Self-Funding Investment to Reduce the Human Error Tax

hand holding tablet pen pointing at tablet screen with graphs
Insights 10 min read

The days when a three-year payback period was acceptable are long gone. Today’s leaders are expected to make strategic decisions, in less time, with faster returns. This is especially true when it comes to digital investments: none are cheap, so the faster an initial investment can pay for itself, the easier it is for companies to jump in.

Most companies (and regulators) agree that we should do whatever we can to reduce the Human Error Tax, the expected cost of human error paid by companies in the form of money and time. After all, who wouldn’t want fewer errors, higher profits, better customer experience, and greater operational resilience? But getting there requires technological investment; and companies now demand a quick return.

But what if we made it possible to reduce the Human Error Tax by self-funding, rather than incurring incremental cost with an undefined return timeline?

Predicting Error


We have tended in the past to react to human errors after they happen. But today’s tighter margins, focus on efficiency, and growing regulatory scrutiny make a strong case for solutions that can help prevent human errors in the first place.

What if we could reduce the Human Error Tax by self-funding, rather than incurring incremental cost with an undefined return timeline?

To do this, we must first predict when an error is likely and then provide a specific prompt to human users, to offset the behavior (input error, missed execution, or miscommunication). For example, to prevent input errors, we may simply prompt the human to ‘slow your input’ on a specific day. The prompt must be given only when there is a high risk of error and with enough time for the human to act.

Machine learning has made this possible. We can train models to identify the factors that collectively contribute to higher likelihood of human error. Unlike linear regression, which finds that x is correlated with y, machine learning can determine that when a, b, c, d and e move in a certain way, then f and g are likely.

If we use this technology to target specific areas of the business where human error is most common, we can not only help to reduce the Human Error Tax, but we can help free up resources to tackle a more complex redesign of end-to-end processes.

Creating Capacity (Money and Time)


Let’s say that human errors cause a business to lose $10 million annually and that employees will spend thousands of hours responding to those errors (rebooking trades; investigating and resolving payment or reporting errors; communicating with clients and regulators; and reporting in committee meetings). Estimating that one FTE is equivalent to 2,000 hours a year, it’s easy to see how the opportunity cost of human error adds up.

In this example, if a business invests $1 million in technology to reduce human errors (and their resultant losses) by 10%, it will have self-funded the investment in 12-18 months. The time saved in stopping just 10% of human errors can also create capacity in employee time.

This becomes the new steady state: preventing human errors and reducing time spent responding to them, the beginning of an upward spiral of self-funding and capacity creation.

The new steady state: preventing human errors and reducing time spent responding to them, the beginning of an upward spiral of self-funding and capacity creation.

Turn the Table on Operational Loss

Interested in learning more? Call or complete the form below.

Please provide full name
Please provide complete email
Please provide role
Please provide company
Please provide phone number

The Value of Momentum


In our example, reducing losses by just 10% can not only offset the cost of the technology, but also create capacity in time. This can free up an employee to organize the next segment of efficiency work, which could include business processes and alignment on areas where there is real versus perceived risk.

Let’s look at some specific examples. In tranche 1, a 10% reduction in human error leads to a $1.0MM reduction in losses and the equivalent of 1 FTE capacity of time. With only a 10% reduction in losses due to human error and the resulting 1 person freed up, a leader can create momentum that leads to self-funding lasting change.

Tranche Self-funded by Capacity created
  1. Use machine learning to predict and help prevent $1.0MM of loss
$1.0MM reduction in losses Creates capacity in time (FTE 1) to create an inventory of opportunities for efficiency
  1. Inventory and execute opportunities to stop doing or reduce non-value-add activity (see table 2)
FTE 1 (2,000 hrs) released by not having to respond to the 10% of human errors reduced in Tranche 1 Creates capacity in time (FTE 2 & 3) for process/control mapping & targeted process enhancements
  1. Inventory and map key processes/controls; identify targeted enhancements or automation
Reuse FTE 1 and add FTE 2 & 3
  • Enhance processes where there is real risk by using intelligence from Tranche 1 & process maps
  • Align risk management activity with real risk
    • Adjust risk management activity for low-risk areas using intelligence from Tranche 1 & 2
    • Design/execute targeted process enhancements based on risk
  • Key stakeholders agree to reduce activity, monitor impact
  • Reuse FTE 1, 2, and 3 to design/execute process enhancements
  • Creates additional capacity across lines of defense by reducing/eliminating low-value risk mgmt. activity
  • May further reduce human error through process re-design and can self-fund resources required for automation

The table below sets forth an example of the inventory described in Tranche 2 with common activity that can be adjusted or eliminated because it is not required, is inefficient, or doesn’t make us safer.

Table: Inventory of Opportunities for Efficiency

Activity Reason for adjusting
Committee mtgs/Forums to discuss operational risk framework activity (events, risk assessments, KRIs, scenarios)
  • participants have already been informed
  • more than required to demonstrate oversight
  • too many participants who don't participate
  • box ticking exercise
PowerPoint decks and materials required for committee mtgs/forums
  • industry of people to gather info for materials
  • backward looking (for previous quarter)
  • have always used this format/material
  • not prescribed by regulators
Reporting about events (at the time of, monthly, quarterly) and entry of information into risk systems
  • inefficient, duplication
  • not differentiated by level of risk
  • have always done it
Emails about operational risk framework activities (risk assessments, KRIs, events)
  • backward looking, inefficient
  • not differentiated by level of risk
Approvals and minutes – number of people in a business or corporate area required to approve risk assessments, events, KRIs
  • approvals by too many people for too many things
  • inefficient – manual approval and minutes
  • minutes not required for all meetings
Effective challenge – done for everything by multiple independent parties
  • may be a box tick, sometimes nothing to challenge
  • inefficient and duplicative

OLI - Predicting and Preventing Human Error


OLI is BMO’s Operational Loss Intelligence solution, used to predict when operational losses, often caused by human error, are likely so that clients can take action to prevent them.

Using machine learning, OLI combines a company’s internal data with external data to identify the likelihood of a human error resulting in a loss. This likelihood is based on patterns that the model has identified about the internal and external data. Once the likelihood of an error reaches a threshold, OLI alerts humans and prompts them to take specific action to prevent the loss.

For example, in a trading business, OLI may predict 75% likelihood of a human error on a desk on a given day, alerting the employee of the higher risk and prompting them to “slow your input”. Repeated high signal may prompt colleagues to review a process for adjustments that enable better resource management or automation.

The same approach can be applied to other areas that pay the Human Error Tax such as payments and operations functions, fraud departments, and third-party vendors to which companies have outsourced.

With the self-funding advantages that machine learning can offer in reducing operational losses from human error, the question isn’t “should I make the investment?” but rather “why would I wait?”

Technology investments deserve to be carefully evaluated to determine how suitable they are for an enterprise, and how quickly return will be generated. But with the self-funding advantages that machine learning can offer in reducing operational losses from human error, the question isn’t “should I make the investment?” but rather “why would I wait?”

Find out how to predict and help prevent operational loss

Call or complete the form below.

Please provide full name
Please provide complete email
Please provide role
Please provide company
Please provide phone number

About BMO

We’re proud to be fueling growth and expanding possibilities for individuals, families and businesses. More than 12 million customers counts on us for personal and commercial banking, wealth management and investment services. As the 8th largest bank, by assets, in North America, we help our customers make the most of their money. In Canada, the United States and across the globe, we’ll continue to build, invest and transform to drive performance that services the good that grows.