How do you know which pages performed poorly for your users and their associated timings? The Performance tab gives an overview of page speed, load times, the most requested pages and the slowest pages.
Five handy metrics at the top of the screen provide a quick snapshot of information:
This Page Speed reflects an ‘average of averages’ by combining the median, average and 90th percentile load times to smooth outlieres in the data and give a high-level indication of the page speed. The categories are: Excellent, Good, Needs work and Poor.
The Avg Load Time displays the site’s average page load time. It is measured by collecting load times at regular intervals across the date range. The load times are then added up and divided by the number of load times collected.
The Median Load Time shows the 50th percentile load time across the date range.
The P90 Load Time shows that 90% of your users experience this load time or less, within the date range.
The P99 Load Time shows that 99% of your users experience this load time or less, within the date range.
In the Most requested pages table, you can view the most requested pages in descending order, starting with the top ten pages. The median load times can be seen in the color-coded segments on the right side of the table:
The graph displays the breakdown of load times, in colour-coded segments, so it’s easy to see how long each part of the page load takes. You can hover over a color-coded segment to get the load time details - this is similar to the detail you get from the network tab in the browser developer tools. The load time details can help determine where time is being spent and where any performance issues lie. For example, if the server time section is high, there could be a problem in server side code.
Load time details:
- DNS: Time it takes for the nameserver lookup to resolve
- Latency: Time it takes for the host server to receive and process a request for an asset
- SSL: Time to complete an SSL handshake
- Server: Time it takes for the server to compute your code
- Transfer: Time the page has to wait for assets to load from the server
- Children: Time for asynchronous assets to process - this refers to all requests loaded by the page up until onLoad (includes scripts, stylesheets, images and XHR requests)
You can also click on a page link to get more detail about a specific page.
In the Slowest pages table, the slowest pages are shown in descending order, with the top ten slowest displayed. As with the Most requested pages table, click on a page link to get more detail about a specific page.
The Average load times distribution table displays the average load time distribution across the date range specified:
The fastest 10%, slowest 10% and average 80% are highlighted. This is a useful way to see the differences between the fastest and slowest load times, and to check whether any actions should be taken to reduce the load differences.
Click the cog icon on the upper right corner to enable optional settings:
- Logarithmic Scale: If the data has a large range of average load times with outliers, viewing the data with this turned on might make the graph easier to read.
- Hide 10% slowest: If the slowest 10% of average load times is significantly higher than the average and fastest load times, turning this option on might make it easier to view the average and fastest data, by removing the outliers.
Inspect page performance
Looking into the performance of a single page aggregates all data in the time period and provides a granular breakdown of load time, load distributions and a full waterfall of how this page loads based on a sample request from the period.
You will also see the slowest asset requests. If this page generates AJAX requests, the timing data for those calls will be displayed at the end of this page.
Exporting to CSV
All performance data on this page is available to be downloaded in CSV format.
This can be done by clicking the setting icon on the module that you wish to export data for and selecting export CSV.
note: For performance reasons we limit the number of records exported to 1,000.
You can read the following articles for more insights on solving web performance issues: