and a written rule reduce the variation, but they do not eliminate it. Have both reviewers write the reason for a disputed attribution.
Protect candidate data throughout the review. Use a study identifier and keep the mapping in a restricted file. Do not include identity document numbers or health information in the analysis notes. Delete temporary copies when the review is complete and follow the organization's retention rule. Write the attribution rule in plain language and attach it to the dataset. A future reviewer should be able to apply the same rule without asking the original author. If the rule changes, keep the old version and note the date. A silent rule change makes the trend unreadable. Name the retention period and the owner who will delete the working file.
Practical use
Pick the document type with the largest attributable delay and the clearest repair. A pre-start checklist with examples may reduce invalid submissions. A named onboarding contact may reduce unresponsive periods. Test one change and measure the same categories again. Keep the change small enough to test within one hiring cohort.
Report the result to the people who control the request, not only to the onboarding team. A late approval or an unclear offer term may sit outside the team that is being measured. The study is most useful when it shows the full path a new hire must travel. Ask each owner to confirm the stage where their action was waiting.
The goal is not to assign blame for a late start. It is to find the point where a routine document request turns into an avoidable wait. That finding gives the next onboarding cohort a better chance of starting on time. Review the delayed cases with the onboarding owner before choosing a fix. The data shows where the wait occurred, but the owner knows which request or form caused it. Ask for one change that would remove the most common avoidable wait. Then measure the same categories in the next cohort.
Sources
References supplied with this research note.
| Source | Publisher | Context |
|---|---|---|
| U.S. Citizenship and Immigration Services, Form I-9 | uscis.gov | Primary reference supplied with this article. |
| U.S. Department of Labor, New Hire Reporting | dol.gov | Primary reference supplied with this article. |
| U.S. Equal Employment Opportunity Commission | eeoc.gov | Primary reference supplied with this article. |