Table of contents
Most closing delays look obvious after the fact. A missing signature, a late payoff, or an unresolved condition appears in the final email and everyone wonders why no one saw it. The more useful question is what the team could see before the closing date. A delay that was visible early is a management problem, while a delay that appeared on the last day may be a genuine surprise.
This protocol measures how much of a closing delay was visible before the scheduled date. It defines leading indicators, sets a warning threshold, and compares the warning date with the actual delay. The result shows whether the team had time to act. It does not assume that every warning could have prevented the delay.
Question and scope
The study asks what share of delayed closings had at least one warning signal before the scheduled date. Scope names the team, the closing types, the period, and the systems used. A team might study purchase closings during a six-week window. The result describes that setting and should not be presented as an industry average.
Define a delayed closing before collecting data. A closing is delayed when it does not fund on the scheduled date, or when the date moves by at least one business day. State whether a same-day time change counts. A study that mixes definitions will produce a number nobody can reproduce.
Decide whether the unit is the file or the closing event. A file may have more than one scheduled date. Use the first scheduled date that the team communicated to the borrower. Later dates are outcomes, not baselines.
Choosing leading indicators
A leading indicator is a measurable fact that tends to appear before a delay. Candidates include an open condition past its due date, an unsigned disclosure, an unconfirmed wire instruction, a missing appraisal, or a title commitment exception. Choose the indicators the team can observe at least two business days before closing. Each indicator needs a threshold. A condition that is one day late may be routine, while a condition that is three days late may predict a delay. Set the threshold from the team's own history, not from a generic benchmark. Record the threshold and the reason beside the indicator.
Limit the set to five or six indicators. A long list produces warnings on every file and no useful signal. A short list that the team trusts is more likely to be used. Review the list after the first sample and remove indicators that never precede a delay.
Setting the warning date
The warning date is the first business day when an indicator crosses its threshold. Record the indicator, the threshold, and the date. If several indicators cross on the same day, record the earliest and list the rest. The warning date is the start of the team's opportunity to act.
Do not move the warning date after the closing. A retrospective warning date is not a measurement. Keep the original observation and note any later correction. The comparison between the warning date and the closing date is the core of the study. Account for weekends and holidays. A warning that appears on Friday before a Monday closing leaves one business day, not three. Convert the gap to business days so the result reflects the time the team actually had.
Sampling delayed and on-time closings
Sample both delayed and on-time closings. A study that only looks at delays cannot show whether the indicators also fire on files that close on time. The false-positive rate is part of the result. A warning that fires on every file is not a useful warning.
Use consecutive weeks and include the files that closed early. Early closings can reveal an indicator that is too conservative. They also keep the sample from being dominated by the hardest files. Record the closing type and the complexity drivers. If the team has few delays, extend the window rather than lowering the threshold. A small sample can produce a dramatic share that will not repeat. Report the count and the confidence range so the reader can judge the precision.
Coding the warning evidence
Record the source of the warning, such as the condition tracker, the closing disclosure log, or the title company update. A warning without a source is an opinion. The source should be available to a second reviewer. Keep the link or the record identifier with the entry.
Record who observed the warning and when. The observer matters because a warning that only one person saw may not have been actionable for the team. The time matters because a warning discovered at the end of the closing day is different from one discovered in the morning. Do not count a warning that was created after the delay was known. A warning must precede the closing date and the delay. A note added during the post-mortem is not evidence of early visibility.
Comparing warning and outcome
For each delayed closing, calculate the business days between the warning date and the scheduled date. Report the median and the distribution. A median of one day means the team had little time, while a median of five days means the warning was visible but not acted on. Compare the warning rate for delayed and on-time closings. A warning rate of eighty percent for delays and twenty percent for on-time files suggests the indicator has some signal. A warning rate of eighty percent for both suggests the indicator is too broad. Report both numbers. Pair the comparison with the action taken. A warning that was followed by a documented action and still ended in a delay tells a different story from a warning that nobody saw. The action log is the bridge between visibility and outcome.
What the result can change
A high visibility share with a low action rate points to an ownership or escalation problem. A low visibility share points to a missing indicator or a late data source. A high false-positive rate points to a threshold that needs adjustment. The study should name the likely cause before recommending a change.
Test one change, such as a daily closing readiness check or a new warning threshold, and rerun the study. The comparison should use the same indicators and the same sampling rule. A change that only improves the documentation of warnings without improving the outcome is not enough. The mortgage operations support lead can own the closing readiness review and the warning log. The loan processing support team can prepare the evidence for each indicator. The split keeps the observation separate from the decision.
Limits and sources of error
The study measures visible signals, not causation. A warning may appear on a file that would have closed late for an unrelated reason. The result describes the opportunity to act, not the guarantee of a better outcome. Report it with that limitation.
The team may notice a warning but not record it. A missing record looks like a late warning. Add a short daily prompt that asks for any new closing risk. The prompt improves coverage without turning the log into a second job.
The sample may exclude files that were canceled or withdrawn. Those files can carry warnings that never became delays. State the exclusion and the reason. A study that silently drops canceled files may overstate the warning rate for delays.
Review cadence
Repeat the study after a change to the closing checklist, the title partner, or the condition process. Use the same indicators and the same window length. A visibility share that rises while delays stay flat may mean the team is documenting more but acting less. The action log should be part of the review.
Share the result with the closing team and the title partner. Ask which warnings were easy to act on and which ones sat without an owner. The discussion can reveal a handoff gap that no indicator captures. The next study should test one of those findings. End with a sentence that states the visibility share, the median warning lead time, and the period. The Consumer Financial Protection Bureau's closing resources describe the borrower's review period, which is one constraint on the closing schedule. The goal is not to predict every delay, but to give the team enough time to protect the date.