Hooked on ChatGPT? Search Habits in the Google Era
A lot of the psychology, product, and UX podcasts I listen to recommend ‘Hooked’ as a must-read. I’m finally reading it, and they were all right; it’s a fantastic, well-researched book. Nir Eyal explains in ‘Hooked’ that it is very difficult for new products that change user habits or routines to succeed.
A product being marginally better isn’t enough; it must be overwhelmingly superior to overcome the inertia of user preference for the familiar. The book vividly illustrates this with the enduring dominance of the QWERTY keyboard. Despite its origins in the 1870s to prevent typewriter jams — a problem that no longer exists — the layout remains dominant. Alternative keyboards promising faster typing speeds have emerged but failed to replace QWERTY. This is largely because switching entails significant effort in relearning to type, and human nature often defaults back to long-established habits.
The Search Engine War
This brings me to the search engine war. ‘Hooked’ frequently references the competition between Google and Bing. Users stay loyal to Google due to their familiarity with it, and switching to Bing or alternatives demands cognitive effort. Googling has now become a habit that people do almost instinctively.
Enter ChatGPT, which allegedly reached a million users barely 5 days after launch and 100 million in two months. Despite its initial version containing bugs and a knowledge cutoff, ChatGPT’s user base continued to grow. Contrary to speculations of it being a mere fad, ChatGPT has sustained, if not exceeded, its initial success. Its only similarity to traditional search engines is in offering responses and having a prompt/search/query box, akin more to chatbots than search engines.
ChatGPT vs. Search Engines
Beyond this, the platform is markedly different. ChatGPT provides direct responses to queries and can do much more than traditional search engines. For average users, it can summarize and edit content, engage in hypothetical discussions, and aggregate information, among other functions. It also caters to specific use cases, such as data analysts running SQL queries, engineers writing code, and even more nuanced API interactions.
However, it’s a powerful tool that is far from perfect. For one, getting precise responses often requires providing comprehensive context. In Lenny’s podcast, OpenAI’s head of developer relations, Logan Kilpatrick, revealed that along with context, nuances like adding a smiley face or using ‘please’ can improve ChatGPT’s performance by 1%-2%. These seemingly minor tweaks require a bit of effort and aren’t common knowledge.
Furthermore, ChatGPT is often criticized for inaccuracies or biases, especially in calculations. Its accessibility for those not tech-savvy or with low literacy (based on my observations, not hard evidence) is another hurdle.
So, how did ChatGPT manage to carve out its niche against Google?
Using Nir Eyal’s insights from ‘Hooked’, ChatGPT seemed destined to fail because it necessitates new learned habits, and Google has been the go-to search tool for decades.
What Do Users Really Want? Understanding Search Behavior
Studies on search behaviors reveal a variety of user needs: most seek answers and prefer search results in different formats (text, image, video, etc). Some users rely on Google’s autocomplete suggestions, related searches, or the “people also ask” feature for queries they can’t fully articulate.
Despite the general goodwill towards Google, some studies highlight user complaints regarding ads and privacy concerns, tied to Google’s use of personal data for tailored search results.
A few studies delve deeper, suggesting a small percentage of users don’t find what they’re looking for, indicated by behaviors like scrolling to the bottom of the first search results page and modifying search queries. Yet, pre-ChatGPT Google search frustrations centered mainly around ads and privacy concerns. A minor percentage might have mentioned the lack of direct answers to their searches if probed, but not to a degree deemed significant enough for researchers to consider it a major problem.
Sure, user researchers or people who do research (PWDR) understand that customers don’t always recognize their wants or needs, sometimes unaware of a problem with a product because they’ve devised workarounds or don’t realize a solution is possible (consider the often-cited faster horse anecdote by Henry Ford).
Yet, the reality remains that few users would have identified the lack of direct responses as a problem. In the business/product world, this seemingly minor issue might not seem worth addressing given Google’s dominance. However, ChatGPT has carved out success by offering direct, customized interactions and more.
This leaves me wondering about the reasons behind ChatGPT’s success. Is it the platform’s perceived or actual utility outweighing the inconvenience of changing search engine habits? Does the association with big names like Elon Musk and Sam Altman play a part? Or is it the diverse applications beyond direct responses? The true answer is likely a mix of these reasons and other unexplored factors.
Sorry to those expecting a definitive explainer — I am merely curious. Thoughts on this topic are welcome, whether in agreement or not.
