Dual is a Four-Letter Word, Part I

Oct 13, 2011 by Phil Simon in Data Integration, Data Management

I’ve railed before against dual data entry with corporate expense reports. Turns out that I’m hardly the only one that despises splitting things into two–especially when those two things used to be one and the same.

Case in Point: Netflix. In case you didn’t hear, the ultra-popular company recently made the unpopular to split its business into two parts:

• Netflix for DVD rentals via mail
• Qwikster for instant, streaming videos

Note that the company decided to reverse course after I wrote this post.

Offended at the price increase, more than 1 million subscribers quickly canceled their Netflix subscriptions. Reed Hastings, the company’s CEO and founder, had to issue an apology of sorts, prompting even more acrimony.

NY Times’ Technology columnist and prolific author David Pogue recently wrote:

I could not have put my reaction any better than this e-mail from a reader:

Netflix’s split into Qwikster is ridiculous. I wasn’t mad at them before; I am now. I don’t want two companies, two credit card charges, two Web sites, two logins, and TWO queues to maintain.

Now I have to add movies to both an instant and a DVD queue? I have to check both sites to see whether a film’s available by instant or by DVD? I have to sync my iOS devices to two different services for instant film delivery and queue management?

This is ridiculous. Life’s complicated enough already. Netflix (excuse me, “Qwikster”) just made it worse. Does Mr. Hastings thinks we’re idiots? Making it two companies, with two charges, doesn’t change the essential price dynamics; we can add.

Was Hastings ultimately right? Who knows? While I’m pretty opinionated, I sure don’t profess to know more than Hastings does about his company and customers, despite having used Netflix off and on over the past ten years. But I do know a thing or two about data management and, from that perspective, this was an ill-advised move.

Netflix’s technology and data management has historically been nothing short of amazing. Collaborative filtering has allowed the company to consistently suggest relevant movies to users, making Netflix even more accurate and sticky. If you like The Godfather, Netflix would recommend Goodfellas. People who enjoy Memento often like The Usual Suspects.

You get my drift.

The new two-headed Netflix will doubtless have data matching problems–and who knows if the company will even try to marry Qwikster users with Netflix ones. It has essentially taken relatively pure data and contaminated it in one fell swoop. Plus, collaborative filtering benefits a great deal from the law of large numbers, something Hastings has just compromised.

Bad idea.

But there’s another, non-data management reason that this move is hard to justify. In the Age of the Platform, customers want convenience. Chopping Netflix up into two discrete sites, bills, and queues ultimately defeats that purpose, as the above angry letter manifests. We customers enjoy one-stop shopping. Imagine, for example, having to create a separate Google profile while on your Droid? What if your iPhone required a new sign-on and password, one different from the one on your Mac or PC?

Simon Says

The word dual is a four-letter word. For all sorts of reasons, try to consolidate as much as you can. If you absolutely must bifurcate your data, make sure that there’s a very good reason (read security, government regulation, etc.)  And be prepared for a backlash.

Feedback

What say you?

Tags: collaborative filtering, Data Integration, Data Management, netflix