There are great benefits to Proof of Concepts (POCs). They’re popular and are often looked at as an easy method of comparison between vendors. But are they really? The downside of POCs are the hurdles you have to tackle, like a decent budget and proper resources, before you see the benefits. Also, possibly even worse, they can delay you reaching your ROI. To explain this more, I’ve outlined the three major problems with POCs, and an alternative method for choosing the best solution, without these same obstacles. But first, let’s talk about how they’re used.
How Are POCs Used?
We usually see POCs utilized before major tech decisions are made. Carriers will use them to determine which vendors stand out from the rest, similar to the way someone would compare supercars, for example.
The difference is, however, when someone buys a supercar, there are many things to consider, and which one can drive the fastest quarter mile is the deciding factor for only some people. In insurance, though, the test drive tends to focus on one, single component: who can find the most fraud from my historical data set? They’re not focused on handling, fuel consumption, over-the-air updates, maintenance costs, value in 10 years from now or “the looks”, like buyers of supercars. For many companies, this is, and has been, the deciding factor for a while.
The challenge that then comes from POCs, is ultimately determining what fraud is, in the first place. If a carrier was only looking to identify fraud they’ve seen in the past, they probably wouldn’t be buying a new solution. So, this means investigators get an extra job. They’re now in charge of evaluating POC results. But wouldn’t it make more sense for your company to immediately reap the benefits of new technology instead of looking back at closed claims?
Here’s 3 reasons why a POC might not be the best choice.
#1: The Hidden Costs of a POC
Aside from the upfront fee of a POC, cost of investigations after the fact must be budgeted too. From experience, we know that many carriers forget about this initially. They don’t realize the amount of effort it takes to evaluate results, and they sometimes end up spending two or three times the anticipated hours. Honestly, it’s time consuming to investigate each vendor separately, but we also know that it’s a crucial part of the POC process.
Luckily, claims that have already been flagged as fraud in the past are easy. Your IT department can handle that analysis. But what about these new claims your investigators didn’t catch the first time? We know that they should be re-evaluated, but whose job is it?
#2: Integration into Core Systems
Secondly, if your company transitions to straight-through processing to clear up time for investigators, you’d expect to see an increase in efficiency and automation, right? Well, that’s only true if the fraud detection solution is integrated into your current core system. If it’s used as a stand-alone that doesn’t work in real-time, you might find yourself with a solution that nobody even uses in a few months’ time.
Claim handlers would need to check the application several times a day and find a way to manually incorporate those results into their daily routine. And because they are already pushed to be more efficient, this is another time-management challenge that everyone can relate to. Remember, integration is key, no matter how “future-proof” your core system may be.
#3: Self-Learning Models
Lastly, there’s one more feature we’re forgetting: self-learning models. These are especially relevant for fraud detection software but can’t be evaluated during a POC either. As we know, most fraud detection companies are now using AI in their core technology to keep up in our modern world. But one thing we need to consider is that as fraud evolves, AI models need to evolve with it. They need to learn from human judgement and intervention.
AI needs a human there to tell it whether it incorrectly flagged a claim, so it won’t make that mistake in the future. Yet not all vendors incorporate this into their software. For this reason, during POCs, self-learning models aren’t evaluated. The POC is a frozen moment in time and it’s nearly impossible to see the impact of self-learning models without monitoring results regularly. It’s even more challenging if a vendor leaves you on your own after implementation too.
So, then what?
The question all insurance companies should be asking themselves now is this: do I really want to do a POC if so many aspects of the software I’m evaluating are missing?
And unfortunately, a POC can sometimes be a throw-away investment for this reason. Comparing it to a supercar test, in the end, you’ll have results that tell you which car is the fastest, but only on that particular day only, and only on a tiny stretch of asphalt. Is that really what you want? We still know nothing about fuel consumption or handling.
In insurance, you need to know which vendor is going to be reliable and ready to race in a daily Supercars Championship. So, how do we do that?
Let’s Do a Pilot Instead
In the car industry, we already know how to evaluate. We ask friends or colleagues about their experiences, and we do a test drive. And preferably, not just one, but a few. What if we could do that for even a few months? Wouldn’t that be the ideal situation?
For fraud, it’s the same idea. Do a pilot implementation instead. You can try out the solution in real-time for a few months, and if you like it, you subscribe to the SaaS solution. If not, you just pay for the costs of evaluation and setting things up. But let me say this, you’ll start finding more fraud on the first day of going live with the trial than you would trying to identify fraud in cases you’ve already closed.
At FRISS, we are more than willing to let you talk to our references. We want you to obtain knowledge about the supercar you want to drive. We want you to have a month-long test drive and tell us not only about how fast the car goes, but about its reliability and long-term fuel consumption too.
If you’re reading this and still considering a POC, I would urge you to go for a FRISS-style test-drive instead. Get a feel for our car, the road, the brand, the people in the garage; I promise your mind will change. And hey, maybe you’ll start to understand why we, over here, say “Insurance is a beautiful thing” too.