PageSpeed Insights analyzes a page to see if it follows our recommendations for making a page render in under a second on a mobile network. Research has shown that any delay longer than a second will cause the user to interrupt their flow of thought, creating a poor experience. Our goal is to keep the user engaged with the page and deliver the optimal experience, regardless of device or type of network.
It is not easy to meet the one second time budget. Luckily for us, the whole page doesn’t have to render within this budget, instead, we must deliver and render the above the fold (ATF) content in under one second, which allows the user to begin interacting with the page as soon as possible. Then, while the user is interpreting the first page of content, the rest of the page can be delivered progressively in the background.
Adapting to high latency mobile networks
Meeting the one second ATF criteria on mobile devices poses unique challenges which are not present on other networks. Users may be accessing your site through a variety of different 2G, 3G, and 4G networks. Network latencies are significantly higher than a wired connection, and consume a significant portion of our 1000 ms budget to render the ATF content:
- 200-300 ms roundtrip times for 3G networks
- 50-100 ms roundtrip times for 4G networks
3G is the dominant network type around the world, and while 4G deployments are in progress around the world, you should still expect that the majority of users will be accessing your page on a 3G network. For this reason, we have to assume that each network request will take, on average, 200 milliseconds.
With that in mind, let’s now work backwards. If we look at a typical sequence of communication between a browser and a server, 600 ms of that time has already been used up by network overhead: a DNS lookup to resolve the hostname (e.g. google.com) to an IP address, a network roundtrip to perform the TCP handshake, and finally a full network roundtrip to send the HTTP request. This leaves us with just 400 ms!
Delivering the sub one second rendering experience
After subtracting the network latency, we are left with just 400 milliseconds in our budget, and there is still a lot of work to do: server must render the response, client-side application code must execute, and the browser must layout and render the content. With that in mind, the following criteria should help us stay within the budget:
- (1) Server must render the response (< 200 ms)
- Server response time is the time it takes for a server to return the initial HTML, factoring out the network transport time. Because we only have so little time, this time should be kept at a minimum - ideally within 200 milliseconds, and preferably even less!
- (2) Number of redirects should be minimized
- Additional HTTP redirects can add one or two extra network roundtrips (two if an extra DNS lookup is required), incurring hundreds of milliseconds of extra latency on 3G networks. For this reason, we strongly encourage webmasters to minimize the number, and ideally eliminate redirects entirely - this is especially important for the HTML document (avoid “m dot” redirects when possible).
- (3) Number of roundtrips to first render should be minimized
Due to how TCP estimates the capacity of a connection (i.e. TCP Slow Start), a new TCP connection cannot immediately use the full available bandwidth between the client and the server. Because of this, the server can send up to 10 TCP packets on a new connection (~14KB) in first roundtrip, and then it must wait for client to acknowledge this data before it can grow its congestion window and proceed to deliver more data.
Due to this TCP behavior, it is important to optimize your content to minimize the number of roundtrips required to deliver the necessary data to perform the first render of the page. Ideally, the ATF content should fit under 14KB - this allows the browser to paint the page after just one roundtrip. Also, it is important to note that the 10 packet (IW10) limit is a recent update to the TCP standard: you should ensure that your server is upgraded to latest version to take advantage of this change. Otherwise, the limit will likely be 3-4 packets!
Before a browser can render a page to the user, it has to parse the page. If it encounters a non-async or blocking external script during parsing, it has to stop and download that resource. Each time it does that, it is adding a network round trip, which will delay the time to first render of the page.
- (5) Reserve time for browser layout and rendering (200 ms)
- Complicated scripts and inefficient code can take tens and hundreds of milliseconds to execute - use built-in developer tools in the browser to profile and optimize your code. For a great introduction, take a look at our interactive course for Chrome Developer Tools.
- How do 4G networks impact above mobile criteria?
- Lower roundtrip latencies are one of the key improvements in 4G networks. This will help enormously, by reducing the total network overhead time, which is currently over 50% of our one second budget on 3G networks. However, 3G is the dominant network type around the world, and will be for years to come - you have to optimize pages with 3G users in mind.
- How will SPDY and HTTP 2.0 help?
- SPDY and HTTP 2.0 both aim to reduce latency of page loads by making more efficient use of the underlying TCP connection (multiplexing, header compression, prioritization). Further, server push can further help improve performance by eliminating extra network latency. We encourage you to investigate adding SPDY support on your servers, and switching to HTTP 2.0 once the standard is ready.
- How do I identify critical CSS on my page?
- In Chrome Developer Tools, open the Audits panel, and run a Web Page Performance report, in the generated report, look for Remove unused CSS rules. Or use any other third party tool, or script, to identify which CSS selectors are applied on each page.
- Can these best practices be automated?
- Absolutely. There are many commercial and open-source web performance optimization (WPO) products which can help you meet some or all of the criteria above. For open-source solutions, take a look at the PageSpeed optimization tools.
- How do I tune my server to meet these criteria?
First, ensure that your server is running an up-to-date version of the operating system. In order to benefit from the increased initial TCP congestion window size (IW10), you will need Linux kernel 2.6.39+. For other operating systems, consult the documentation.
To optimize server response time, instrument your code, or use an application monitoring solution to identify your bottleneck - e.g., scripting runtime, database calls, RPC requests, rendering, etc. The goal is to render the HTML response within 200 milliseconds.
- What about Content Security Policy?
- If you are using CSP, then you may need to update your default policy.
First, inline CSS attributes on HTML elements(e.g.,
Have more questions? Please feel free to ask and provide feedback in our discussion group at pagespeed-insights-discuss.