The Disaggregation of Search

For the first time in decades, the way we search for things on the internet is being disrupted. Google, which by every measure is the world’s leader in search, has built a remarkable business surfacing information when we have questions. But a series of new innovations, namely ChatGPT and LLMs, coupled with new frontiers in search UI (e.g. chat-based search), are going to rapidly transform how we get the information, and hopefully knowledge, we seek over the coming decade.

I once read that the history (and future) of business is just an endless cycle of aggregation and disaggregation. I went to Google to figure out where this theory came from and the results I got back were not helpful. So I went to ChatGPT and found exactly what I was looking for:

Before Google the way we found information on the internet was by combing through verticalized directories that were compiled by humans working for companies like Yahoo and MSN. Google, through an innovative approach to programmatically and algorithmically indexing the web that yielded results far superior to any other search engine, aggregated the entirety of search such that we only had to go to one place to get the information we needed. Goodbye, directories. Hello, one search bar to rule them all. We’ve seen its evolution over the past two decades and it has had a profoundly positive impact on the world.

Today, it is hyperbolic to say that search is broken. But it is an understatement to say that it could be better. The amount of information available on the internet has grown by many orders of magnitude since Google’s inception, and the way in which we interact with information (i.e. through web and mobile applications) has changed dramatically. It has become impossible for a single search aggregator to answer all of our questions in a systematically excellent and satisfying way.

Over the past several months, a new suite of tools (e.g. ChatGPT and LLMs) and experiences (e.g. chat-based search) have been introduced to hundreds of millions of people across the globe. These are the beginnings of the infrastructure that will power the disaggregation of search. Jim Barksdale once famously stated:  “there are only two ways to make money in business: One is to bundle; the other is unbundle.” It is my belief that search is going to be unbundled over the upcoming decade (and then ultimately re-bundled when AGI can satisfy all of our search desires in a uniquely personalized way, but that’s for another (several?) decades). And it’s going to happen by entrepreneurs building on the aforementioned tools to target underserved and overlooked customer segments that Google cannot prioritize satisfying.

The defining characteristics of these search dis-aggregators are that they will be differentiated from conventional search in UX (e.g. chat-based and other novel approaches), vertical or use-case specific to improve the thoroughness of answers, and powered by LLMs that uniquely improve as usage grows. The tools are finally here to create search and information-finding experiences that are 10x better than what Google can do across a wide variety of verticals. We are beginning to see it with companies like Consensus and Phind and several more highlighted in these good pieces by Connie Chan and Justine Moore at a16z and Talia Goldberg at Bessemer.

I have spent the better part of the past decade building businesses that used Google and SEO as their primary customer acquisition channel. If I were to build something new I’d focus on creating a 10x better experience than Google can for a specific vertical or use case and relentlessly work to establish it as the most trusted brand for providing the right information in a compelling way within our area of expertise. It’s an exciting time to unbundle your bit of search.