The executive gap, and how to price it

Ask a chief operating officer how long month-end close takes and you will get an answer in hours. Ask the four people who actually run it and you will get an answer in days. Both are given in good faith. Both are wrong in a specific, measurable direction, and the size of the difference is the most useful number an audit produces.

Why the two answers diverge

Leadership describes the process as designed. The frontline describes the process as executed. Between those two descriptions sit all the things nobody writes down:

• The rework loop that fires on roughly one in four items
• The handoff that waits because a single approver is the only one with access
• The step that exists purely to correct an earlier step
• The spreadsheet that reconciles two systems that were supposed to be integrated

None of these appear on a process map. All of them appear in an interview, as soon as you ask a person to walk through the last time they did the work rather than describe how the work is supposed to go.

The gap is a quantity, not an anecdote

The temptation is to treat the discrepancy as a communication problem and average the two figures. That destroys the finding. The frontline record is the measurement; the leadership estimate is the belief. What matters is the difference between them, multiplied by the people it affects.

-- The cost of one executive gap, per finding
select
  p.name                                   as process,
  f.frontline_hours - m.manager_hours      as gap_hours_weekly,
  (f.frontline_hours - m.manager_hours)
    * 48                                   as gap_hours_annual,
  (f.frontline_hours - m.manager_hours)
    * 48
    * o.avg_hourly_rate
    * f.affected_headcount                 as annual_cost_of_gap
from frontline_record f
join manager_estimate m using (process_id)
join processes p        using (process_id)
join organizations o    on o.id = p.org_id
where f.frontline_hours > m.manager_hours
order by annual_cost_of_gap desc;

Run that across every process and the output is a ranked list of the places where the operating picture is wrong, ordered by what being wrong costs. Note the final multiplier: affected headcount. A two hour weekly discrepancy in a task performed by one person is 96 hours a year. The same discrepancy across fourteen people is 1,344 hours, which is most of a full-time role that nobody has ever budgeted for.

A worked example

Take a month-end close described by leadership as a three day process, and suppose the interview record puts the median at six and a half days. The figures below are illustrative, chosen to show the shape a finding takes rather than to report a particular company.

The interesting row is the last one. Leadership had no estimate for the longest wait because nobody had ever been asked to measure a wait. Waiting is invisible in every system of record a company owns. It shows up in a calendar as nothing at all.

A process that takes six days and one hour of work is not a slow process. It is a fast process with five days of queue attached to it.

The practical consequence

When the gap is measured first, the automation decision changes. In the example above the obvious project is document extraction, because re-keying is the most visible pain. The ranked finding is the approval queue, which accounts for more than half the cycle and which no software had been proposed for.

Extracting documents faster addresses five steps of manual re-keying. Routing approvals differently addresses two days and four hours. The first is an AI project. The second is a routing rule and an afternoon.

Measuring first is what tells you which one you are looking at.