Whilst decompressing in my hotel room during a work visit to Canada a number of years ago, I was struck by a TV ad on the Canadian equivalent of the BBC. The screen was full of a concentrating man with his head held intently next to a radio, listening as the broadcaster read the latest headlines. The camera drew back to reveal that our hero was behind the counter at a newsstand and surrounded by newspapers, presumably current. The ad, for CBC radio, had this simple punchline: “Some people want the news while it’s still news.”

I remembered this bit of advertising cleverness early this week whilst surfing a web page called “Real-time guidance for major hurricane Melissa.” It prompted a question. How real is real time?

The “Late EPS track guidance plotted by model” I clicked on was time-stamped 58 minutes before I viewed it. The “SHIPS Intensity forecast” was from 179 minutes earlier. The latest information on the “CIRA TC Real-time Page for MELISSA” was nearly five hours old.

I don’t mean to criticise the excellent and invaluable work of the people who compile such web pages, but what does “real time” mean, exactly?

When I watched old movies to learn English, Hitchcock was always a favourite. His tense thriller Rope was famously made in “real time,” although the maximum length of a spool of film meant the shots could be no longer than about ten minutes, so it wasn’t acted out in one go.

When I watch F1 in streaming “real time,” my son sometimes (very annoyingly) tells me the result five minutes before the checkered flag falls, because he’s heard it on the radio. And when I listen to “live” DAB radio in the living room, the digital always lags a little bit behind the same broadcast received on FM Radio in the conservatory. I want to believe that “live” is a synonym for “real time,” but both may be echoes that arrive unfashionably late.

So how real is the “real time” data delivery we so often see touted in insurance technology marketing?

Many insurance-platform-building companies (including mine) claim to deliver information in real time. Right now, someone somewhere is using clever software to track the impact of Hurricane Melissa on their risk portfolio in real time. But how timely is it?

The answer usually has very little to do with the output. Instead, it’s like watching a VHS tape on a UHD TV. What goes in determines the quality of what comes out. When data goes in late, it inevitably comes out at least as late, and probably a little bit later (a truth that holds even when real-time delivery is promised).

Why? No matter how fast an exposure management platform, for example, is able to calculate the impact of a loss scenario on a portfolio, then render that impact into the desired format, then deliver it to just the right people, the output can only be real-time minus the age of the data originally ingested. Plus, it can be only as timely as the schedule on which that old data was fed to the system.

In other words, if you’re a CUO looking at a real-time (live?) update of your estimated loss based on five-hour-old data scraped from a real-time page for the event, your updated estimate will be at least five hours old – no matter how quickly it has been calculated and delivered. If the data from that source is updated, say, four times per day, you could be wallowing in an 11-hour out-to-lunch break. That could be the difference between a Category 5 landfall and simply being at sea with your analysis.

In the midst of the reinsurance renewal season, when cat loss budgets all looked pretty healthy until the late-breaking tropical cyclone, real time data may have a big impact on capacity deployment decisions. It should allow you to better manage your portfolio, and the best exposure management systems will illustrate the impact of loss events on portfolios almost instantly. But the updates are only as current as the data delivered, and only as recent as the last data upload.  

After all, yesterday’s news can always be delivered in real time.