Facebook’s new search concept is going to be seen as a threat to Google, and it is. It’s also a threat to Facebook and it may even be a threat to the whole of online advertising. Revolutions have a way of turning from change to destruction, and while that’s certainly not intended here it is a very possible outcome.
The essence of Facebook’s Graph Search is that it looks through social network information and not web pages. The idea is that recommendations from friends, negative or positive, would manifest themselves as product or product class mentions and could then be translated by natural language query processing into results. That Facebook will kick the problem to Bing under conditions where it can’t yield a reliable result is less a problem for Google than the shift from web-forward to social-forward data as the core of product/service recommendations.
The only thing is that in the product review area, 1) Facebook users haven’t been posting reviews in most cases and 2) retail sites like Amazon already have reviews associated with nearly every popular product. So arguably Google has little new risk here.
The flip side is that Facebook may have little upside. Unless they can extract a lot of utility from their social results, Facebook is just creating a front for Bing and clearly they’ll have to revenue-share with Microsoft. How many advertisers might look at the Bing relationship as a cheaper and easier “in” for social advertising? Facebook could lose as much as it gains.
It also appears that the early Graph Search isn’t fully mobile-integrated, and that’s troubling because you have to wonder why Facebook would not have recognized from day one that mobile users behave differently online, and need different forms of advertising support to make decisions. If you’ve been supplying mobile apps, how could you not have realized that you need to identify these posts and treat them differently in analysis? And despite what analysts on the Street think, LBS recommendations first and foremost rely on accurate location data and a complete base of “local” options. Will Facebook have that?
By making a major launch out of what, from a stock perspective in particular, is a bit of a yawn, Facebook may be signaling that it really has no revolutionary ideas on how to monetize itself. Given that reports yesterday indicated it had lost 1.4 million active users, the bloom may be off the Facebook rose in more ways than one.
I also want to offer a follow-up on my blog on Juniper’s SDN announcement yesterday. My main problem was that what Juniper announced was Network Functions Virtualization (NFV) and not SDN, and that’s particularly interesting given that they picked a partner event where strategic initiatives like SDN or NFV are unlikely to be resonant. Reports from the show indicated that many of the attendees couldn’t follow the story at all. There is in fact an NFV kick-off this week in France; why not aim the pitch at that event given it’s an NFV pitch no matter how you title it? Could it be that Juniper believes some competitor is about to jump out and grab the SDN spotlight, somebody like Cisco? It could be, and it would explain why the focus and the forum of the partner event were selected.
I also listened to a later section of the Partner conference to a session on Juniper’s semiconductor evolution. When the new chip family was first presented at the QFabric launch in early 2011, I saw the potential for the chip to be used to link SERVICES into a delivery chain, and asked a question of the Juniper CTO on that very topic. He agreed that the potential was there, and I wrote all that up in March. Juniper’s CTO also gave us a kind of slogan for the future of networking: “Centralize what you can, distribute what you must!” So sense my frustration here; we have had for two years the ingredients of yesterday’s story. We’ve had what might easily be considered the launching slogan of both SDN and NFV. Why after TWO YEARS are we not getting an effective presentation of a product? In fact, the chip story and the “SDN” story were in different presentations. In fact, the “SDN” story evolved from the Universal Edge preso of last October, which at the time had no clear SDN link at all. But then neither did yesterday’s pitch.
Earth to Juniper. Software is software, not boxes. If you want to have a software story, to be a software company, then you have to start by creating a software architecture and tracing its support roots down to the box level. The stuff can’t be presented as ships in the night. This is a simple articulation problem, Juniper. You have the pieces and you’ve had them longer than anybody. You’ve even had the critical insights, as CTO Sindhu proved with the cloud, and with SDN/NFV principles. Why can’t you learn to sing?