There has been an increasing importance laid on the website loading time when it comes to user experience and its potential effect on a site’s Google search engine ranking. This is because Google is becoming serious in the campaign to load websites faster. Remember that in Google Webmaster Tools, a new section has recently been added: Dashboard – Labs – Site performance. It reports site performance in terms of average page load time.
Not only that, but Google has acquired “Page Speed” ( http://code.google.com/speed/page-speed/ ), a Firefox plug-in that will measure a website’s load speed in terms of “Page Speed Score,” a scale of 1-100.
If you are serious about making your websites more user-friendly to both users and to Google, you should be looking for ways in improving the average page load time of your website. A slow-loading website delivers a poor user experience, and websites that provides poor experiences to visitors are less highly ranked by Google.
This article will provide an in-depth study on the average page load time of the top ranking websites in Google. The following are the main objectives:
- To measure the average website loading time response of top ranking websites. This will be used as a standard/benchmarking value to improve other websites.
- To establish an acceptable maximum value for an average website loading time. Those values that exceed the maximum will be considered “slow loading websites.”
A statistically valid sample size of 36 website URLs will be examined. These website URLs rank in the top position for their targeted keywords. Below is the complete list of those URLs:
The browser used is Google Chrome. The tool used to measure website loading time in seconds is: http://www.numion.com/stopwatch/ . However, below are factors that can affect the accuracy of the results:
1. Speed of the Internet connection
2. Browser cache (if the website has been previously loaded, of course it will load fast).
3. DNS cache by ISPs/browsers
To increase the accuracy of the results, the following methods are employed:
1. During the measurement, only one browser and one tab is open. No other Internet activity is engaged in, except for taking the measurement.
2. Clear the browser cache, then close the browser after every measurement (for each website URL).
3. Clean the DNS cache every after measurement (for each website URL also). In Windows this can be done with the DOS command: ipconfig /flushdns
4. Do two trials when measuring the website loading time. Using two trials will improve the accuracy and precision of the measured variable.
In relation to the above methods, the following is the final measuring flowchart:
In Google Chrome, entire clearing the browsing data and cache can be done by using the following procedure:
Step 1: Click the wrench icon (for “Customize and Control Google Chrome”).
Step 2: Click “Options.”
Step 3: In the “Under the Hood,” click “Clear Browsing Data.”
Step 4: Check all items and set “clear data from this period” to “Everything.”
Step 5: Finally, click “Clear Browsing Browsing data.”
Step 6: Close the Google Chrome Browser.
Once the entire process has been completed for all 36 website URLs, it is repeated, because the measurement process consists of two trials.
The data has been gathered and is available here for download: http://www.php-developer.org/data-gathering-average-page-loading-time-of-google-top-ranking-websites/
Once the data gathering process has been completed for the two trials, the average of the two trials will be computed. So for example, if the data for the first trial is 13.56 seconds loading time and for the second trial it is around 14.23 seconds, the average measurement of both trials is (13.56 seconds + 14.23 seconds)/2 = 13.895 seconds.
The average measurement (for example, 13.895 seconds) is the official average loading time measurement of that specific website URL. Since there are 36 unique website URLs, there are also 36 different website loading time measurements.
Once the data has collected for the 36 website URLs, the next thing to do is to average the 36 measurements. Aside from taking the average, a standard deviation (measurement of the variability) is also measured.
Once the computation has been finalized, the analysis worksheet will look like the one below:
The Excel analysis file (refer to the download link provided in the earlier section) shows the following important stats:
N (sample size) = 36
Observed Mean = 11.832 seconds
Standard deviation= 6.147 seconds
%confidence level to be used = 95%
Below is the distribution of these measurements:
The above chart shows roughly that the 36 measured samples result in an average loading time that falls between 7 seconds and 13 seconds (that is where those vertical bars in the graphs are tall).
However, to really estimate the population’s true average loading time (entire websites ranking in the first position in Google), you need to use some statistical tools to estimate the true average. In statistics this is called “confidence interval of the mean” and is discussed here: http://www.itl.nist.gov/div898/handbook/eda/section3/eda352.htm
One of the online tools is available here: http://www.dimensionresearch.com/resources/calculators/conf_means.html
If you enter the statistics we observed from the 36 samples, you will arrive at a result that says the true mean of all of the websites’ average loading time should be: 9.82 seconds – 13.84 seconds at 95% confidence level.
The above range means that if all position 1 websites in Google are measured (there will be many, which is why it is called a “population” and not a “sample”), the average should be somewhere between 9.82 seconds to 13.84 seconds website loading time.
It is also interesting to measure the upper statistical control limit which is the borderline (maximum) of the website’s loading time. The formula will be:
Statistical maximum = Average + Standard deviation
Based on the results, the average measurement is 11.832 seconds while the standard deviation is 6.147 seconds. Therefore the statistical maximum will be:
Statistical maximum = 11.832 seconds + 6.147 seconds = 17.98 seconds
Conclusions and Recommendations
Now that you have all the data and analysis results, it is time to use the information from it to consider improving your actual website loading time. The following are the conclusions and recommendations based on the results:
1. If your website’s average loading time (measured using the methodology discussed in the earlier section) is between 9.82 seconds and 13.84 seconds, then your website URL loading speed is on the “average.” You may or may not improve further as you wish.
2. If your website’s average loading time is below 9.82 seconds, your website is loading VERY fast as compared to the average websites ranking in Google at Position 1. You do not need to improve in this case.
3. If your website average loading time falls somewhere 13.84 seconds and 17.98 seconds, your website loading speed is below average. In this case, you will need to improve further to make your website load faster.
4. Now if your website has an average loading time of more than 17.98 seconds, it is VERY slow. In this case, you will lose a lot of visitors, and that affects your sales and conversion. In this scenario, you will have to make drastic technical improvements to your website in order to make it load faster.
There are several tools that can make your website load faster. These are beyond the scope of this article. Although below you will find useful guides: