Commentary: Don't Believe the Video Quality Hype
I’m going to take a radical position in the following article, both for rhetorical effect and to challenge conventional wisdom. It’s not a position I endorse wholeheartedly, but it’s one worth pursuing to its extreme to expose some of the flaws in the opposite belief, which is so commonly held that it’s practically accepted as gospel.
I am going to flirt with some heretical thinking that would, had streaming existed a few hundred years ago, have gotten me burned at the Streaming Media stake. However throughout i ask you to remember that the earth is no longer flat and moons orbit planets other than the sun, and if you really beleive quality of service is the only important influencing factor in people’s consumption of video, then frankly I have probably lost you already.
So let’s get down to it.
The only people who really give a damn about time to first byte and latency are the CDNs who are selling you the idea. Frankly, wholesale reselling of IP and commodity cloud compute services focussed on streaming is utterly uninteresting without the constant measurements of miniscule performance differences (most of which are truly out of the CDN's control).
What’s more, the only people who give a damn about “quality”—in terms of video compression—are those who are selling retina screens, 4K cameras and 48-foot front-room projection systems, or managing brands. No one else cares. No one really thinks worse of your brand, or loses interest in the engaging plot of a well-written story, because of a little articfacting in the periphery of the field of view.
No one cares about about short startup delays when initiating the next episode of thier current binge-watching endeavor, and, contrary to popular marketing, no one really has an expectation that they should be able to channel surf in an OTT streaming environment, precicely because it is NOT broadcast TV; the process of “Click…that’s crap...click...that’s crap...ad infinitum” has been replaced by a process of “Click...search...that’s what I want to watch…click...skip ad...watch.”
So there. I said it. I can envision my erstwhile peers in the sector writing in the comments box: “"What the hell are you on about Dom???"
So now you want some logic to pick over, do you?
Let's start with a point that has troubled me for a while: Advertising. There are plenty of carefully compiled data sources—surveys of at least two of the consultancy's paying clients' clients, or experiemental analysis of quadrillions of user sessions, that claim a miniscule increase in startup delay of an advert reduces click-through rates by a small nation’s GDP. And I am sure it does.
But in reality, no one gives a damn for your advert, and at the first possible opportunity they will click “skip ad” or otherwise ignore it.
So lets make it clear that if you are going to jump from a 1% to a 2% click-through by starting your advert sooner, you are in reality wasting only 98% of your money and not 99%. And quite probably most of those click-through responses are mistakes where people have missed the “skip-ad” button.
And while ad-supported free-to-consumer services to be exploding around you newbies to the streaming sector, after watching user patterns closely for nearly two decades it is clear to me that punters are quickly launching themselves into subscriber VOD and audio on demand models as soon as they become available, in a desparate at attempt to run screaming away from the tedium of advertising and all its appaling, distracting and annoying crassness. There is simply nothing “quality” about advertising—no matter if it is delivered with 125ms startup delay or 35ms. In fact the only real purpose it is serving is to create a two-tiered market: those who can afford multiple subscriber accounts (who subscribe thier way out of adverts but are actually the ones with money to spend) and those who cannot afford multiple subscriber accounts, and who are flooded with adverts for products that they cannot afford.
And, interestingly, that subscriber service engages the user in a totally different way. When was the last time you sat waiting for your Netflix stream to start and were impressed when the next episode of Breaking Bad or House of Cards started up a whole 95ms faster because you happened to pick it up from a better CDN path? That difference—the one the CDN has sold as their raison d'etre—really, really...doesn't matter one tiny bit. Indeed when it comes to watching the “next episode” in a series, frankly most viewers will wait not milli or microseconds, but minutes. “Click...next episode...pause....go to kitchen for coffee or glass of wine... return...play.”
Yes, in practice startup latency is of course something we want to keep reasonably low. But if you listen to many vendors in the sector, you might believe that if you are not “predictively” starting a video stream before the user has actually requested, it you are never going to reach or engage that audience in any commercially meaningful way. The truth is that the only part of the audience who will turn away from their binge, or churn away from thier subscriber service because of the type of differences that you may see between two CDNs, will be pedantic streaming media consultants. The other 98% of the general populous will be checking Facebook or Twitter on their phone—itself nearing “first screen” status in my opinion—while that stream is starting.
Pragmatism about buffering and start-up delay is already the norm among end users; if it wasn't, this industry would never have become mass market.
It has been nearly 20 years since I started watching small, postage stamp-sized movies on my dial-up connection at 56Kbps. It was in the days when the deluded among us were buying premium hi-fi CD players with anti-vibration mounts (you know how that bass vibration just gets encoded on the fly by the laser that is reading the CD and gets fedback back into the system, eh??).
Clearly unable to sell vibration mounts for digital hi-fi components any longer, the same piranha-like sales people turned thier attention to selling bigger and brighter, blacker and thinner, and ever-higher resolution screens. It’s not that we had a problem before, but by creating a problem for us—a perceived short-deal on the “quality we should expect”—we could go out and buy a swankier status symbol TV than our friends and neighbours had. And this had benefits because to get the new picture out to the audiences on their new TVs, the same sales people got to sell new cameras, editing tools, and fancy bits to the production workflows. Ka-ching!!!
Of course, in the traditional TV world, this helped to create tonnes of defunct electronic waste which had to be land-filled and replaced everytime there was even a small change to how the signal was transmitted, where in the IP streaming world we tended to carry on pointing our media player at a URL and clicking ''full screen"—and our machine would magically scale without needing to be replaced every time a new video aspect ratio was produced. Streaming has proven to be a right pain in the backside to those who had relied for years on incompatibility between different generations of distribution paradigms to drive churn in the TV hardware market.
Now, a little olive branch to my peers: Obviously, if the picture is clear and sharp, that is better than low-res and blocky or artifacted. If the startup time is short, that is better than a lengthy wait for things to kick in. But what will drive much more consumer satisfaction is if the discovery process and UI is simple and engaging, and if the quality of storytelling in the video material captures the audience’s attention.
With audiences now able to be more fickle, they are clearly making easy-to-execute choices about where they engage with curated and relevant content. Importantly, many reports show that a significant proportion of internet traffic is pirate content: and this has often been attributed simply to the fact that it is easier to find than legal content.
So, yes (my final olive branch) obviously the delivery of streaming has to at its very least be “good enough”– but over-engineering the quality at the expense of choice and discoverability of good enough content for the end user must, I repeat, MUST be avoided.
Always invest in better content and discovery models as your priority to focus on customer acquisition. Then, as a retention strategy, reward your subscribers with better quality as they become loyal.
Taylor Swift pulled off a publicity stunt when she complained about free music services. Maybe the industry needs a way to stop her from cheapening its technology.