When Google announced Search, plus Your World, they made a dramatic decision: they’re embracing the paradigm of a web of people, rather than a web of verbs. And it’s a dangerous move.

There are two broad strategies for improving search engines: making them more aware of intents, or making them more aware of preferences. Most of Google’s design and algorithmic moves have been in the direction of intents: Local captures the intent to travel somewhere nearby; Flight Search captures the intent to fly somewhere; and “query deserves diversity” pushes Google to identify subsets of intents and display as many of them as possible on a search page, rather than naïvely ranking by pure textual and link-based relevance.

Social signals work differently: they do take the naïve approach, but the weight each signal based on the preferences (search behavior, click behavior, and link-sharing behavior) of people in the same social network.

In some ways, social signals just mean doubling-down on Pagerank: ranking pages purely by “Best”-ness, rather than returning ten results for ten different intents behind a single keyword search.

That’s the broad bet they’re making. The narrower bet is that a convenient, mediocre search engine and a convenient, mediocre personal search engine will supplant competitors a few keystrokes away.

It’s a classic Google bootstrap: they built their social search capabilities in part on Twitter’s data, then built their own in-house dataset to search instead. Now, they can provide a weaker experience in a familiar interface.

It’s also a classic platform-domination move. Using a broadly popular product to push a narrow one is what gave Microsoft their browser beachhead (and, more interestingly, their office beachhead—MS Office’s momentum is certainly helped by the huge mass of task-specific VBA macros written to sling data back and forth between Word docs and spreadsheets). But Microsoft was competing in a market with low adoption; Facebook’s monthly unique visitor count approaches 50% of the population in several countries.

It’s far too early to say whether or not this will work. It probably doesn’t make Bing the best search engine. But it certainly narrows the gap. Google has decided that being a worse search engine can bootstrap them into owning identities and social streams. Winning will be tough, but retreating will be murderously expensive. (Imagine how much Twitter will charge for data when Google desperately needs social data and can’t harvest it themselves.)

Who wins from this move? Google has made their search engine worse for many day-to-day purposes, so winners will include:

  • Vertical search: Facebook and LinkedIn (for people search), Kayak (travel searches), Amazon (product searches).
  • Vanilla search: Bing.

Losers:

  • Name/ search: LinkedIn, Ancestry.com, Twitter, private companies like MyLife.
  • Social search: Greplin, Foursquare. These companies don’t have the footprint Google does, and haven’t identified themselves (as a search engine in Foursquare’s case, or as a brand in Greplin’s).

What to watch:

The jury’s very much out on this one. Account growth on Google+ is one weak proxy for the success of this campaign, and post volume (harder to measure but much more important) is another. Once Google brings back true real-time search, it will be feasible to estimate changes in search volume by repeatedly running the same queries and looking at the timestamps of Google+ results.

Meanwhile, there are two easy measures of defeat. One is if Bing gets a bigger share of the market, specifically at Google’s expense. And another is if Twitter and Facebook continue at roughly their same trajectory. For Google to really win here, they need to destroy the #1 and #2 social media sites. Being #1 at search is not necessarily enough to do that.

One Response to Google Can Win Personalization, or Lose Search

  1. [...] cynical version: as we’ve previously argued, it’s Google’s classic embrace-extend-extinguish strategy. Build up the infrastructure [...]

Share Let others know too.
Google Can Win Personalization, or Lose Search