Find the recoveries hiding in your claim notes
Reads claim notes for another party who may have caused the loss, the evidence and anything barring recovery. The recoveries team decides every referral.
Try it on this example
First account of the loss: Came home from work to water coming through the lounge ceiling under the airing cupboard. Turned the water off at the stopcock. The boiler was only put in a few weeks ago.
Handler and adjuster notes, in date order
- Do the notes say enough about the cause of the loss to judge whether someone else caused it?Yes97%
- Who other than the policyholder may have caused the loss, from the notes?Contractor or tradesperson100%
- Do the notes name or identify the other party who may have caused the loss?Yes98%
- How strong is the evidence in the notes that someone other than the policyholder caused the loss?Documented or admitted100%
- Do the notes say a specific appliance, part or product failed in itself?No86%
- Do the notes say the failed part, photos or a report on the cause have been kept?Yes98%
- Do the notes suggest the policyholder or their household caused or added to the loss?No91%
- Do the notes mention anything that may bar or limit a recovery?No85%
- Do the notes say a recovery, a referral to recoveries or a demand to the other party has already been made?No88%
These are real answers stored from one run on this example.
The prism behind it
Find the recoveries hiding in your claim notes
Fields
- First account of the loss
- Handler and adjuster notes, in date order
Context
We are a UK home insurer. This prism looks for recoveries: money we can claim back from someone else who caused a loss we paid. It runs as a file of closed claims, one row per claim, and on open claims when a note is added. Each row gives the first account of the loss and the claim notes joined in date order. When the notes are longer than one call can take, code keeps the first account and the most recent notes. Recoveries we pursue: a contractor or tradesperson whose work caused the loss; the maker or seller of an appliance or product that failed; a neighbour, landlord or management company whose property or neglect caused the damage, such as a leak from the flat above; a utility or public body, such as a burst water main or a power surge; a driver or vehicle that hit the property; any other business whose actions caused the loss. We do not pursue: an unknown or untraced person, weather alone, wear and tear, or the policyholder's own household. The recoveries team decides every referral. Amounts, time limits for bringing a claim, minimum values and whether a recovery is worth the cost are decided by code and by people, not here.
Questions
Do the notes say enough about the cause of the loss to judge whether someone else caused it? Yes / No
Many closed files record only the type of loss and the payments. Judge whether the cause itself is described. Yes: The notes or the first account describe what caused the damage, clearly enough to say whether anyone else was involved. No: The cause is not described beyond a label such as "escape of water" or "storm", so no view on another party is possible.
Who other than the policyholder may have caused the loss, from the notes? Choice
Choose the party the notes point to most strongly. Choose No other party when the notes put the loss down to weather, wear and tear, the policyholder or their household, or a person nobody can trace.
Do the notes name or identify the other party who may have caused the loss? Yes / No
Count a firm name, an address, a vehicle registration, or a make and model of the product. Do not check whether the details are right. Yes: The notes give enough to identify the other party. No: No other party is identified, or none is involved.
How strong is the evidence in the notes that someone other than the policyholder caused the loss? Scale
Judge the strongest evidence the notes describe. Do not judge whether a recovery would succeed.
Do the notes say a specific appliance, part or product failed in itself? Yes / No
A product that failed because it was fitted or used wrongly does not count; that points to the fitter or the user. Yes: The notes say an appliance, a part or a product was faulty or failed while being used normally. No: No product failed in itself, or the fault was in how it was fitted or used.
Do the notes say the failed part, photos or a report on the cause have been kept? Yes / No
Yes: The notes say the part, the product, photos of the cause or a report on the cause is kept by us, an adjuster or the policyholder. No: The notes do not say any such evidence was kept, or say it was thrown away.
Do the notes suggest the policyholder or their household caused or added to the loss? Yes / No
Count something the household did or failed to do, such as leaving a tap running or not reporting a leak for weeks. This reduces or rules out a recovery; it is not a finding about the policyholder. Yes: The notes suggest the household caused or added to the loss. No: Nothing in the notes suggests this.
Do the notes mention anything that may bar or limit a recovery? Yes / No
Count a waiver of subrogation, a contract clause limiting what the other party pays, an agreement between insurers not to claim from each other, the other party having gone out of business, or the policyholder settling directly with them. Yes: The notes mention at least one such bar or limit. No: No bar or limit is mentioned.
Do the notes say a recovery, a referral to recoveries or a demand to the other party has already been made? Yes / No
Yes: The notes say we, an agent or the policyholder already referred the claim to recoveries or wrote to the other party or their insurer to claim the loss back. No: No recovery step is recorded.
Lens columns
notes_sufficient, notes_sufficient_probability, responsible_party, responsible_party_probability, party_identified, party_identified_probability, fault_evidence, fault_evidence_average, product_failure, product_failure_probability, evidence_kept, evidence_kept_probability, insured_contributed, insured_contributed_probability, recovery_bar_mentioned, recovery_bar_mentioned_probability, recovery_already_open, recovery_already_open_probability
Run it on your own text
Add this prism in the app, change any question, and test it on a file of your own.