The Case of the Disturbing MBO Business Processes

I recently finished writing a management case study, the subject of which is our firm’s mapping of the current state of the application-to-investor sale process for the mortgage banking operation of a community bank.

The bank is a client of ours, and the case study stems from our work performed on their behalf; the focus of the study is on the current state of that process, so it does not delve into the solutions from the subsequent redesign effort (in any event, that information would be proprietary).

I wrote the case study, because it strikes me, that, although this bank’s situation is hardly unique in terms of the current state of its processes, its situation will conger disturbing thoughts about the condition of business processes in any enterprise.

As with most case studies, it is a bit lengthy for casual reading. But – it is a worthwhile read if you want to understand how good process analysis works, understand where good process redesign and improvement starts, and understand the financial cost of all the waste, uncertainty, and variation found in the design of the workflow of the processes that most of us have to use, every day.

You can download the full case study at, but here is some of what it has to say:

These comments on the financial cost of process variation:

  • Almost every loan application that was opened (93%) was sent to Underwriting, but only about one-third (38%) ever made it to the Closing stage. Underwriting wasted about half of its current capacity. Processing wasted about one-third of its capacity. That is a significant amount of wasted capacity.
  • The financial cost associated with that waste, whether viewed as the lost Contribution (Gross Income) from the sales of loans that the operation could have closed (if it wasn’t consuming its time and effort on loan applications that cancelled), or as the cost of the production capacity (in the form of overhead) that was not utilized (and, therefore, an unnecessary expense).
  • The annual payroll, advertising, and administrative/processing expense for the mortgage banking operation was projected at about $5.4 million per year. If the case was made for, say, a 12% allowance related to unavoidable cancellations [i.e., an 88% pass-through rate], that would only push the conversion rate from the 38% range to the 50% range [88% – 38% = 50%]. There were only two possible conclusions: Going forward, either the mortgage banking operation would have to live with $2.7 million in excess, unused capacity – capacity that it didn’t need – or, it would face the prospect of giving up the Gross Income from the closings that it failed to get through the system.
  • During 2010, there was a total of 1,193 loans submitted to Underwriting. If unavoidable cancellations were 12% of the loans that made it to Underwriting, the operation would have still closed 1,050 loans. But, there were only 450 actual closings. The difference [600 closings] corresponded to the additional 50% cancellation rate that was avoidable – therefore, preventable – and, therefore, either wasted capacity or lost throughput. Using the [bank’s] forecast assumptions of $185,000 average loan, and an average Gross Margin of 3.2%, the 600 lost closings had cost the bank about $3.6 million in Gross Income that year.
  • There were a range of issues that had caused loan applications to cancel, but a percentage of any cancellation rate is preventable, and the cost of preventable cancellations, whether measured in the cost of wasted loan production capacity or in lost income, is likewise preventable.

These observations on process lead time:

  • In the engagement letter for this project, there was “an expressed determination that the cycle time for the Application-to-Close process should not exceed 25 business days”. That was the only measure of performance for the redesigned process that was discussed. That duration of lead time was a target, not an indication of the duration of lead time for the current process.
  • We looked at the 2010 operating data, and determined what the level of applications-in-process would have been, with the reported throughput, under the assumption of achieving the targeted 25 day lead time. The lead time – for any process – can be calculated according to the following formula [universally known in production physics as Little’s Law]:  Lead Time = (Units-in Process/Units Completed) x Daysn.
  • Presuming that every loan application passed to Closing eventually closes, there were 450 closings throughout 2010. Therefore, if n = 360:

(X/450) x 360 = 25

X = (25 x 450)/360

X = 31

  • The indicated inventory (applications-in-process) that the process should have been working on to produce 450 annual loan closings with a lead time of 25 days was 31 applications-in-process. However, there was an average of 299 applications being acted on – being passed between different stages of the process – at any point in time during 2010.
  • It was impossible, therefore, for the current process to be achieving the targeted lead time of 25 days.

To be fair, the level of average inventory in 2010 wasn’t likely the 299 applications-in-process indicated, either, because that would have meant the calculated lead time for an application going through this process would be almost 240 days.

If there was, as indicated by the operating data, a progressive decline in the number of actions being taken as applications moved downstream in the process, then a very rough average level of inventory for the overall process could be estimated from the average number of actions that occurred at the front-end and the back-end of the process.

In 2010, there was an average of 107 opening actions each month. There was an average of 38 closing actions. The averages were 72 actions and 450 closings:  (72/450) x 360 = 58 days

Then again, calculated as the monthly lead time based on the average number of loan closings indicated each month (38 closings), a 60 day lead time would have had 76 applications-in-process; a 90 day lead time would have had 114 applications-in-process.

What was the true level of applications-in-process? Who knows? The truth was, the system so distorted the metrics, that it could not even determine the number of applications-in-process.

Then, there were the following comments on the characteristics of the process associated with little or no added value:

  • In a process that is intended to take only 25 days to complete, there are at least 39 instances when the work of one person (or department) is handed off to another person or department.
  • There were 29 instances when the completed work of one person or department was subjected to the review, inspection, or approval of a different person or department.
  • Characteristic of the process was the iterative, piecemeal review approach to gathering additional documentation, looking for missing documentation, and meeting other requirements. The process constantly – repeatedly – reached back and checked to be sure that it had everything, checked to make sure everything was right.
  • It became an excuse for not getting the work completed and done right the first time;
  • It promoted multi-tasking, because applications were never really completed.
  • It prevented major rework, but it slowed the process.

Because it slowed the process, it also reduced throughput, lengthened lead time, delayed closings, kept too many applications-in-process, and consumed resources.

Finally, this set of observations, culled from the case study’s epilogue:

The analysis of the current state included the time-consuming task of flowcharting the actual workflow, which confirmed that the only existing documentation of the process was a transcribed outline that the senior-most manager in the mortgage banking operation had sketched on a large erasable board in one of the break rooms.

This was a community bank, but it was not a small banking operation. Yet – all this mortgage banking operation had to show for its existing process was an outline of the management-prescribed workflow that SAI had to transcribe from a whiteboard.

There was never any relevant, reliable operating data available from the process, at any time during the AS-IS.

Following the dissemination of the AS-IS Report, as the project was reconvened to begin the SHOULD-BE element, the first order of business the team undertook was to reverse the illogical order of the process, in which loan applications were completely underwritten before the applications were ever processed.

The revealing part, was that this action flowed from the insight gained during the AS-IS and was arrived at completely ad hoc, before the formal redesign work began.

The team members had taken it upon themselves, in informal, unstructured conversations and dealings, to consider elements of logical process design; they walked into the SHOULD-BE session, prepared to act on what they had seen during the AS-IS.

After all, it was the team members who performed the actual process work, not the managers. The team members could see the problems. They knew – intuitively, instinctively – that loans couldn’t be underwritten before the application was even processed. They could see the stakeholder benefits – to the borrower, the investor, the bank, and to themselves – of having a redesigned and improved process.

The team members understood the benefits of doing more and better with the same or less – more loans funded and sold, more revenue, better results, produced with the same or fewer resources, in less time, with fewer errors, with less pointless effort, with greater satisfaction and less frustration.