Dear subscribers,

My first tech job was at Microsoft working on Bing. Back then, it felt futile trying to eke out single digit market share gains against Google’s monopoly.

A decade later, search as we know it might get disrupted.

I’ve been using the new Bing and its chat interface already gives me better answers than the search results page for many queries. Let’s cover:

• Where traditional search is falling short

• Microsoft’s “copilot for the web” vision and Google’s innovator’s dilemma

• What it’ll take to win the AI search race

Where search is falling short

Search is arguably the most profitable internet business of all time. For two decades, Google has dominated this market with 90% share and 40%+ profit margin.

But search also hasn’t changed much in 20 years. You still type in keywords to get a list of blue links. If anything, search has devolved with far too many ads and SEO-optimized pages. Perhaps the best example of this is food recipe queries:

Peter Yang @petergyang

Biggest pain point that AI can solve is so I never have to read a recipe website with tons of SEO optimized content before the actual recipe again

1:06 AM ∙ Feb 8, 2023

---

196Likes10Retweets

Search also struggles with:

1. Open-ended questions. If I type “Help me plan a Japan trip”, search doesn’t return a great response because there is no one right answer.

2. Context. If I follow up with “great attractions for kids”, search doesn’t remember that I’m looking for attractions in Japan.

That’s why people have been using hacks like adding “reddit” to queries to get better results. The thing is:

Which brings us to Microsoft.

Microsoft’s “copilot for the web” vision

Microsoft is going all in on AI chat search with its “copilot for the web” vision.

Below’s an example of Bing helping me plan a winter trip to Japan:

Even though it’s early days, AI chat’s potential to address search’s flaws is clear:

1. Open-ended questions get concise answers. For example, I can get a personalized itinerary for kid-friendly activities in Japan.

2. Context is preserved. I can make a follow up request to “Stay in Tokyo and Kyoto hotels only” and it’ll update the previous itinerary.

Of course, AI chat is not perfect. Flaws include:

1. Overconfidence. AI chat will sometimes confidently give the wrong answer. Bing addresses some of this by including source links in the replies.

2. High costs and slow speeds. AI chat costs an order of magnitude more per response than regular search. These costs will likely come down over time.

3. Publisher and advertiser risk. AI chat could hurt publishers and advertisers if users can get the answer without clicking any links.

All of these risks are very real. Nevertheless, it’s easy to see why this experience might be better for end users:

Microsoft, of course, is aggressively going after this space under Satya:

Peter Yang @petergyang

Damn - Satya is ruthless

11:37 PM ∙ Feb 8, 2023

---

2,910Likes211Retweets

The company has a lot to gain by going after Google’s core business (80% of Google’s revenue is from search). Even a modest gain in search market share will result in billions more dollars. Microsoft Azure also has a lot to benefit from powering OpenAI’s models. Indeed, Satya wants to make Google dance:

So how can Google respond?

Google’s innovator’s dilemma

The innovator’s dilemma is when companies don’t want to risk their existing business to go after an innovation that’s at first inferior. The innovation gets better over time and ultimately disrupts the incumbent’s business entirely.

Today, AI chat is slower, more expensive, and worse for publishers and advertisers than regular search. But for open-ended queries, AI chat is showing promise to be better in the one dimension that matters - giving end users the answer.

Google understands this risk. It has declared “code red” internally and announced a competing AI chat product in Bard. The company still has huge moats, including:

1. Brand that’s synonymous with search.

2. Customer base (users, advertisers) that’s an order of magnitude larger than Bing.

3. Distribution via the leading browser (Chrome) and OS platform (Android).

4. Top talent and data in AI, maps, and more.

But instead of resting on its laurels, I think Google needs to disrupt itself before Microsoft or another company disrupts it.

Winning the AI search race requires walking a tight rope between the needs of users and advertisers + publishers

I’ve written about obsessing over customer problems and building with community before. Winning the AI search race is no different.

For both Microsoft and Google, making this transition successfully requires balancing competing needs between two different types of customers:

1. Users just want the answer. Ideally, they want the answer directly in the chat conversation instead of having to click a link.

2. Advertisers and publishers want users to buy their products and services. Today, that means getting users to click their links in search to visit their webpage.

To satisfy these competing needs, both Microsoft and Google have resorted to compromise solutions like showing the chat interface next to the traditional search results page…

…and displaying traditional ads in the chat results:

I think the winner will be the platform that moves beyond these compromises.

One thing is for sure - it’ll be fascinating to see how this plays out.

You can access Bing’s AI chat by following the instructions here. It has already replaced ChatGPT as my daily AI assistant.