Methods of Testing Internal Linking with Chris green
1 like448 views
Chris Green presented at the Crawl & Indexation Summit on September 11, 2020, discussing the importance and challenges of testing internal linking strategies in SEO. He emphasized the necessity of questioning established practices and advocating for a testing-based approach to assess the impact of changes made to internal linking structures. Green also shared methodologies for experimentation using Oncrawl, with specific processes and potential outcomes based on the results of various modifications to a site’s linking strategy.
Methods of Testing Internal Linking with Chris green
- 1. Crawl & Indexation Summit Sept 11th, 2020 Methods of Testing Internal Linking Methods of Testing Internal Linking WITH CHRIS GREEN
- 2. Crawl & Indexation Summit Sept 11th, 2020 Methods of Testing Internal Linking
- 3. Chris Green OnCrawl Ambassador @ChrisGreen87 +10 years experience in SEO Head of Marketing Innovation at Footprint digital Speaking today
- 4. Judging the impact of internal linking changes is notoriously difficult, and...
- 5. We make too many changes without testing!
- 7. Changes we make with good intentions can often do more harm than good
- 8. Experiments with Internal Linking | @chrisgreen87 Testing this isn’t easy
- 9. Experiments with Internal Linking | @chrisgreen87 I want you to leave the session more willing to challenge and test your own knowledge when optimising in the future
- 10. We fear asking “dumb-sounding” questions
- 11. “I knew that 5 years ago!”
- 12. “Everyone should know that” “I knew that 5 years ago!”
- 13. “It’s obvious” “I knew that 5 years ago!” “Everyone should know that”
- 14. “It’s obvious” “I knew that 5 years ago!” “Everyone should know that” “BAD TEST!”
- 15. Experiments with Internal Linking | @chrisgreen87 Don’t be afraid of this. Even if it’s unhelpful & damaging
- 16. EVEN with a more testing-based mindset, we’re all too stuck in our ways
- 17. Quit looking for things to confirm your decisions, start looking for ways to debunk them.
- 18. I want to share with you my methodologies, findings and hopefully inspire you to try something similar.
- 19. The Tests Using OnCrawl I recently ran some internal linking tests. I use their Inrank score as a proxy for how Google measures the flow of value between pages. To try to understand what these changes do before deployment
- 20. I wanted to explore how valid my ‘gut feeling’ would be under test conditions
- 21. Meet the Test Site
- 22. The Site • 3,762 posts • 146 authors • 4 post categories • 0 tags
- 23. BUT a crucial point to remember here is I added no robots/crawling control in any form. Whilst it would have been interesting to do so, the scope of this was the impact of internal linking only.
- 24. What would you do to optimise for this site?
- 25. Experiments with Internal Linking | @chrisgreen87 “It depends”
- 26. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags
- 27. The Tests
- 28. The Tests
- 29. Experiments with Internal Linking | @chrisgreen87 The Process 1. Crawl the site
- 30. Experiments with Internal Linking | @chrisgreen87 The Process 1. Crawl the site 2. Make a change
- 31. Experiments with Internal Linking | @chrisgreen87 The Process 1. Crawl the site 2. Make a change 3. Crawl the site
- 32. Experiments with Internal Linking | @chrisgreen87 The Process 1. Crawl the site 2. Make a change 3. Crawl the site 4. Reset, make another change
- 33. Experiments with Internal Linking | @chrisgreen87 1. Crawl the site 2. Make a change 3. Crawl the site 4. Reset, make another change 5. Repeat until done You need to be patient for this! The Process
- 34. Some tips 1. Write down everything 2. Crawl setup 3. Categories/tags 4. Plugin settings etc 5. Take Database Backups often 6. Take Codebase Backups often 7. Ensure Google cannot crawl your site!
- 35. Results
- 37. ��
- 40. 90%> of site content is still 5> levels deep.
- 41. We are looking at the trade off between how quickly a search engine (or user) can access a page vs. how much value is passed to that page.
- 43. You can read more about it here http://bit.ly/internal-links-ebook
- 44. Back to my hypothesis
- 45. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags ✔
- 46. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags ✔ ❌
- 47. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags ❌ ✔ ❌
- 48. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags ❌ ✔ ✔ ❌
- 49. Experiments with Internal Linking | @chrisgreen87 EVEN if my implementations of these new elements needed better planning, my initial thoughts/steps were not as good as I would have hoped!
- 50. In order to test my approach I created a proxy, but what if you can’t?
- 51. Experiments with Internal Linking | @chrisgreen87 Three Methods 1. The ‘Paper’ Test 2. ‘Live’ testing 3. Scrape & ‘Hack’
- 52. The ‘Paper’ Test 1. Create a version of your site using ‘pagerank games’ thanks to JohnMu 2. Then adjust to show changes 3. Compare changes in output log_PR values (using excel or similar) http://bit.ly/pagerank-games http://bit.ly/networkX
- 53. It has to be pretty close* to your site… *the closer, the better
- 54. Python is a skill that not all SEOs have - including myself (yet?)
- 55. The ‘Paper’ Test 1. Crawl your site 2. Download all the links data 3. Import into Knime 4. Calculate PageRank 5. Change Links Data (using excel or similar) 6. Re-run through Knime https://www.knime.com/
- 56. There’s got to be an easier way here!
- 57. ‘Live’ testing 1. Using JavaScript (via a tag manager) 2. Edge workers (EdgeSEO) 3. Trigger via custom user agent (NOT GOOGLE) 4. Crawl (with JS Enabled) 5. Compare with original
- 58. If you don’t have EDGE SEO POWER!
- 59. Experiments with Internal Linking | @chrisgreen87 Google Tag Manager has your back!
- 61. ‘Live’ testing Whatever you want to change in here
- 62. ‘Live’ testing Custom UA <- don’t mess this bit up! Firing trigger - could be anything, usually will be page rules though
- 63. Experiments with Internal Linking | @chrisgreen87 It does not need to look pretty - just have hrefs on the page which your crawler will pick up
- 64. Experiments with Internal Linking | @chrisgreen87 I guess this is cloaking - don’t take excess risks!
- 65. If the logic needed is too complicated?
- 66. ‘Scrape & Hack’ Method http://bit.ly/ethical-content-scraping 1. As per my initial experiment 2. Developer skills likely needed 3. You need a server & CMS which can function like the live site 4. Internal link structure needs to be the same
- 68. Experiments with Internal Linking | @chrisgreen87 How do we know Inrank (or PageRank) or any other metrics - for that matter - is a good indication of a successful test?
- 69. Experiments with Internal Linking | @chrisgreen87 We don’t… not really. It’s better than not testing (remember the original example) AND this needs to be coordinated with monitoring & tweaking.
- 70. ▢ Added related posts ▢ Added more categories ▢ Added date taxonomies ▢ Ignored tags ❌ ✔ ✔ ❌
- 71. Go out, test it. Create a basic proxy for the site you are working on, check for relative change to what is there already.
- 72. Experiments with Internal Linking | @chrisgreen87
- 73. www.oncrawl.com Enjoy a 1-month free trial