Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

NLP Powered Outreach Link Building

283 views

Published on

Gareth Simpson's slides from TechSEO Boost 2019

Published in: Marketing
  • Be the first to comment

NLP Powered Outreach Link Building

  1. 1. Gareth Simpson | @seekerdigital | #TechSEOBoost Gareth Simpson @simpsongareth NLP Powered Outreach Link Building
  2. 2. Gareth Simpson | @seekerdigital | #TechSEOBoost Specialist Outreach and Link Agency
  3. 3. Gareth Simpson | @seekerdigital | #TechSEOBoost
  4. 4. Gareth Simpson | @seekerdigital | #TechSEOBoost What we’ll be covering 1. Link building process 2. NLP productivity hacks
  5. 5. Gareth Simpson | @seekerdigital | #TechSEOBoost The Four Pillars of SEO • Content: Relevance & keywords • User Experience: Engagement & reputation • Links: Authority & trust • Technical: Website build quality
  6. 6. Gareth Simpson | @seekerdigital | #TechSEOBoost The Importance of Backlinks https://ahrefs.com/blog/competitive-analysis/ “Because when we analysed over 2 million keywords, we also found that it was possible to outrank sites with higher DR scores by building backlinks to individual pages. So if your site is new and is in a competitive niche with high DR scores, then your initial plan should be to build links directly to pages you want to rank. I should also point out that Google deny overall domain authority influences rankings, however, all our testing finds a strong positive correlation between DR and rankings. That being said, as we also found links to individual pages trump overall DR…”
  7. 7. Gareth Simpson | @seekerdigital | #TechSEOBoost Link Building Process
  8. 8. Gareth Simpson | @seekerdigital | #TechSEOBoost
  9. 9. Gareth Simpson | @seekerdigital | #TechSEOBoost
  10. 10. Gareth Simpson | @seekerdigital | #TechSEOBoost
  11. 11. Gareth Simpson | @seekerdigital | #TechSEOBoost
  12. 12. Gareth Simpson | @seekerdigital | #TechSEOBoost
  13. 13. Gareth Simpson | @seekerdigital | #TechSEOBoost The Outreach Problem • Link building is hard to scale • Requires effective PM • Mundane work • Lots of data to process
  14. 14. Gareth Simpson | @seekerdigital | #TechSEOBoost 12.5% of staff time is lost in data collection. That’s five hours a week in a 40-hour work week.SOURCE: GARTNER
  15. 15. Gareth Simpson | @seekerdigital | #TechSEOBoost Data Piping • Platforms • Ahrefs & Majestic • Pitchbox • Data studio / Sheets • Connections • Zapier • Piesync • API
  16. 16. Gareth Simpson | @seekerdigital | #TechSEOBoost Decision Bottlenecks • Human intervention • Read the situation & act • Blockers in the workflow
  17. 17. Gareth Simpson | @seekerdigital | #TechSEOBoost https://www.slideshare.net/ipullrank/building-your-outreach-machine Published on June 9th, 2017
  18. 18. Gareth Simpson | @seekerdigital | #TechSEOBoost ML Benefits for Outreach • Automating decisions • Processing mass amounts of data • Enables you to work faster, smarter and with more accuracy • Frees up smart people from doing mundane work • Higher level applications offering more developed models
  19. 19. Gareth Simpson | @seekerdigital | #TechSEOBoost 1
  20. 20. Gareth Simpson | @seekerdigital | #TechSEOBoost Natural Language Processing • Content classification • Natural language understanding • Provides awareness to: • Automate repetitive tasks • Emulate human decisions 2 0
  21. 21. Gareth Simpson | @seekerdigital | #TechSEOBoost NLP without a PHD
  22. 22. Gareth Simpson | @seekerdigital | #TechSEOBoost App stack 2 DATA ML ENGINES
  23. 23. Gareth Simpson | @seekerdigital | #TechSEOBoost NLP Powered Outreach ENABLING BETTER COMMUNICATIONS
  24. 24. Gareth Simpson | @seekerdigital | #TechSEOBoost Contact Research & Enrichment 1. Manual contact research 2. Identify the appropriate person 3. Slow & error prone
  25. 25. Gareth Simpson | @seekerdigital | #TechSEOBoost
  26. 26. Gareth Simpson | @seekerdigital | #TechSEOBoost
  27. 27. Gareth Simpson | @seekerdigital | #TechSEOBoost The Process 1. Pull job title from data source e.g. Pitchbox or Vuelio 2. Run through MonkeyLearn’s job industry classifier 3. Filter table on contacts relevant for pitch 4. Run through MonkeyLearn’s Seniority Classifier to ID 5. Enrich contacts for additional info e.g. social profiles 6. Pipe populated data into Pitchbox for sends.
  28. 28. Gareth Simpson | @seekerdigital | #TechSEOBoost Email Filtering 1. Replying to 1000s of emails every month 2. Yes/No/Maybe/Questions/Auto 3. Not best use of our expert negotiators time
  29. 29. Gareth Simpson | @seekerdigital | #TechSEOBoost Cut through the noise • We’re sitting of 4 years of email decisions which became our training data • Summarise email • Classify for intent, sentiment, urgency etc • Send to Pitchbox as custom tag for filtering 2 9
  30. 30. Gareth Simpson | @seekerdigital | #TechSEOBoost Training data 3 1 2 3 4 Human processed data Features & tags Good data in > good data out Refine over time
  31. 31. Gareth Simpson | @seekerdigital | #TechSEOBoost 3
  32. 32. Gareth Simpson | @seekerdigital | #TechSEOBoost 3 By 2020, 85% of customer interactions will be handled without a human SOURCE: GARTNER
  33. 33. Gareth Simpson | @seekerdigital | #TechSEOBoost
  34. 34. Gareth Simpson | @seekerdigital | #TechSEOBoost
  35. 35. Gareth Simpson | @seekerdigital | #TechSEOBoost Reactive PR • HARO • Twitter • Response Source
  36. 36. Gareth Simpson | @seekerdigital | #TechSEOBoost HARO 1. Journalist submits request 2. PRs check email digests 3. Coordinate content and respond 4. Journalist uses quote and credits
  37. 37. Gareth Simpson | @seekerdigital | #TechSEOBoost
  38. 38. Gareth Simpson | @seekerdigital | #TechSEOBoost Twitter • #journorequest • #prrequest • #bloggerrequest
  39. 39. Gareth Simpson | @seekerdigital | #TechSEOBoost Categories Categorize your content using a five-level classification hierarchy. Concepts Identify high-level concepts that aren't necessarily directly referenced in the text. Emotions Analyze emotion conveyed by specific target phrases or by the document as a whole. Entities Find people, places, events, and other types of entities mentioned in your content. Keywords Search your content for relevant keywords. Metadata For HTML and URL input, get the author of the webpage, the page title, and the publication date. Relations Recognize when two entities are related, and identify the type of relation. Semantic Role Parse sentences into subject-action-object form, and identify entities and keywords that are subjects or objects of an action. Sentiment Analyze the sentiment toward specific target phrases and the sentiment of the document as a whole. Custom Models Identify custom entities and relations unique to your domain with Watson Knowledge Studio.
  40. 40. Gareth Simpson | @seekerdigital | #TechSEOBoost HARO Email Extraction 4
  41. 41. Gareth Simpson | @seekerdigital | #TechSEOBoost #journorequest 4 FILTERCLASSIFY NOTIFICATIONSHASHTAGS
  42. 42. Gareth Simpson | @seekerdigital | #TechSEOBoost Reactive PR 1. Digital PRs log in once per day 2. Filter by keyword / category 3. Export contact & info request to Pitchbox.
  43. 43. Gareth Simpson | @seekerdigital | #TechSEOBoost Summary • Map out a process for link building • Leverage APIs to speed up data transfer • Use NLP to help with automating decisions • Start with MonkeyLearn & Zapier, then move to native API Get them links!
  44. 44. Gareth Simpson | @seekerdigital | #TechSEOBoost
  45. 45. Gareth Simpson | @seekerdigital | #TechSEOBoost Thank You – gareth@seeker.digital

×