2 Feb 2015

The Next Phase of Search

Last week, Google announced that it would integrate third party apps into Google Now. While this may seem like just another update to Google's contextual search offering, it has the potential to deeply influence interaction models and app discovery in the mobile age. But in order to understand its potential, we first need to understand how search evolved on the web.

The History of Search, PageRank and AdWords

The early days of search looked very much like the image above -- a selection of websites sorted by category that you could "search" through based on keywords in the page titles or the description. Search companies like Yahoo!, Altavista, etc. monetized by serving banner ads on these pages. The longer users stayed on the search directory, the more likely they were to click on an ad. The problem for users was that it took long to find what they were looking for.

In order to solve this problem, Sergey Brin and Larry Page invented the PageRank algorithm to rank web pages based on the number and quality of inbound links from other pages. This was to act as a "voting system" to help identify high quality web pages relevant to what users were searching for. They attempted to sell this technology to existing search companies, but ran into a problem. None of the companies were interested for one simple reason -- the goal of PageRank was to give users what they wanted instantly, thereby reducing the odds of a banner ad click.

In other words, PageRank was a superior technology on metrics that may have mattered to users, but it was inferior on the metrics that mattered to advertisers, i.e. PageRank was disruptive. Therefore, it needed to be combined with a disruptive business model in order to succeed. This led to the birth of Google and one of the most iconic online advertising products, AdWords. Instead of serving vaguely targeted banner ads, AdWords served ads based on what users were searching for.

The Next Phase: App Interactions and Google Now

This business model served Google very well until the rise of smartphones. While mobile search is still an attractive revenue opportunity, cost-per-click has dropped as more and more interactions occur within apps. While distribution is centralized through app stores, the discovery model itself isn't very different from pre-PageRank search engines -- scores of "searchable" apps listed under categories.

This discovery problem has created a revenue opportunity for messaging apps like LINE and WeChat. These apps offer a curated list of apps to their engaged user base in exchange for a cut of revenue from app developers. But this just uses a different channel to implement the same discovery model. Facebook has (successfully) attempted to capture usage data and target app-install ads, but users don't really go to the Facebook News Feed to find apps. That's just a happy coincidence. In both cases, the goal is to keep users on Facebook or on the messaging app. More engaged users are likely to use the app discovery channels and generate revenue.

In contrast, the goal of Google Now is to provide users with information or prompts at the right time and place, so users can immediately take action or jump into an app. Google's bet is that, instead of searching for keywords, users may be signaling intent based on their actions. This can then be used to give users "access" to the right app at the right time. At the moment, this only includes a list of about 40 apps that need to be installed on a user's phone but Google could easily open up app integrations to all developers if the model shows promise.

Is there any possibility of creating an AdWords-equivalent business model here? Google's announcement may give us some clues:

Today, you have two ways to get information from these apps—either remember to constantly open them up and look, or get a notification, which you may forget to act on if it shows up at the wrong time.

In the morning, catch up on news of the day with cards from The Guardian. On your commute, Pandora can give you recommendations for music to play, based on what you like, or you can be reminded to complete your daily French lesson on Duolingo. During your downtime, you can take care of the groceries, with a card from Instacart reminding you to stock up on the things you often order. If you’re planning a trip and looked up places to stay on your Airbnb app but couldn't make up your mind, you’ll see Now cards from Airbnb for the location and dates you’ve researched. And when you land at an airport, you’ll see a card to order a Lyft.

In the case highlighted above, would The New York Times, Spotify or Uber be interested in acquiring users of competing apps? In this scenario, advertisers wouldn't be bidding for keywords, but for user contexts, i.e. Uber would have to "outbid" Lyft for a user who has signaled his intent by landing at an airport. The resultant AdWords-equivalent business model could initially manifest itself as "contextual app install ads". However, this business model would remain relevant even if the app interaction model evolves from "installing" apps to "accessing" hosted app store services. In other words, Google could align this new business model with the goal of matching users with the right app/service at the right time. With access to a billion smartphones and growing, reach will certainly not be a problem.