Squeezing real value from your data

In this session of Talking Tech with Tavio and Tony, they look at extracting real value from your data.

Tavio Roxo, the CEO of OWLS Software and Tony van Niekerk, Editor of COVER Publications.

One of the main themes from Insurtech2022 was how to relate better to the client, how to understand the client better and how to meet them where they are. That is the whole idea of humans when we want it; trying to find that balance. What came out is that we need to understand the client on a granular basis, and this is where data comes in.

So, everybody is talking about data being the new oil and all those sorts of nice things and everyone has realized that this is a critical issue. However, it does not seem like we are really walking the talk and getting it right. What is your opinion on that?

Tavio: I do not think we are Tony, because data is very binary. Data is either right or it is wrong and there is no 80% correct. A large portion of the data which insurers and the insurance industry are utilizing is data that finds itself having been created out of perhaps older legacy systems. That data in its natural form is dirty, it is not data that you can rely on.

So, a large portion or a large function of what must happen is the data must be extracted, it must be transformed, and it must be manipulated. It must be cleaned up before it can be utilized.

Tony: So where do you think are the shortcomings currently that prevent us from getting it right? We do have the tools to do everything that you have mentioned now, to be able to do that.

Tavio: Even though we might have the tools, there is a fair amount of effort between taking the underlying data and then modifying it, transforming it, and manipulating it in order to consume it and rely on it.

Even if that process takes a few weeks, it still is not live data that allows you, as an insurer, UMA or as broker, to rely on it 100%. But more importantly, because you must do all these things to it before you can work on it, you cannot have your system rely on it either. So, the system cannot then utilize the data to allow automation, to allow straight through processing, and so on, allowing all sorts of functions that you ultimately want to get to, but you cannot do it because there is a data gap.

Tony: It can be said that we need to look five or 10 years into the future at what data we do not need now but that we might need in the future and collect that data now, to be able to use it in the future. Your thoughts?

Tavio: There are two aspects to this. The first is that, even if you ask the question about the data that you need now with the view that you are going to only need that data in 10 years’ time, you must validate that data. You must have a pretty sophisticated rules engine that ensures that by the time that data hits your database, it is the right data. Looking at many insurers, UMAs or brokers you will see that they have a policy listing of all their policyholders and under ID number they might have the first six digits, or they might just have a date of birth or under registration number they might have TBA, which is, to be advised or to be provided.

You have the data, but it is not data that you can rely on or use. So, I think a big component around data ought to be the validation and the verification of that data before it hits your database.

Secondly, your old data must be transformed and manipulated so you can use it. And then new data, ensuring that when you now onboard a new client, that you are checking whether that ID number is valid, in real time, checking whether those banking details are valid in real time or checking Mead & Mcgrouther to see if a vehicle exist.

So, you need to use systems with these various integrations embedded in them to ensure that when you are working with the data. It must all happens seamlessly because you cannot rely on your policyholder or your client to give you the correct data. They give you the data that they give, and you must be able to, at that point in real time, say to them; thank you very much, but I see that your surname is spelled incorrectly.

Tony: Then how do you make sure that data stays up to date because, as you say, you are collecting the data and the data might become stale over a period.

Tavio: That is always going to be a difficult one, because it ultimately relies on the policyholder informing you of the changes in their world. So, if they change address, they should notify you.

There are ways that you can get around it. You can run macros on your data sets once every quarter or once a year and you can do a mapping exercise to see whether that address is still linked to that ID number against the various credit bureaus, for instance. You then ping your client and say, we noticed that there is an updated address for you now of 115 versus 118, can you click and agree that this is the new address or not.

But you and I both know that the chance that the client will agree and will accept is low. So, for you to keep the data from not being stale, is a bit more of a difficult challenge. The primary focus should then be on new data, to make sure that when you get it that very first time, that it is in fact correct.

Tony: This process is of great value to everyone from the one man show independent broker to the UMA and the insurer. It supports the whole value chain.

Tavio: Absolutely, and that is where your primary focus must be. In the brokers environment, where they are at the coal face with a client, ensuring that you do not onboard any new business with a TBA on a license registration number for a new vehicle. You should not be able to do it, the system should block you, the technology should block you. You should be able to get the correct information from the outset.

That will go a long way towards ensuring that the data you are working with is good data. Because once you have that correct reg number, you can cross map it against the correct address. The address can be validated in Google and now your insurer can run claims experience heat map on an address which is real, not on an address that you know is perhaps incorrect. They can run information relating to the underlying vehicle code based on the reg number and they know that is correct information versus using, “I think it is a 1.5 litre Toyota Corolla”.

The best part is that, once you get the correct data, not only does it mean that you can optimize what you currently have, you can streamline your processes, save costs and have happier clients while managing their expectations better.