Brighton SEO Notes – Part 2

23 Apr 2019

SEO Team Director

Brighton SEO Notes – Part 2

Site Speed

 

Speed & Performance Optimisation: How to Meet Users High Expectations – Rachel Costello

Slides here

    • Talk based on her blog post
    • Notes:
      • Consider the worst/slowest devices your users access the site via – get insight via GA devices to know % on terrible phones
      • People are more likely to blame their internet connection than the website for slow load times
      • Prioritise above-the-fold content for best ‘perception’ of speed – even if the rest loads super slow
      • org do a cool speed map
      • Look at the number of unused scripts per page
      • GT metrix is good for testing whether a CDN or different server location could help
      • Implement “progressive enhancement”:
        • Load HTML
          • Load CSS
            • Load JS
          • Or implement “graceful degradation” (BBC)
            • Load everything
              • Pull JS back
                • Pull CSS back
                  • Just have HTML loaded
                • Use cool things:
                  • Preload pages
                  • Prefetch links
                  • Preconnect server
                  • JS will use AI and GA data to determine most likely next click
                  • Use ‘Network information API’ to sent the best weight page to the user
                    • Will also check if connection changes during the session
                    • Slow connection; get a nice image. Fast connection; get a great video

 

How to get a 100% Lighthouse performance score – Polly Pospelova

(my favourite talk of the day)

Slides here

  • Notes
    • Lighthouse can be 100%!
    • HTTP v HTTP2 is the equivalent of a relay race v 100m race – which ends first?
    • Use next gen formats: WebP images
    • Use adaptive image sizes
      • code will contain loads of image link variations but only one will load
      • create a placeholder that is filled dynamically
      • makes for really bloated HTML – but still quicker
    • Use lazy load
    • Critical metrics:
      • All JS and CSS converted to HTML and put in the head
      • DOM gets loads bigger but that’s okay
      • Everything then gets cached
      • Cookies used to decide whether this is sent inline or not
    • Load only what you need
      • Break up all css & js into tiny individual files (only works with HTTP2!)
    • ModernJS is 300% faster
    • Speed in an ongoing task, like all optimisation

 

How to Trim JS, CSS and External Stuff to Slim Down and Speed up your Site – Chris Simmance 

Slides here

  • Notes:
    • Progressive Jpeg has a fade in effect that means an image loads instantly and then gets better quality
      • By the time their eye is drawn to the image it should be fully loaded
    • Stop using trackers – they are the worst
      • They also still work in the footer
      • Use GTM
      • Never use plugins always hardcode
    • Fake minimalism is the worst (pages with nothing on them but are thousands of lines of code in the back-end)
      • See coke example
    • What PHP version are you using? 7.2/7.3 is the best – easy to update
    • Apache Gzip code is really easy to implement
    • When caching – use “varnish cache” not folder cache

 

 

Making SEO life super easy

 

Living on the Edge: Elevating your SEO toolkit to the CDN – Nils de Moor

Slides here

  • Notes
    • Use CDNs when your client has one and probably doesn’t have a CMS
    • Can edit:
      • txt
      • Redirects
      • Security header
      • Retrieve access logs
      • Deliver AB testing
    • Edge SEO < sounds cooler than it is
      • Lamda@edge
      • Cloudflare

 

How To Use Chrome Puppeteer to Fake Googlebot & Monitor Your Site – Tom Pool 

Slides here

  • Notes
    • Chrome Puppeteer AKA Headless Chrome
    • Use it to scrape
    • Run from command line
    • Need to install NPM & Node.JS
    • Can screenshot a webpage or bulk list of webpages as if you were Googlebot – maybe replace ‘fetch as google’
      • Install puppeteer on your server, display as googlebot view
    • Content King tells you when clients change:
      • Title
      • Meta
      • Canonical
      • Content
      • Etc
        • Puppeteer can do it for free
      • Differences get saved as a ‘differences’ txt file
      • Need a Raspberry Pi (CronJob)
      • Can get it to email you when there are difference with the txt file

 

CLI Automation – Using the Command Line to Automate Repetitive Tasks – Mike Osolinski

Slides here

  • Notes
    • Basically a load of command line shortcuts for things that take ages:
      • Bulk CSV merging < we do this
      • Bulk file renaming
      • Bulk lighthouse audit < we do this
      • Bulk image optimisation
      • SF crawls and extractions
      • Software Powershell helps

 

 

Other things

  • If you don’t have a .ru site you’ll never rank in Russia (Yandex)
  • Every. Single. Client. Needs to be tracking Brand & Generic split of traffic – minimum

 

 

SEO Team Director

As Team Director of the SEO department at ROAST, Paige plays a crucial role in the development and expansion of the ever-growing SEO team.