Why is Data Accuracy so Important for Insurers?

One of the most frequent discussions we are having with insurers lately is about data accuracy. Many questions arise when we mention that open data is helping our clients to optimize their performance. We often find skepticism involved in the idea of using "external data" to make such important decisions as assessing a business risk. And that is reasonable as the first thing that usually comes to your mind about online data is the idea that anyone can put anything on the Internet.

In this post, I will try to summarize the questions we are asked and the answers we usually provide. I will also do a humble attempt of helping the skeptics gain a bit more confidence in the use of external data.

Now, what is data accuracy?

Data accuracy and reality could be synonyms. The closest to reality a data point is, the more accurate it will be. In other words, an accurate data point is a data point that describes an aspect of reality accurately.

Why is it so important?

Any type of business makes decisions based on data. Whether it is an excel file, a database, the result of an algorithm, or just personal experience, we all make most of our (professional) decisions based on data. And the insurance industry as its whole is based on data-driven decisions. Therefore, having accuracy in our data will enable us to make better decisions.

Why is data accuracy so important for insurers?

There are numerous problems that come with data inaccuracy. If a business is insured under a different address, activity, and size, the risk that the insurer is covering is completely different from the reality. When we find these mismatches, very uncomfortable situations take place when the business has a claim. For example, we have seen how insurers and business owners end up in court because the insurer claims they cannot cover that claim as that was not the risk considered in the policy while the business owner says that they were paying for something they were not getting. And this is just one of the possible outcomes that could be solved if those 3 or 4 data points were corrected on time.

How do we achieve data accuracy?

Two basic steps: find proper data and avoid errors.

First, to find proper data there must be a system that knows what we need to find. Our solution, for example, uses the name and address of a business to collect information. Then, the information collected from a few websites is again used to expand the search. The system finds online data and analyzes how similar it is to what we are trying to find. For example, if you are trying to find an address, anything that doesn't look like an address can be discarded.

Secondly, there must be a system in place to clean errors. When it comes to open data, our systems are ready to discard data that do not match. For example, if we collect financial data of a business but the name, address, or the number of employees do not match with the information we previously collected, our system will automatically discard and erase this data.

But really, how can you trust online data if anything could be fake?

Sometimes one wants to answer a question with another question. I avoid doing so because I am a politeness believer. But these questions always come to my mind: well, how do you know your current data is not fake or wrong? How often do you update your data? How do you know the data received during the submission was 100% trustworthy? Is there any data quality verification process you performed during the submission process?

Of course, it is legit to have doubts about online data. But we should also question our current data quality. In that diagnosis, similar insurers can actually be very different (from "my data is correct" to "of course I need to improve my data quality").

Answering the question, we know we can trust open data depending on where it comes from. There are hundreds of data sources available. We rate them depending on the quality of data and how trustworthy they are. For example, government databases are something we can completely rely on. The accuracy might not be 100%, but you would not suspect that a government is interested in lying or faking data about a registered business.

Another example is the website of the business. We can believe that a business is interested in sharing where their address with accuracy as they want to attract as many customers as possible. Especially if we find that exact same address again in Google for Business, Yelp, Tripadvisor, the property register, social media, etc. Crossing the same information in different sources is a key to validating data quality, and therefore, validating data accuracy.

Conclusion

There can always be mistakes or mismatches. But leveraging open data is the best way we have available to get the closest possible to reality while avoiding human tasks or interactions.