Industry Thoughts

How we build for speed, accuracy, and customer ROI

September 23, 2023

minute read

  • Sorcha Nolan
    Technical Lead

I’ve always loved a good puzzle.

In fact, that’s part of what drew me to a career in engineering: The ability to solve problems through logic and build solutions that people find useful. 

As Technical Lead for the Fraud team at Inscribe, solving problems is what I do day in and day out. My team’s main goal is to help our customers, which are mostly banks, fintechs and other financial services organizations, quickly and accurately identify and prevent cases of document fraud through AI-powered tools and intelligent automation. 

But in the world of engineering, speed and accuracy can often be at odds with one another. That tension, coupled with the high stakes of working in the world of fraud prevention, is what keeps our jobs interesting, as we strive to not only stay a step ahead of our competition but also the very fraudsters we’re fighting against. 

Achieving speed and accuracy is also what generates so much return on investment (ROI) for our customers. So I thought I’d share some of the principles that ground and guide our team in our fight against fraud and how they help us develop and deploy the system changes that will deliver fast, clear, actionable results consistently and continuously.

Four engineering principles we use to build a solution with maximum ROI 

1. Design systems to enhance speed – even as more services are added

Speed is an important factor for our customers. As a fully automated platform, we aim to reduce human wait time as much as possible. That means designing our system and streamlining processes so that we can produce accurate results in mere seconds – and introducing new techniques that will help us consistently reduce processing times.

At the same time, we recognize that the value of our product should grow over time. Our team is regularly adding new features, capabilities, and improvements to provide more benefits to our customers. And therein lies the catch: Expanding the scope and services of a tool like ours can lengthen the overall time it takes to complete a process.

To that end, it has always been critically important for our team to maintain or even enhance speed as we add more features and services. This means that part of our innovation process isn’t just to identify new products or features, but to implement them in a way that doesn’t slow down processes or otherwise impact our product’s performance. It means taking a smarter approach to engineering whenever we want to add a new feature.

2. Strike a balance between speed and accuracy.

For our product to be successful, it must deliver results that people can trust to make critical business decisions.

Like many engineering teams, we recognize that the principle of diminishing returns applies to our product: There comes a point where the value of a miniscule increase in accuracy – at the expense of speed – is negligible or even counter-productive. In other words, our team needs to identify approaches or workarounds that will help us achieve optimal levels of both speed and accuracy.

 So that’s what we do. One example of this is how we run some of our fraud detector processes synchronously as opposed to sequentially. This allows our team to continue to enable two distinct features without adding time to the process or sacrificing accuracy.

3. Monitor for continuous improvement.

At the heart of every engineering decision is data. Our team relies on data monitoring tools to monitor speed, errors, and other attributes in any number of ways – for the system as a whole, for specific queries, parts of a process, or any other area we need to know more about.

To that end, our team leverages several different dashboards that graph patterns that reveal how long each part of the system is taking, as well as the total time from the initial customer request through all the different fraud processes and parsing until completion. We can look at jobs individually if there is a document that's taking a long time, or we can review an aggregate and analyze processing times over a given period.

We also build alerts into the system to inform us if something has slowed down unexpectedly. This is especially helpful if we are rolling out a new feature that performed well in testing but, once deployed, is taking longer than anticipated to process.

 Using our data monitoring and analysis system in this way helps us identify the slowest parts of our processes and find ways to optimize those areas. This can mean taking a different approach, finding a workaround, or even temporarily rolling back features and reworking them if the increase in time isn’t sustainable. 

4. Think outside the box – but get inside the head of the customer.

Customers value speed. Customers value accuracy. Customers want to get as much return on their investment as possible. Those ideas are at the core of everything our team does – and we are constantly striving to find new ways to increase all three: speed, accuracy, and ROI. 

To maximize value we need to think about the user experience from the perspective of the customer. From their point of view, when does the process begin and when does it end? What additional steps do our team take as part of the end-to-end process that don’t have any bearing on the customer? How can we isolate those pieces and complete them later so that we can shorten the customer journey and improve their experience without negatively impacting performance?

We recently used these questions to come up with a faster end user experience for our document collection portal — ultimately reducing the amount of time that end users had to wait for a “Success!” message by 80%. How did we do it? By reordering some of the verification processes so that only the most critical ones occurred in the moment (with the rest occurring after the end user receives the confirmation message). With this new configuration, we were able to reduce end user wait times from 15 seconds to just 2-3 seconds.  

The secret to reducing risk: Strategic engineering

At Inscribe, we believe that reducing risk and outsmarting fraudsters starts with strategic engineering. Our team works tirelessly to deploy the features and services that will help our customers stay a step ahead of digital criminals and identify the very first signs of fraud.

By following these four guiding principles, our engineering team proves that speed and accuracy need not be mutually exclusive – and that with our continued energy and ambition, we can continue to increase both to help us win the fight against fraud.

Interested in joining our team in the fight against fraud? Check out our Careers page to browse open positions that can have you on the frontlines against fraud. 

Start your AI journey today

See how Inscribe can help you unlock superhuman performance with AI Risk Agents and Risk Models.