Se ha denunciado esta presentación.
Utilizamos tu perfil de LinkedIn y tus datos de actividad para personalizar los anuncios y mostrarte publicidad más relevante. Puedes cambiar tus preferencias de publicidad en cualquier momento.

Digital Olympus Technical SEO Findings Whilst Taming An SEO Beast

870 visualizaciones

Publicado el

A few of the recent findings we discovered whilst working on an SEO beast which cover crawling, server log file analysis, site speed optimization and database optimizations. Technical SEO insights

Publicado en: Marketing
  • Sé el primero en comentar

Digital Olympus Technical SEO Findings Whilst Taming An SEO Beast

  1. 1. Default  Ecommerce  Site  Crawl  Settings
  2. 2. Default  High  Quality  SEO  Blog  Crawl   Settings
  3. 3. Oliver’s   Experiment   And   Increased   Clicks
  4. 4. SITE  DEFAULT  SETTINGS  IN  CRAWL  GSC SITE  TYPE SITE DEFAULT   SECONDS  BETWEEN   REQUESTS SITE DEFAULT   REQUESTS  PER   SECOND CMS  TYPE CDN USED? SHARED HOSTING  /   DEDICATED  SERVER Business  Directory  – Data driven 10 0.1 NONE  -­‐ CUSTOM  BUILT   APPLICATION NO DEDICATED  SERVER Gambling  – data   driven 10 0.1 NONE – CUSTOM   BUILT  APPLICATION NO DEDICATED  SERVER Travel  /  Auto 5 0.2 CUSTOM BUILD  /   WORDPRESS  HYBRID CLOUDFLARE DEDICATED  SERVER Distribution Service   (Database  Driven  Site) 0.003 333.333 NONE – CUSTOM   BUILT  APPLICATION NO SHARED  HOSTING Ecommerce  fashion 0.001 100 SHOPIFY SHOPIFY SHOPIFY  PLATFORM Business  Blog 0.008 125 WORDPRESS NO SHARED  HOSTING Ecommerce  Fashion 0.003 333.333 SHOPIFY SHOPIFY SHOPIFY  PLATFORM
  5. 5. SITE  DEFAULT  SETTINGS  IN  CRAWL  GSC SITE  TYPE SITE DEFAULT   SECONDS   BETWEEN   REQUESTS SITE DEFAULT   REQUESTS  PER   SECOND CMS  TYPE CDN USED? SHARED HOSTING  /   DEDICATED  SERVER Ecommerce  – Home &   Garden 0.016 MAGENTO NO DEDICATED  SERVER Jobs board  – Dynamic   data 333.333 0.003 UMBRACO NO DEDICATED  SERVER Ecommerce -­‐ Gifts 125 0.008 WORDPRESS CLOUDFLARE SHARED HOSTING Business Consultancy   – Brochure  Site 333.333 0.003 WORDPRESS NO DEDICATED SERVER Jobs  Board 100 0.01 UMBRACO CLOUDFLARE DEDICATED  SERVER
  6. 6. Popular  CMS  ’Rule  Patterns’  (URL  Parameters) ALL  WILL  HAVE  SOME  COMMON   PATTERNS  WHICH  CAN  BE  LEARNED   FOR  EFFICIENCY BUILDS   ‘HINTS’  ON   WHAT  NOT   TO  CRAWL
  7. 7. Custom  Build   PHP  Site   “DEFAULT” SETTING”
  8. 8. I  slid  the  slider  up  to  ‘high’  crawl  request  too
  9. 9. What  happened?
  10. 10. 1  week  later  -­‐ -­‐ No  difference  at  all  to  crawl
  11. 11. NADA ZILCH…  ZERO
  12. 12. Putting  in  a   crawl  request   via  Google   Search   Console
  13. 13. What  happened?
  14. 14. NADA ZILCH…  ZERO
  15. 15. 1  week  later  -­‐ -­‐ No  difference  at  all  to  crawl
  16. 16. Can  you   handle  it?...   NO,  YOU   CAN’T
  17. 17. No?     …Do  Nothing
  18. 18. Optimizing  For  Speed
  19. 19. NO  CDN
  20. 20. CLOUDFLARE  WILL   NOT  INCREASE  YOUR   GOOGLE  PAGE  SPEED   SCORES
  21. 21. Finding  4  – PageSpeed   Module   Conflicts
  22. 22. Conflicts  with  AJAX   &  JSON • CONFLICTS  WITH  JSON  AND   AJAX  FORM  FILLS  WHICH  PULL   ELEMENTS  AND  INDIVIDUAL  DATA   RATHER  THAN  A  FULL  PAGE  LOAD
  23. 23. Savings  from  Gzip Compression Varvy’s Gzip Compression  Testing  Tool  -­‐>  https://varvy.com/tools/gzip/
  24. 24. USING  GZIP   COMPRESSION  HAS  A   COST
  25. 25. IT  CAN  INCREASE  TTFB Source:  https://blog.cloudflare.com/ttfb-­‐time-­‐to-­‐first-­‐byte-­‐considered-­‐meaningles/
  26. 26. TTFB  is  Often  Increased  With  Gzip Compression Compression  has  a  cost  because  it  increases  TTFB   but  the  web  page  loads  quicker.   So,  it’s  better  for  users  but  TTFB  is  negatively   impacted  whist  the  page  is  unzipped Gzip Compression Source:  https://blog.cloudflare.com/ttfb-­‐time-­‐to-­‐ first-­‐byte-­‐considered-­‐meaningles/ Gzip Compression   effects  TTFB  negatively. But  the  page  loads   faster  in  the  end
  27. 27. IS  TTFB  REALLY  ALL  THAT      ?? Source:  https://blog.cloudflare.com/ttfb-­‐time-­‐to-­‐first-­‐byte-­‐considered-­‐meaningles/ H
  28. 28. You  should  reduce  your  server  response  time   under  200ms (Google  Developers)
  29. 29. OPTIMIZE  FOR  USERS  FIRST CRITICAL INTERACTION PATH
  30. 30. CSS  TOO  LARGE  ==  8k+  Lines
  31. 31. Remove  7.5k  lines  of   unnecessary  css code? What  do  you  think   happened?
  32. 32. NOTHING   BAD
  33. 33. Getting  There
  34. 34. MERGE  &  MINIFY  JS  &  CSS BUT  YOU  MUST  KEEP  THEM   IN  THE  SAME  ORDER  AS  THEY   WERE  IN  ORIGINALLY  IN  THE   ONE  COMBINED  CODE
  35. 35. THIRD  PARTY   ANALYTICS   REDIRECT  CHAINS
  36. 36. <link  rel="dns-­‐prefetch"> <!-­‐-­‐ Prefetch DNS  for  external  assets  -­‐-­‐>   <link  rel="dns-­‐prefetch"  href="//fonts.googleapis.com">   <link  rel="dns-­‐prefetch"  href="//www.google-­‐ analytics.com">   <link  rel="dns-­‐prefetch"  href="//opensource.keycdn.com">   <link  rel="dns-­‐prefetch"  href="//cdn.domain.com">
  37. 37. APPLICATION  LOGIC   IS  A  NIGHTMARE
  38. 38. UNOPTIMIZED   DATABASE  CALLS
  39. 39. RE-­‐OPTIMIZE  YOUR  DATABASE APPLICATION  LOGIC  IS  A  KILLER &  MULTIPLE  DB  CALLS
  40. 40. Optimize  &  Minimize  SQL  Queries Application  Logic   Problems (Google  Developers) %LIKE%  &   ^PREGEX$  in   SQL
  41. 41. OPTIMIZE  YOUR   WORDPRESS  DATABASE MANY  PLUGINS   LEAVE  ‘THINGS’   BEHIND  WHEN   DELETED
  42. 42. Add  ‘Keep-­‐Alive’  Connection
  43. 43. Messy  Crawl  with   Builder  Cache   Included  Files (DEPENDENT  FILES   ARE  GIVEN   PRIORITY)
  44. 44. Cache  Control  in  .htaccess #  One  year  for  image  files <filesMatch ".(jpg|jpeg|png|gif|ico)$"> Header  set  Cache-­‐Control  "max-­‐age=31536000,  public" </filesMatch> #  One  month  for  css and  js <filesMatch ".(css|js)$"> Header  set  Cache-­‐Control  "max-­‐age=2628000,  public" </filesMatch>
  45. 45. Content   fingerprinting   &  CSS  /  JS   Files https://www.yourdomain.com/css/exa mple-­‐styles.css?cache=1711101236 https://www.yourdomain.com/js/exa mple-­‐js.js?cache=19683983
  46. 46. 7  JS files reduced To  ONE
  47. 47. 1.75  seconds  to  fully  load  now
  48. 48. DEFER   LOADING  OF   JAVASCRIPT Varvy.com
  49. 49. 1.77  seconds  now  to  fully  load Defer   javascript till  after   full  load
  50. 50. Yslow score  is   improved  after   defer  javascript
  51. 51. Base  64  Encoding BINARY  TO  TEXT   ASCII  ENCODING
  52. 52. BASE  64  ENCODER ONLY  SUITABLE  FOR   SMALL  IMAGES  SUCH   AS  LOGO
  53. 53. 1.69  seconds  with  no  CDN  &  still  improvements   available
  54. 54. WILL  WE  ADD  A   CDN?
  55. 55. THE  SOUNDEX  &  METAPHONE   ALGORITHMS https://en.wikipedia. org/wiki/Soundex SPELLING   MISTAKES
  56. 56. SQL,  SOUNDEX  &  SEMANTIC   POSSIBILITIES  
  57. 57. SOUNDEX  EXAMPLE  GENERATOR
  58. 58. Thank  you
  59. 59. Sources  &  References • http://ohgm.co.uk/crawl-­‐rate-­‐defaults/ • https://developers.google.com/speed/docs/insights/Server • https://varvy.com/pagespeed/defer-­‐loading-­‐javascript.html • https://www.red-­‐gate.com/simple-­‐talk/sql/t-­‐sql-­‐programming/fuzzy-­‐ searches-­‐sql-­‐server/ • Bar-­‐Yossef,  Z.,  Keidar,  I.  and  Schonfeld,  U.,  2009.  Do  not  crawl  in  the   dust:  different  urls with  similar  text. ACM  Transactions  on  the  Web   (TWEB), 3(1),  p.3. • https://www.seroundtable.com/google-­‐grammar-­‐seo-­‐24571.html
  60. 60. Sources  &  References • https://www.seroundtable.com/google-­‐measure-­‐page-­‐speed-­‐ 24821.html • https://www.w3resource.com/mysql/string-­‐functions/mysql-­‐ soundex-­‐function.php • https://www.base64encode.org/ • https://en.wikipedia.org/wiki/Metaphone

×