We all understand the need for an accurate address. Without one we face tangible issues. Post gets misdirected, taxi drivers can’t find us, and devastatingly, our takeaway arrives late and cold, usually after a phone call... these are just a few examples.
Whereas for many of us, giving or getting an incorrect or inaccurate address is inconvenient, for insurers, this information is absolutely crucial to the continued financial strength of the business, and the ability to remain compliant with PRA and FCA regulations. Insurers need to know, with pinpoint accuracy, where the property they are being asked to insure is located; this information is essential to understanding the risks associated with that location and therefore, determining their level of exposure, the price, and any underwriting terms that need to be applied to the policy.
Cracking the geocode
Whereas many insurers still rely on postcodes to assess the risk being placed on cover, geocoding presents a more accurate and valuable way of understanding risk.
Geocoding is the process of taking an address and determining the geographic coordinates associated with it. Following the initial creation of an address by the local authority, this information is provided to Ordnance Survey (OS), who then survey the property and produce the associated co-ordinates. This is the geocode.
In the UK, the OS surveys the property resulting in very high-quality data. Elsewhere however, many geocodes are developed using the process of interpolation, where the data is produced through an understanding of a number of points on the street, inferring where the intermediate properties are using a set of rules or an algorithm. Though still more accurate than a postcode, the process of interpolation may lead to data that falls short of what insurers need to develop an accurate picture of the risk.
Some services such as Google or Microsoft Bing provide good, general-purpose solutions which many of use daily for navigating around cities or in our cars. However, the underlying data is typically not from an authoritative source such as the Ordnance Survey. In addition to the risk of false positives using these services, some place restrictions on how the data can be used. Storing or caching the geocodes is often restricted in the T&Cs which can undermine the usefulness of this data when it comes to exposure management or risk analysis.
As well as the provenance of the coordinates, it is important to consider some other factors when using geocoding to assess and price risk:
Firstly, we need to understand the positional accuracy of the code. This describes how close the geocode is to the actual point on the earth's surface that we are trying to locate. For insurers who are interested in risk associated with a particular building, the key is that the geocode is within the building outline of the relevant property. If this is true, the positional accuracy is sufficient to meet the insurer’s requirements. Even with a high-quality source such as the Ordnance Survey there is a lifecycle associated with an address e.g. a new build may initially be assigned coordinates of a neighbouring postcode and then only be surveyed at address level on or after the completion date.
The other consideration is whether the geocode has been assigned to the correct building in the address dataset. Data quality is a challenge for all insurers, in particular commercial addresses. Incorrect postcodes, typos in the address, superfluous information such as “The Warehouse next to the Garages” can all confuse geocoding services and lead to a ‘no match’ or even worse, an incorrect match or ’false positive’.
Different sources of addresses can also be a problem, the postal service view may be at odds with the local authority as they have different needs i.e. delivering mail vs taxation and providing services. The organisation name can also be challenging and can mean different things to different people: Atlantic Foods Ltd and The Codfather, for example, may be two ways of referring to the same commercial premises.
If the address match is incorrect, automated risk lookups will be based upon the location of the geocode and not the building, this could have serious implications in relation to the risk profile of that location and the exposure faced by the insurer.
We must also consider the concept of relative accuracy. This isn’t just a geocoding issue but relates to all spatial data. If the maps don’t align well with the risk layers and the geocoded points, then the picture being presented will be confusing and misleading. Most countries in the world have their own spatial referencing system and there are international systems which are used to speak a common spatial language, converting between these must be done carefully to avoid issues. The Ordnance Survey use British National Grid (BNG) in most of their databases, many international services will use some form of the WGS84 referencing system. There are several ways of converting between these, picking the right one is important as it could put an address tens of metres from the actual location and into an incorrect flood band for example.
The impact of a bad address
There are two main areas that can be directly impacted by the presentation of an incorrect or inaccurate geocode to underwriters: peril risk scoring and exposure management.
Firstly, the underwriter may receive information relating to peril risk score or exposure values all based on an incorrect location. This could lead to an underwriter accepting business that is actually outside of their appetite, or conversely declining business that is actually high quality and within appetite.
Where peril risk assessment is being used to inform pricing, we can see how this can quickly lead to a situation where poor risks are underpriced and good risks are overpriced, directly impacting the bottom line of a business when a claim is made.
Effective exposure management helps insurers to understand where their exposures are located and the risks present in that location in order to secure adequate reinsurance protection to protect the account from all perils including serious catastrophe events.
Poor quality geocoding can make effective property level exposure management difficult because it will be challenging to identify the property in question. This can lead to an incorrect assessment of accumulation risk taking place potentially resulting in either risks being accepted that have unknowingly taken us beyond our limit or if the assessment tells us that we have no capacity then we have potentially declined business unnecessarily. Either way these are negative impacts to the portfolio performance.
Insurers must work in partnership with specialists with local knowledge and a deep understanding addresses and geography, and align this data with that from numerous other sources to provide a full and accurate picture of the risks they place on cover, otherwise a cold pizza may just be the least of their concerns…
To chat about this subject, and find out how you can embed geocoding into your insurance business, contact me on: