TL;DR: Barbados’ motor insurers collectively possess years of claims data showing when, where and under what circumstances accidents occur. What if insurers pooled selected, anonymised data and used it to warn motorists about historically high-risk days, times and locations? A Friday-morning Facebook post, TikTok video, Instagram post or email might tell motorists that historical claims show an elevated number of accidents during particular hours or at particular locations. The idea is simple: use yesterday’s accidents to try to prevent tomorrow’s.
I’ve been working with computers and technology for more than 20 years in various industries, roles and across different countries. Spending that long around technology inevitably shapes how you look at problems. One of the things I’ve been interested in is how we can use technology and the data we already have to solve problems or to do things better.
I especially thought about this during my years as the Chief Technology Officer of an insurance company in Barbados. In particular, I wondered whether all the motor claims data being collected by insurance companies could be used not only to deal with accidents after they happen, but perhaps to help prevent some of them from happening in the first place.
That’s the idea behind this article. Insurance companies collect large amounts of information as part of their everyday operations. And one question I found particularly interesting during my time in the industry was this:
How can an insurance company use the data it already collects not only to process claims, but potentially to prevent some of those claims from happening in the first place?
Motor insurance is one class of insurance that provides an interesting example.
Every Accident Creates Data
Every time a motor accident results in an insurance claim, data is generated. Data like: When did the accident happen? Where did it happen? What day of the week was it? What were the weather and road conditions? How fast was the vehicle travelling? What were the circumstances?
This isn’t hypothetical. The motor claim forms of several insurers in Barbados ask questions relating to the date, time and place of the accident, as well as weather conditions, condition of the road, et cetera (see here, here, here and here for examples). One claim tells you very little. Thousands of claims accumulated over several years can potentially tell you quite a lot.
(Note: This article assumes that at least one vehicle is insured – a large number of motor vehicles in Barbados are un-insured, see here, and that someone files a claim).
Thousands Of Accidents And Millions Of Dollars
In January 2025, then General Insurance Association of Barbados (GIAB) president Randy Graham said Barbados was experiencing between 6,000 and 7,000 motor accidents per year, or approximately 500 to 600 per month (see here).
The financial consequences are significant. In February 2025, Graham told Barbados TODAY that motor claims for 2024 had reached approximately $87.6 million, compared with $86 million the previous year. He argued that reducing the frequency of accidents was necessary to make a meaningful difference to claims costs (see here).
In January 2026, Co-operators General Insurance CEO Anton Lovell estimated that Barbados was recording approximately 15 road accidents every day, with his company alone experiencing an average of seven to eight accidents daily. Subsequent reporting put the average repair bill per collision at around $10,000, with injury accidents potentially costing considerably more (see here).
And in August 2026, Co-operators said it was reviewing its motor insurance premiums because the rising number of road accidents was continuing to drive up claims (see here).
So perhaps there is another way insurers could attack the problem.
What If The Data Warned Us Before The Accident?
Imagine that an insurance company analysed five or ten years of its motor claims data. Maybe the analysis showed — and I am making up the following numbers for illustration — that accidents were substantially more likely to occur on Fridays than Wednesdays, and that on Fridays between 4 and 6 PM was particularly bad, and in or around particular roads, junctions or other areas
(A caveat: raw counts like this partly reflect traffic volume, not just risk — a busy junction or rush-hour window will show up more in claims data partly because it’s busy, so the real prize would be pairing this with traffic-flow figures to find what’s dangerous beyond what its volume alone explains).
Now imagine that on Friday morning the insurer sent its customers the following message via email, WhatsApp, Facebook, IG, TikTok, et cetera:
Driving today? Our historical claims data shows that Friday between 4 p.m. and 6 p.m. is one of our highest-risk periods for motor accidents. These specific locations have historically recorded particularly high numbers of claims. Take your time and get in good!
Suddenly, claims data isn’t only something sitting inside an insurance company’s database waiting to be analysed by actuaries, underwriters and claims departments. It becomes preventative information.
A driver who sees that message Friday morning may remember it at 4:30 that afternoon:
“Oh yeah. This is the time they said accidents tend to happen.”
Maybe that person slows down. Maybe they leave a little more distance between their car and the one in front of them. Maybe they pay more attention approaching a junction that appeared in the insurer’s warning.
Will that stop every accident? No! Could it influence the behaviour of some drivers? It might, and the cost of sending out the messaging (email, Facebook post, et cetera) is likely to be a lot less than the cost of processing a motor claim, the time and productivity lost on the road as a result of the accident, the inconvenience to other road users and more!
Imagine If Every Insurer Participated
This is where I think the idea becomes much more interesting. A single insurer sees only its own claims. But collectively, Barbados’ motor insurers potentially possess an extraordinary dataset about what is happening on our roads.
Collaboration between insurers and public authorities on road-safety issues isn’t new. In August 2024, when Barbados announced its new Road Traffic Accident Investigation Policy, GIAB publicly supported the change. GIAB president Randy Graham said at the time that insurance companies were paying approximately $85 million to $90 million every year to clients for vehicular accidents. The Barbados Government Information Service also reported that insurance-company responders had received training from The Barbados Police Service to gather information at accident scenes (see here).
So there is already precedent for the industry working together. For the idea I am proposing, insurers wouldn’t need to share any Personally Identifiable Information (PII) such as customers’ names, individual claim values, policy numbers, vehicle registration numbers or other information that could identify individual policyholders.
Instead, participating insurers could agree on a deliberately limited and appropriately anonymised dataset containing information such as (remember, this type of data is already being collected by individual insurers via their motor claim forms):
- date and broad time period of the accident;
- general location;
- type of collision;
- weather and road conditions, where reliably recorded; and
- other non-personally-identifiable variables useful for identifying patterns.
Aggregate it. Analyse it. And publish the useful patterns back to the public — a short weekly note on what the data showed, not a report that sits unread.
A Potential Win-Win For Everyone
The interesting thing about this idea is that the incentives are unusually well aligned: Motorists benefit if there are fewer accidents. Government and emergency services benefit. Insurers benefit from fewer claims.
And ultimately, the wider public bears some of the consequences of frequent accidents through potential increases in premiums, traffic congestion, lost time, damaged property, injuries and demands on emergency services.
There is another interesting point. Insurers are already trying to change driver behaviour. In November 2025, CBC reported that Co-operators General Insurance had organised defensive-driving training for its policyholders because of concerns about the frequency of accidents (see here), while Sagicor offers a safe-driving rewards programme (see here).
So the underlying principle isn’t particularly radical. The question is whether data itself could become another road-safety tool.
The Data Already Exists
That’s the part I keep coming back to. We don’t need to install thousands of sensors around Barbados. We don’t need a multimillion-dollar artificial intelligence project. And we definitely don’t need to sprinkle the letters “AI” over the idea simply because that is fashionable.
The raw material already exists. Insurance companies have accumulated motor claims information over decades. Information such as accident date, time, location, weather, road conditions and vehicle speed can already form part of the claims process.
The first question is whether that information can be standardised, appropriately anonymised, aggregated and analysed across the industry in a way that protects customers while revealing useful patterns.
The second question is even more interesting: Can those patterns actually change driver behaviour? That part should be measured rather than assumed.
Start small. Take several years of historical claims data from participating insurers. Agree on a handful of common fields. Analyse it. Pick one recurring pattern — perhaps a particular day and time window — and run a public-information experiment for several months.
Establish a baseline. Track accident frequency. Try different messages and different delivery methods. Determine whether motorists remember the warnings. See whether accident patterns change. Measure what happens. If it works, expand it. If it doesn’t, the data tells us that too.
After more than 20 years around technology, I’ve learned that innovation doesn’t always mean inventing something new. Sometimes it simply means looking again at something you already have and asking a different question.
Insurance companies traditionally use historical claims data to understand and price risk. Perhaps we should also ask whether it can help us avoid some of that risk in the first place. Use yesterday’s accidents to try to prevent tomorrow’s.
About the author: Amit Uttamchandani is an experienced Caribbean technology and business executive and data enthusiast. Since 2017, he has used caribbeansignal.com to explore the numbers, trends and ideas behind issues affecting Barbados and the wider Caribbean.