By

Emiliano Pasqualetti, CEO and Chief Data Scientist
and Daniel Ruzzini Mejia, CO-Founder and CSO

Analyzing anonymized data from the DomainsBot Name Suggestion service we wanted to investigate how users search for domains, especially in relation to the availability of what they search as a .com domain and how their behavior changed during the last year. 

Our goal is to identify trends that can help us improve our service for the benefits of our own users and the internet community at large. The takeaways of this first study are:

      1. The number of individual users searching for a domain grew by 37% since 2020. However, users are submitting significantly less queries per session.
      2. The number of people searching 5 times or less during a search session grew by 33% while those who searched 6 or more times shrunk by 26%.
        At this time we do not know if people are finding what they want faster or if they are giving up faster than before.
      3. The chances of finding an available .com matching exactly what the user searched for are less than 13% for a 2 word query and less than 40% for a three word search. Users need to type a query of at least 20 characters to increase the odds to 50%. These results are in line with last year.
      4. Users tend to search for domains with a certain number of words consistently. About 75% search for domains made of 2 or 3 words and they tend not to add additional words to increase their chances of finding an available .com domain to register.

In conclusion, we observed a significant decrease in the average number of queries per session as well as a relatively low but stable chance of finding an available .com within the most popular search groups (2 or 3 words). We have also noticed little user interest in increasing the odds of finding an available .com domain by searching for longer queries or more words than their original idea.

We want to keep investigating how users search, register and use domain names. In particular we plan to focus our next studies around the following questions:

  1. How did the evidence of less queries per session affect registration trends?
  2. How many users register and develop a domain in a non .com extension even if the .com equivalent is already registered (possibly by someone else)?
  3. Is a synonym based domain suggestion effective to alleviate the user pain of not finding what they search for available as a .com?
  4. Is there evidence that a growing number of users look at the secondary market when they cannot find what they are looking for as a .com domain available in the primary market?

We believe the answer to these questions can help our industry understand how to address potential pains for the consumers.

Introduction

We want to investigate the behavior of users searching for a domain name in the primary market. In particular we want to understand what are the chances that a users find exactly what they are looking for as an available .com according to:

    • The query number within her session (first, second, third query etc…)
    • The number of words in the query (e.g. “HealthyVeggies” counts as a two words query)
    • The length of the query (e.g. “HealthyVeggies” counts as 14 characters)

We also want to compare these numbers with the previous year to understand how the .com availability is changing the user behavior. 

Naming convention

  • Exact Match We define an exact match as a domain whose SLD matches exactly what the user searches for. In case the user searches for a full domain the comparison is on the SLD alone. E.g. The domain “example.com” is an exact match for the queries “exact”, “Exact”, “EXACT” or “exact.live” but not for the query “Exact123” or “Ex-act”.
  • Query We define a query as the string or words that a user types when searching for a domain.
  • Query number Users typically search more than one query during their search session. We refer to the query number as the number indicating the order in which the user searched.  For example, the first query in a session will have the query number set to 1, the second to 2 and so on.
  • Domain availability In the context of this study we will refer to a domain availability in reference to the exact matches only and at the time of the query.

Methods

We have analyzed an aggregated and anonymized set of requests to the DomainsBot Name Suggestion from May 1st 2020 to May 31st 2020 and from the same period in 2021. The majority of requests were in the English language, followed by Italian, Spanish, Portuguese and French.   From this initial dataset we have discarded requests matching any of the following conditions:

  • Request without an anonymized session ID
  • Requests not allowing suggestions for a .com exact match.
  • Requests from a session ID used more than 1000 times in a month 
  • Requests for a query searched more than 40 times in a month

For the purpose of this analysis we have excluded dashes from the word count (e.g. “Healthy-Veggies” would count as two words).

Results

 

Part 1 – Number of queries per search session

Chart 1 – distribution of queries per session

Study - Chart 1

The number of individual users searching for a domain grew by 37% since last year.  However, on average, each user performs less searches per session today than in the previous year. In May 2020 about 20% of the users searched only once. This number jumped to about 30% in May 2021. We have trimmed the series in this chart to 20 queries per session because the trend continued to decrease without any further changes. Please note some users searched hundreds of times during the span of the observed month.

Chart 2 – grouped changes in distribution of queries per session

Study - Chart 2

Chart 2 shows the changes in the distribution of the queries of users searching for up to 5 times or more than 5 times, respectively, within a session.   The amount of queries performed by users searching up to 5 queries in a session grew by 33% in 2021 while the queries from those who searched 6 or more times in a session decreased by 26%.

Chart 3 – changes in distribution of queries per session

Study - Chart 3

Chart 3 shows the growth of the queries performed by groups of users who performed exactly N requests during the month of observation. 

The percentage of queries performed by users searching just one time grew by 55% (30.65% in 2021 vs 19.73% in 2020) followed by the queries of users searching just two times per session (24%) and then three queries per session (14%). The rate has been instead steadily negative for each cohort of queries performed by users searching more than 6 times during their session. It is interesting to note the growth trend is a monotonic function (meaning it is always decreasing).

Part 2 – .com availability per number of words or query length

Chart 4 – Chances of users finding exactly what they searches as an available .com by number of words

There is a strong positive correlation between the number of searched words and the chances of an available .com exact match. In a nutshell, the more words are searched, the more likely a user is presented with an available .com. It is interesting to observe that users searching for 2 word domains have less than 13% chances to find them available as a .com exact match and less than 40% chances when searching for 3 word domains. The odds did not change significantly against 2020.  Please note that the vast majority of available one word queries were IDN (domains containing non ASCII characters).

 

Chart 5 – Chances of available .com by length of the query

Study - Chart 5

Similarly, the chances of an available .com domain increase with the length of the query measured in number of characters. Users need to type a query of more than 20 characters to increase their odds of finding an available .com exact match to more than 50%.  No significant changes occurred between 2020 and 2021.

Part 3 – Attitude toward changing search pattern during session

Chart 6

Study - Chart 6

This chart shows the distribution of queries by number of words. It also shows how users did not change their attitude toward their preferred number of words to use while searching for domains between the two periods. 75% of the times, users search for just 2 or 3 words domains with a clear preference for 2 words. No significant changes seem to have occurred during the last year.

Table 1

avg. words per query 2020 2021 difference (year over year)
first query of the session 2.39 2.38 -0.42%
last query of the session 2.47 2.47 0.00%
difference (last vs first) 3.35% 3.78%  

It is surprising to observe that users tend to have a consistent attitude towards their preferred number of words in a domain.  For instance, a user searching for a 2 word domain might also search for a 3 word domain but usually keep looking at 2 word alternatives afterwards.   The average number of words per query seems to grow between the first and the last query of the session but by a minor 3% to 4%.   No significant changes year over year.

Chart 7

Study - Chart 7

In Chart 7 we plot the average chance of finding an available .com exact match according to the query number.  This chart is the most insightful of this study.   The availability of the .com exact match is remarkably consistent across every group of queries within a session with an average of 21.7% in 2020 and 22.4% in 2021. It appears that the majority of users performing a single lookup have the lowest chances while those who perform at least 2 and less than 10 tend to have better odds.    This chart also confirms the evidence that users tend not to increase the number of words during a search session. If that was the case we would expect the chances of available .com exact matches to increase with the query number.

Conclusions

In conclusion, we observed a significant decrease in the average number of queries per session as well as a relatively low but stable chance of finding an available .com within the most popular search groups (2 or 3 words). We have also noticed little user interest in increasing the odds of finding an available .com domain by searching for longer queries or more words than their original idea.   We want to keep investigating how users search, register and use domain names. In particular we plan to focus our next studies around the following questions:

  1. How did the evidence of less queries per session affect registration trends?
  2. How many users register and develop a domain in a non .com extension even if the .com equivalent is already registered (possibly by someone else)?
  3. Is a synonym based domain suggestion effective to alleviate the user pain of not finding what they search for available as a .com?
  4. Is there evidence that a growing number of users look at the secondary market when they cannot find what they are looking for as a .com domain available in the primary market?

The answer to these questions can help our industry to understand how to evolve to address a potential pain for the consumers.

Finally, we are aware that 2020, the first year of Covid-19, was an exceptional year, including for our industry and that this study should be expanded taking in consideration additional years in the past and the future.

For any questions please contact us at: insights@domainsbot.com

Compliance

This study uses anonymized and aggregated data collected from DomainBot’s proprietary Name Suggestion service.   In compliance with GDPR, no personally identifiable information has been stored nor used during the study. We receive and process a fully pseudo-anonymized session ID in compliance with both privacy and security requirements.    The actual content of the queries was not processed directly during the creation of this study. The number of words, their length and the .com exact match availability were inferred, when possible, from the logs of our response to the service request.