5 Ways DataOps Can Make Life Better for Businesses in Highly Regulated Industries
September 12, 2022
If you’ve never heard of DataOps, think of it as a collection of technology, business practices, and operating principles that enable businesses to operate more effectively and efficiently via automation, orchestration, and predictability. It includes establishing rules for storing, processing, documenting, extracting, and developing with data at scale.
Today, the leaders in financial services are those not only embracing data but also implementing DataOps to ensure their data is fueling their businesses as effectively and efficiently as possible. Those that hope to stay operational for the long term must follow suit.
Read more: Financial Services’ Long-Time Data Advantage
In this piece, I’ll highlight five concrete benefits implementing DataOps can bring to businesses in highly regulated industries like banking and mortgage, then explain which elements of DataOps functionalities make those benefits possible.
Benefit 1: Increase Products per Customer and Reduce Customer Churn
Fewer than one in five homeowners refinance their mortgage with their primary mortgage lender – but it’s not because they were dissatisfied with their experience. Fully 69 percent of borrowers apply for their mortgage online, which means they’re probably starting the process by Googling “apply for mortgage” or something similar, and then applying with whichever lender shows up in results.
DataOps can help you be top of mind when existing borrowers want additional products.
You could start by using the data you have about customers to build a model to predict refinance behavior – say, reaching a certain amount of equity, plus some delta between their primary loan’s interest rate and current rates, plus certain credit behavior.
This model can predict when a customer is likely to refinance and you can proactively contact them with information about how to do that with you. As you gather more data, you can improve the model further.
DataOps comes in particularly handy here: it establishes clear procedures for documenting changes you make so you can easily track their performance and tweak as needed moving forward. As the model is built, trained, and deployed, important processes like managing version control, automatically deploying computing containers for processing, and data orchestration to trigger the model, process data, and push back the results quickly are all aspects that become well managed by your DataOps processes.As you automatically deploy your scalable basket analyses, churn scoring, or feature creation you can more cleanly track and manage changes, versions, roll-backs, deployments, and observability.
The result: more products per customer, lower customer churn, and more data points that drive even more insights and have it be done in an organized, repeatable, and scalable process.
Go deeper: 3 Data Trends in Financial Services
Benefit 2: Make Compliance Easier + Reduce Fines and Penalties
I mentioned above that part of DataOps involves documentation procedures. This can prove valuable in the context of compliance and of preventing fines and penalties associated with noncompliance. Let’s take a look at how.
So you’ve got your “likely to refinance” model and you’re using it to retain existing customers who might otherwise have gone elsewhere. In fact, in the years you’ve been using and refining this model, you’ve started outperforming the industry overall: thanks to your model and outreach, 30 percent of your primary loan borrowers who refinance do it with you.
All that “refining” translates, on the back end, to pulling more data points, writing more code, putting it in the data frame, cleaning it up, QA’ing it, and so on. And because of the industry you’re in, all of that – the code, the features, the data, the model – may be subject to regulatory compliance and audits.
With a DataOps program in place, complying and passing audits is a breeze. The documentation practices and workflow in place mean you can easily and clearly communicate where data came from, why changes were made, and how the model is being used to inform investment decisions.
Benefit 3: Preserve Knowledge through Staff Turnover
As you can imagine, the same documentation protocols that make compliance audits easy help preserve knowledge through staff turnover.
Without DataOps in place, you might have a brilliant data scientist who builds a model without documenting their process. When they leave, that knowledge capital disappears with them. Your company loses information about what worked in past versions, what didn’t work, what they were tracking to potentially iterate on next, and so on. Instead of steadily moving forward, you have to spend several months attempting to recover what you used to know.
With DataOps in place, all those things are documented as a matter of course. Staff turnover and technical failures will have much smaller impacts on your performance.
Benefit 4: Shorten Onboarding Time
Just as a DataOps practice makes an employee departure less disruptive, it also makes onboarding more efficient.
Because DataOps encompasses both technical workflows and business operations, businesses that embrace this model operate in such a way that makes it easy for new hires to observe how and why things are done the way they are.
For example, say the success of the refinancing campaign inspires a bank loan officer to request a similar model to predict when mortgage customers are likely to be shopping for an auto loan.
A newly hired data scientist could look at the documentation around the refinance prediction model, learn about which feature sets have worked best, and build their initial model powered by those insights.
Benefit 5: Make it Easier to Extract Value from Data
Throughout this piece, I’ve referred to models that let businesses derive real value from their data. Having a DataOps practice in place is key to making that a reality.
Without DataOps, a bank or lender might develop a data-fueled model, deploy the model at the code layer, then spin up a virtual EC2 instance to read the data, process it, and push it back to the database.
That’s all well and good, but all that code has to live somewhere. What happens when you have to make a change to accommodate a new data point or fix a bug? Where do you document that change so that auditors can track it and future team members can learn from it?
DataOps offers a playbook for managing version changes, history, and updates so that team members can work together on a unified end result. It establishes rules of engagement so that data scientists’ time and energy can be spent building and improving rather than explaining things to each other.
DataOps Is Necessary for Efficient Data Use
Market leaders in mortgage and banking are doubling down on tech investments right now because they know that’s the only way to improve operational efficiency in the long term.
This doubling down, of course, requires upfront investments in both technical system improvements and staff training, but the downstream benefits are significant. What’s more, mortgage lenders, servicers, and banks won’t be able to remain competitive much longer without making such investments. Eventually, those that don’t embrace the kinds of process improvements that DataOps make possible will find that they can’t compete with businesses that have made the transition.
Not sure where to start? Get in touch. We’d love to help you identify the biggest opportunities in your organization and lay out a roadmap to help you get where you want to be.