By

In recent years, a significant number of online or subscription-based computer models have emerged, promising policyholders and intermediaries reliable, quick and cheap estimates of reinstatement costs.

While these tools showcase the power of artificial intelligence (AI) and advanced modelling, they have inherent limitations. Not all the factors impacting reinstatement costs can be captured from address data, plans, or satellite images alone. As a result, their outputs must be approached with caution.

In this article, we’ll explore the challenges in arriving at appropriate declared values using modelled assessments, and why some of the results from these platforms and tools might not deliver the accuracy, or peace of mind, policyholders expect.

The challenges with modelled assessments 

While desk-based assessments have a place in assisting policyholders review declared values there are a number of reasons why their use can be a challenge:

1. Relying on imagery to determine characteristics

AI models often use satellite or street level imagery to assess the characteristics and materials of a property. This can lead to…

Misinterpretation of use and design

With modern re-use and adaptation of buildings it is very easy to mistake a warehouse for a factory, or to assume a mixed-use city-centre office building is just an office, when it could also include a bar, restaurant and/or residential accommodation as well. 

Structural guesswork

Using Google Earth or similar to calculate the size and structure of a building means relying on variable quality imagery. Structures can be below tree cover or built into embankments or below ground. 

Material identification issues

Modern AI models claim to be able to determine the building type and materials used. However, buildings may be overclad, concealing their original finishes. What may appear to be brickwork can easily be brick-effect panelling, and internal structural frames are entirely hidden, requiring a physical inspection to verify the details.

2. Inaccurate address information 

Vague or missing information

Addresses for locations provided by policyholders may be missing or incomplete, e.g. only showing a street name or the industrial estate name rather than a specific address. This is a common issue with industrial and remote locations in particular. 

Errors in postcodes or building numbers

Mistakes in address details can mislead models, causing them to associate the risk with the wrong property entirely.

Unmentioned adjoining facilities

Addresses might not indicate or include nearby (insurable) structures like car parks, substations, pumphouses, or equipment stored offsite. These assets could still fall under the policyholder’s insurance responsibility, exposing them to significant underinsurance risks.

As a result of these issues, tools or models that rely on post office or other standard address data may struggle to match locations to insurable risks.

3. Insurance responsibility

Site boundaries, occupancies, and insurance responsibilities are rarely clear from desk-based assessments

Multiple-occupancy properties

There can be multiple occupancies within the same building or location. The demarcation between these occupancies may not be clearly defined in an address or in the location details provided.

Determining responsibilities of leased properties

Where the property is held under a lease, who is responsible for the insurance? Is the lease on a full insuring and repairing lease? 

Even if the lease addresses the building insurance position, it may be silent on who insures the tenant’s improvements. Depending on the property at the time the lease was granted, the extent of the tenant’s fit-out may be difficult to determine. 

Lease questions that most computer models and desktop assessments don’t answer include:

  • Have previously installed improvements reverted to the landlord on a lease renewal? 
  • Who is responsible for the previous occupier’s fit-out if these are retained by the new tenant? 
  • Does the landlord insure the tenant’s improvements and then recharge the cost to the tenant, or is the tenant responsible for insuring their own improvements? 
  • How have fixtures and fittings been treated? For example, does the value arrived at for the building by a model represent a bare shell structure or does it allow for internal partitioning, electrics, kitchen and toilets? 
  • Does any demarcation adopted in these models match to the lease and insurance policy terms?

4. Inconsistent floor area data

Accurate floor area measurements are essential for reinstatement cost calculations, yet they are frequently misrepresented in computer models:

Measurement standards

Insurance assessments typically use gross internal floor area (GIFA), but policyholders regularly provide net lettable area (NLA) data from leases or recent property valuations, which may exclude common spaces, voids and other non-lettable areas. This discrepancy can skew cost estimates.

Overlooked features

Basements, mezzanine floors and canopies are frequently omitted in desktop assessments. Similarly, temporary structures like tented spaces or modular buildings, some of which may have been in use for decades, may not be picked up in floor area data.

5. Hidden factors

The specific conditions of a site will influence the reinstatement cost of a property and many of these can only be verified through a physical inspection on site. For example:

  • Is the property on a sloping site or have other topographical characteristics that could impact reinstatement costs?
  • Do the soil conditions in the area require piling or other substantial changes to the foundations?
  • Are there easements across the site above ground or below ground such as utilities that could impact the cost or timelines needed to reinstate?
  • Are there structures near to or abutting the property that could increase costs of reinstating or could mean that the expected costs of shoring up and protecting these properties in the event of an insured peril would be higher?
  • Is the site close to underground infrastructure (e.g. Underground tunnels in London) that would impact costs.

These issues can lead to significant differences in the final assessment and many of these can only be determined through a physical inspection and verification on site.

6. The limits of satellite imagery

Several of the computer models used for desktop assessments state that they can automatically measure floor areas, building heights etc., and arrive at declared values from satellite and street level imagery. However, this imagery has inherent shortcomings:

Data staleness

Even the most sophisticated commercially available satellite imagery can be several months old at specific locations across the UK, meaning that recent structures may not show on these images.

Blind spots

Overhead views cannot reveal basement levels, void spaces, or mezzanine floors – features that significantly affect reinstatement costs.

Identifying materials

Several companies offering opinions on reinstatement costs based on AI models also state that they can automatically determine construction materials from external images (e.g. Google Street View) and satellite imagery. As mentioned, these models will struggle to tell the difference between a brick building and a steel- or concrete-framed building with brick effect wall panels, but they also won’t be able to categorically determine if a building contains asbestos which would materially impact the costs of demolition and debris removal. 

Internal finishes

Satellite and other external photography rarely gives any indication of the extent of the internal finishes. There is a big difference between the costs of reinstating a bare shell industrial unit and one with internal high-tech assembly areas with cleanrooms, HVAC systems and high end glazed partitioning.

Why a site survey is essential

At Charterfields, we recommend site surveys to ensure comprehensive and accurate reinstatement cost assessments. Unlike desktop tools, a site survey provides:

  • Accurate structural details – Surveyors can confirm building structure, materials, and internal finishes, capturing nuances that remote tools might miss.
  • Detailed floor area measurements – A site visit ensures all areas – including basements, mezzanines, voids, and temporary structures – are accounted for.
  • Clarity on insurance responsibilities – A professional surveyor can review leases, tenant agreements and site boundaries to clarify coverage obligations.

By taking the same approach as loss adjusters would during a claim review, a site survey by a professional surveyor minimises disputes and protects policyholders from the financial risks of underinsurance.

Conclusion 

Online and desk-based assessments are only as good as the information provided. Even with the best and most accurate information (which is rarely available) they are still prone to miss aspects of the site and property that could have a substantial bearing on the correct reinstatement cost.

While desktop assessments and AI models offer speed and convenience, there are clear limitations in modelling reinstatement costs.

Policyholders relying on computer-modelled cost assessments should understand the limitations of these tools and avoid assuming they provide fully reliable declared values suitable for insurance cover.

At Charterfields, we provide expert site surveys that go beyond numbers, ensuring you have peace of mind and comprehensive coverage. If a detailed site survey is not possible, we offer desktop assessments which, unlike most computer models, are matched to analysis and reviews by senior surveyors. Get in touch to find out more.