Several years ago, when I still worked at ITV, a colleague in the press office was desperately trying to tell a journalist that a programme he was about to label a flop was actually pretty successful when put in context. The show had held up well on the +1 channel but, what's more, was one of the most successful shows the network had seen on online catch up, pulling in some seriously impressive numbers.
The journalist, however, was unmoved. "Nobody," he informed my colleague, "cares about online numbers."
Netflix, you suspect, would beg to differ.
Even five years ago - an age in digital terms - it was obvious to see what a huge part online, on demand and data would play in the future of TV. The numbers kept going up and the insights led to some rather fascinating discoveries.
Of course, there's always somebody ready to step in and show you how to do things better and recently Netflix has been making a big play about the mechanics behind the site, as it continues its aim to ascend into the same league as the major broadcasters (assuming, by whatever terms its judged, it hasn't already got there).
The online streaming site's recently victory at the Emmys for House Of Cards and the critical acclaim lavished on Orange Is The New Black, both original and expensive commissions from Netflix, suggest that the company is flexing its way into the establishment. But a look behind the commissioning process reveals a more data driven way of thinking.
It seems quite basic to say that because their data told them that Netflix viewers who liked the original BBC House of Cards series also liked David Fincher and Kevin Spacey - the remake's director and star, respectively - the company decided to invest in the new commission.
But it's the ability of Netflix to investigate at a "molecular level" that provides the real insight to getting viewers hooked. As this article notes, the site can tell when subscribers fast-forward, binge view or abandon a series and provide, with a degree of confidence, when a series works and what will hook each individual subscriber's attention next. With the right analysts and decisions that's powerful stuff.
Of course all this should be qualified. Firstly, Netflix don't release their data, so it's unclear how many watched House Of Cards, let alone some of the more interesting data from the site.
Secondly, broadcasters have a degree of insight into this. Similar data is available for the iPlayer, ITV Player, 4OD and Sky Go, while it's fascinating to see how usage of the BBC Radio Player app breaks down.
But these are just catch-up and live streaming services not live viewing through the TV itself. It will give a reasonable degree of insight but this will naturally be skewed towards the catch up demographic, not necessarily the more traditional TV viewer.
Instead, live TV is, for the moment, reliant on BARB, which relies on a sample of households to estimate viewing figures. In this day and age that's an awful lot of data missing.
It's why I'm quite excited about the next generation of TVs and viewing devices. For broadcasters and advertisers, data around all aspects - from overall numbers to the way people actively engage with programmes is incredibly useful.
Better data should, theoretically, lead to better programmes (although there's also a danger of it leading to common denominator programmes too). It at least should lead to better insight, recommendations and user experience across broadcaster websites (I'm personally a big iPlayer fan).
If - and it's a big if - audiences are happy with broadcasters tracking this sort of data through the living room TV then this has the potential to seriously alter the way broadcasters approach programming. And the journalist who harangued my colleague all those years ago will, you suspect, be asking rather different questions about online numbers in the years to come.