A never-before-seen study of more than 332 million Google queries has just shed light on users' search habits. This in-depth analysis, carried out between January 2023 and September 2024, reveals surprising data about how Americans use the Google search engine.
Key takeaways:
- More than 44% of Google searches concern brands, contrary to common beliefs that suggested a higher proportion.
- A third of searches are purely navigational, with users using Google as a homepage to access their favorite sites.
- Entertainment overwhelmingly dominates searches with 25% of the total volume, far ahead of other categories.
- Between 60% and 70% of the total search volume has monetization potential.
An unprecedented deep dive into Google's data
This study, conducted in partnership with Datos (a Semrush company), is based on a representative panel of 130,000 American devices over a 21-month period. It is the first time an analysis of this scale has been carried out on Google users' search habits.
One of the most striking aspects of this study concerns the distribution of queries. The data indeed reveal a massive concentration of search volume on a relatively small number of terms. The 10,000 most-searched terms account for 46% of all searches, 148 of which make up nearly 15% of the total volume.
Among these popular queries, unsurprisingly one finds major sites such as YouTube, Gmail, Amazon, Facebook, and ChatGPTThis concentration reflects a growing trend: web traffic is increasingly focused on a limited number of sites, and Google plays a predominant role in this centralization.
A rigorous methodology despite certain limitations
To analyze and classify more than 320,000 different search terms, the researchers had to innovate in their methodological approach. They turned to language models (LLMs), and more specifically to GPT 4o-mini via the ChatGPT API. After numerous tests involving different models (Claude, ChatGPT, Gemini, and Mistral), GPT 4o-mini proved to be the most effective, achieving a remarkable accuracy rate of 96%.
The classification process took place in several stages. First, 1,000 queries were manually classified according to various criteria: brand versus generic terms, search intents (navigational, informational, commercial, or transactional), and thematic classification into 24 categories of interest. This dataset was then used to refine the prompts used for automated classification. A major finding of this process was the importance of using more descriptive classification categories to obtain better results.
It is nevertheless important to note certain limitations in the data collection.
Indeed, the study does not include:
- Searches made via the Google app
- Specialized searches (Google Shopping, Google News, Google Hotels, Maps, Images, Videos, etc.)
- The negative click searches, that is, searches where Google displays instant answers in the search bar even before the query is completed
- Searches made via Google's Gemini AI tool
Although important to mention, these limitations do not invalidate the study's conclusions. On the contrary, they help better contextualize the results and understand their real scope. Despite these restrictions, the sample of more than 331 million searches remains sufficiently large and representative to draw meaningful insights about users' search behavior.
The mystery of the long tail finally solved
The long-tail analysis of searches reveals very interesting data. For the month of September 2024 alone, out of more than one million search terms used, 59% were searched only once. More precisely, these 612,981 unique searches represented only 2.2% of the total search volume.
This extreme distribution highlights the following phenomenon: on average, each panel user performs 5 searches that no one else performed. The study suggests that these could either be legitimate one-off searches or the influence of a small number of users with particularly atypical search behaviors.
This finding puts into perspective Google's claim that " 15% of daily searches are new ". The study also suggests that these new searches come not so much from the long tail as from daily news : new music albums, video game launches, natural disasters, unexpected political appointments and thousands of other events that generate unprecedented searches every day.
If we exclude queries with fewer than 100 searches over the 21-month period, users perform an average of 121 Google searches per month. Including all long-tail searches, this figure would reach about 200 searches per month per user.
Search intents decoded
This study also provides a precise insight into the real intentions of Google users. The distribution of searches reveals four major categories of intent:
- 51.8% of searches are informational, that is to say users are simply looking to obtain information on a topic that interests them or that they need
- 33% are navigational : these users already know where they want to go and use Google as a faster, safer, or more efficient way than typing a URL directly
- 14.5% are commercial, without necessarily implying an immediate purchase intent. This category mainly includes product comparisons and searches for product information, both in B2B and B2C
- 0.69% are transactional, indicating a clear intent to buy something, sign up for a service, or hire a provider
Note that the distribution of search volume differs significantly from the distribution of unique keywords. Navigational searches, in particular, account for a much larger share in terms of volume than in number of unique queries. This difference is explained by the frequent repetition of the same navigational searches among users who want to access their favorite sites.
The dominance of entertainment and brands
Contrary to popular belief, the entertainment-related categories overwhelmingly dominate Google searchesActors, films, TV series, musical artists and video games represent about 25% of total search volume. This finding would explain why Google invests so much in the user experience for these types of content.
Furthermore, the study reveals that 44% of searches concern brands, a figure lower than initial estimates but still significant. This data suggests that Google has become the default "homepage" for many brands, highlighting the crucial importance of managing online reputation.
Implications for digital marketing
These different findings can have major implications for marketing strategies. In essence, the study suggests that it is now more advantageous to be the searched-for brand more advantageous to be the searched-for brand than to try to rank for millions of query variations. Google has become a place users go after identifying a need, rather than a space for discovering new brands or services.
Furthermore, between 60 and 70% of total search volume presents monetization potential. However, domains such as arts and entertainment, gaming, education or adult content, although representing a significant share of searches, remain difficult to monetize for the majority of site publishers and businessesFor example, it is difficult to sell derivative products to people simply searching for information about their favorite video game or celebrity.
The analysis also highlights theevolution of the search ecosystemGoogle's direct answers, including AI previews, already dominate several of the largest search categories, notably in the fields of arts and entertainment, gaming, sports, finance and reference content. This trend could intensify, pushing companies to diversify their visibility strategies online beyond traditional organic search.
Once considered an excellent starting point for marketing, search now appears more as the reward for effective marketing carried out on other channels. Discovery journeys have shifted to social networks, YouTube, podcasts, newsletters and events, suggesting the need for a more diversified marketing approach that is less dependent on SEO.
The article “Exclusive study: what 332 million Google searches reveal” was published on the site Abondance.
