.svg){kind=icon}
.svg){kind=icon}
Operational Mortgage Lending Metrics
Framework to quantify the performance of mortgage operations
At this point, the concept of a metric is common knowledge among operations leaders.
Metrics are critical tools for tracking performance and facilitating continuous improvement.
However, it can still be a challenge to consistently find practical applications of metrics in mortgage operations that drive tangible results.
To make it easier to identify, define, and measure metrics in mortgage operations, I set out to break down what metrics are and how they are applied to operations.
Below, you’ll find my analysis of:
If I’ve missed anything or you have anything to add, please let me know, and I’ll update the post.
Before we get into the specifics of operational metrics, let’s define what a metric is.
A metric is a quantifiable, standardized measurement derived from recorded data.
At first glance, this definition might be hard to fully grasp, so let’s break down what each part means:
The definition of metrics is often closely associated with analytics software and business operations.
However, the concept of metrics is not unique to any of them.
Metrics are widely used in business, science, health, and other areas where performance, outcomes, or conditions must be measured and analyzed.
And metrics are technology agnostic; theoretically, they can be defined and measured using nothing more than a pen and paper.
Here are a few examples of metrics:
The problem metrics solve is the difficulty in assessing and comparing facts without a standard scale.
Metrics are essential for analysis and decision-making because standardized and quantifiable measurements enable objective assessment and comparison by placing facts on the same scale.
The terms "metric" and "KPI" (Key Performance Indicator) are often used interchangeably in business and analytics. Still, they have distinct meanings and serve different purposes.
Let's draw a line between a metric and a KPI for the sake of clarity in this article.
First, every KPI is a metric, but not every metric is a KPI.
A KPI is a metric that represents progress towards a specific goal.
In other words, KPIs are metrics that indicate whether you're getting closer to or further from your goal.
For example, if the goal is to increase loan origination profitability, then KPIs might include:
KPIs are goal-specific. As company goals change, so do the KPIs.
Meanwhile, operational metrics are function-specific. As long as the function exists in the company, the metric will exist too.
A metric prototype is a blueprint that defines how a specific metric should be measured. It’s essential for ensuring consistent and reliable measurements.
A single metric can be described with four components:
For example, the metric $ Total Production Volume (per Loan Officer per Month) consists of:
LoansSum of Loan AmountLoan OfficerLoan Closed DateLoan Status is ClosedBelow is a more detailed overview of each component of a metric prototype.
A metric dataset defines what data should be measured by the metric and where to get it.
Without data, there are no metrics.
A dataset is a collection of data organized for a specific purpose.
In the context of metric datasets, this typically consists of structured data where each row represents a unique record, and each column represents a specific attribute.
Metric datasets are usually defined by the common attributes of the records within the dataset.
In the example above, the $ Total Production Volume (per Loan Officer per Month) metric measures records within the Loans dataset, which can be extracted from a LOS.
Here are more examples of datasets:
You can learn more about operational mortgage data and data types here.
A metric measure defines how to quantify the records within a metric dataset.
It's expressed through quantitative operations needed to translate raw data into a numeric value, such as:
In the example above, the $ Total Production Volume (per Loan Officer per Month) metric measures Volume by calculating the Sum of the Loan Amount.
Here are more examples of measures:
Metric dimensions define how to categorize records within a metric dataset into segments before measurement.
Categorization of the dataset enables the assessment and comparison of the metric's value across different segments.
Seeing the value of metrics across different segments adds depth and meaning by providing context for numerical measures, such as geographical location, product types, customer segments, or periods.
A single metric can be broken down across multiple dimensions at once.
In the example above, the $ Total Production Volume (per Loan Officer per Month) metric categorizes the dataset across two dimensions:
Loan OfficerLoan Closed DateDimensions are usually defined by the qualitative attributes of a record within the dataset, such as:
By analyzing multiple dimensions, you can gain insights into correlations between metric values across dimensions, such as the product mix that each loan officer produces.
A metric filter defines which records within a metric dataset should be measured.
Filtering the dataset narrows the scope of the measurement to obtain a desired data point. A filter can be defined by a set of rules that a record must satisfy to be included in the measurement's scope.
Each rule is usually defined by the following:
In the example above, the $ Total Production Volume (per Loan Officer per Month) metric measures only records that meet the filter rule below:
Loan statusEqual toClosedA single metric can apply multiple filter rules to narrow the dataset to a desired scope.
One of the most common filter rules in metrics defines the timeframe of the metric, e.g., last month.
Here are more examples of filter rules:
loan.status = Closedbranch = Branch 2closed time > Jan 1st
For clarity, it helps to understand how a metric prototype becomes an actual metric (a numerical value).
This transformation occurs through metric measurement.
Metric measurement is the process of calculating the value of a metric based on its prototype.
Here's what the process looks like:
In our example, $ Total Production Volume (per Loan Officer per Month) would be measured following this process:
Loans.Closed.Loan Officer and Month.$ Total Production Volume for each segment.
Let's see what makes operational metrics different from other metrics.
Operational metrics involve the same four prototype components as any other metric.
What makes them different is what are these 4 components:
If, in most general terms, Metric is a quantifiable, standardized measurement of recorded data (recorded facts about the world)
Then, operational metrics are quantifiable measurements of data about a business's operational activities.
The sections below provide a deeper overview of what's unique about the Metric datasets, measures, and dimensions for operational metrics.
Usually, an operational metrics dataset consists of records describing facts about operations produced by a specific business function.
An operation is a sequence of actions undertaken to produce a desired output.
Most operations can be described by the following facts:
This framework applies to operations of any size, from the smallest ones like sending an email carried out by a single agent, to huge ones like loan originations that involve multiple departments and agents.
Here are a couple of examples of operations:
1. Manufacturing a Widget:
2. Processing a Bank Transaction:
The dataset of operational metrics is usually scoped to operations produced by a single business function. Each function has different inputs and outputs, and thus, it is impractical to measure operations like "Sending an email" and "Underwriting a loan" together since they have little in common.
In the section below, you can find out how data about operations is measured.
From the previous section, you've seen what data operational metrics quantify. Now, let's explore what's unique about the measures in operational metrics.
Operational metrics measures quantify the performance of the business function.
How you measure an operational dataset determines the insights you can gain about function performance.
Measures will be unique for each function, but the aspects of function performance they measure generally remain the same.
There are numerous ways to measure function performance. And since the more metrics you measure, the more noise you get, my rule of thumb is to:
Measure only performance aspects you want to improve or don't want to get worse (while you're working on what you want to improve)
Here are five types of operation measures that I find most useful for operational improvement:
Below is an overview of what each measure type entails.
This type of measure quantifies the volume of work product (operation output) produced by completed operations.
The volume of a work product can be defined as the total amount of output produced, considering not only the number of units but also the complexity, size, or scope of each unit.
For example, if the work product of a cleaning function is a cleaned apartment, then the total count of cleaned apartments would not provide an accurate volume measure. This is because apartments have varying square footage, and three 850-square-foot apartments represent less total area than two 1,500-square-foot apartments.
In this case, a more accurate measure of volume would be the total sum of the square feet cleaned.
Here are examples of work product volume measures:
This type of measure quantifies the quality of the work product (operation output) produced by completed operations.
Work product quality can be defined as how well the work product fulfills its purpose.
To quantify the aggregated quality of the work product, you first need to understand what "quality" means in your specific context. The definition of quality is unique for each work product type.
Once you know what quality entails, you need to identify which attributes within the dataset represent it. The quality of a single work product unit is usually represented by one of the following:
Once you understand which attributes of the records represent quality, you can calculate an average of numeric attributes or the percentage of binary attributes.
In most cases, the quality of a product (how well it fulfills its purpose) can only be assessed when it is consumed or used, whether simulated by QA or actually by the end user:
Since data about the quality of the product is usually only available after usage, you need to build a feedback loop to collect this data to measure quality metrics.
Here are examples of work product quality measures include:
This type of metric measures the financial aspect of producing a unit of work product (operation output).
The financial aspects of the work product represent the cost, revenue, profit, and margin per unit of the work product.
It is measured by dividing total revenue/expenses by the total number of work product units produced over the same period. Profit and margin are the difference between these values, expressed as absolute ($) or relative (%).
Here are examples of the work product financial measures:
This type of measure quantifies the efficiency of a function in transforming operation input into output.
Efficiency can be defined as the ratio between input and output within a specific timeframe. The less input (time, resources, and materials) it takes to produce a unit of output, the higher the efficiency.
Since input and output are not always easy to quantify, below are common proxies that quantify efficiency through:
Cycle time is typically measured by calculating the average time from the operation's start to finish. The completion rate can be calculated by comparing the percentage of completed operations to those that started. The step rework rate is measured by counting the percentage of operations that required more than 1 touch of the same step compared to all completed operations.
Here are examples of operation efficiency measures:
This type of measure quantifies the compliance of completed operations with predefined standards.
Compliance can be defined as the extent to which an operation is carried out according to standards set by the company, market, or regulators.
Operation compliance focuses on the operation and how it was carried out rather than the work product. For example, it examines how long it took, what steps were taken, and in what order.
Operation compliance is usually binary (an operation is either compliant or not), so it is calculated by finding the percentage of completed operations that meet the standard compared to the total number of operations.
Like work product quality, operation compliance can be assessed after an operation. Therefore, you’ll need a method to record what was (and was not) done and in what order and have a feedback loop to collect this data.
Here are examples of operation compliance measures:
The attributes of the metrics dataset define the dimensions of the metric.
So, the operational metrics dataset's attributes define the operational metrics' dimensions.
Even though attributes in the operational metrics dataset will be unique for each business function, records within the dataset follow the same pattern:
Below is an overview of the common operation metrics dimensions and the insight you can gain by applying them to different measure types.
Each section looks at a dimension type in isolation, but you can stack multiple dimensions to create the metric you're looking for.
Applying different dimensions to the same measure gives insight into the various aspects of function performance.
Operation time dimensions break down measures by date and time attributes of the operations.
Operation time attributes describe when an operation started, was completed, or any other timestamped event that happened in between.
Here are examples of operation time attributes:
Operation time dimensions allow you to see the metric's value for each period, enabling comparison of metric values across different periods, such as June versus August.
Metrics broken down by operation time dimensions provide insights into how the metric's value has changed over time, which helps identify patterns and trends.
Here's what insights you can gain by applying the time dimension to each measure type:
The operation time dimension is most commonly stacked with all other dimensions.
Operation agent dimensions break down measures by the operation agent attributes.
Operation agent attributes describe who’s carrying out an operation and their traits.
Here are examples of the operation agent attributes:
Operation agent dimensions are useful for assessing the performance of the employees in specific aspects of the operation (e.g., quality, volume, compliance) to identify team members who need additional training or aren’t a good fit for a company.
Here’s what insights you can gain by applying the operation agent dimension to each operation type:
Operation output dimensions break down the measure by the work product (operation output) attributes.
Work product attributes describe the final work product produced due to operation.
Here are examples of the operation output attributes:
Operation output dimensions give insight into the work product mix produced by the operation and the correlation between the type of work product produced and the value of the specific aspects of the operation (quality, volume, etc.).
Here's what insights you can gain by applying operation output dimension to each operation type:
Operation input dimensions break down measures by the attributes of the operation input.
Operation input attributes usually describe the demand for the work product.
Here are examples of the operation input attributes:
Operation input dimensions give insight into how much and what input is involved, as well as whether there is a correlation between operation input and work product quality, cost, or operation compliance.
It is handy to identify specific input types that produce superior results and then either adjust the input mix to get more off or improve the function/agent to produce better results with underperforming input types.
Operation output dimensions give insight into the work product mix produced by the operation and the correlation between the type of work product produced and the value of the specific aspects of the operation (quality, volume, etc.).
Here’s what insights you can gain by applying operation input dimension to each measure type:
With what makes metrics operational out of the way, let’s see what’s unique about operation mortgage metrics.
The primary distinction of operational mortgage metrics from other operational metrics lies in the functions they measure.
Operational mortgage metrics measure the performance of functions within a mortgage company.
Thus, datasets, measures, and dimensions going to be unique to the operations produced by the mortgage operation functions such as:
I hope this post gave you insight into Operational Mortgage Metrics and a few ideas on how you can apply them to improve your mortgage operations
If you’d like to learn more about how you can apply technology to your mortgage operations, consider joining our mortgage technology newsletter.
I’m a software consultant with background in software engineering. Currently, I run a mortgage software consulting and development company that builds custom tools and automation solutions for mortgage lenders.
