Building a A Zero-Code AI SEO Workflow

You can’t build an AI SEO expert. You can build an AI SEO audit workflow that analyzes your site’s existing content, traffic data, and crawl data for opportunities. I’ll show you how I do it, from the inputs — ScreamingFrog, Google Analytics, DataForSEO — to the prompts and project.

Transcript

1. A Zero-Code SEO Assistant That doesn’t suck Ian Lurie
2. https://ianlurie.me/ai-seo-workflow Slides, markdown, backup materials
3. Ian Lurie Digital marketer for 31 years Dungeons & Dragons player for 45 years Nerd since birth
4. I have a reputation
5. Nothing but love
6. AI is going to change everything! Expertise is just a prompt away!
8. Not expertise. Collaboration.
9. Not prompts. Workflows.
10. A great AI workflow is a collaborator that uses real data and is trainable
11. OK, on to SEO
14. I need an SEO AI workflow that is a collaborator, is trainable, and uses real data.
15. Ta-da
16. The tools
17. Claude.ai is easy to work with, connects well to MCP servers, and is my favorite.
18. Screaming Frog is a desktop site crawler that integrates with a lot of the other tools we’re gonna use
19. OpenAI provides an embeddings toolset that’s easy to use in Screaming Frog
20. Data For SEO provides rankings data (and works with MCP)
21. Zapier MCP provides a (relatively) trouble-free Model Context Protocol server
22. Also Google Sheets, Google Analytics, Google Search Console, and your favorite text editor.
23. Don’t have all these? That’s OK! Use what you’ve got.
24. Rather use Gemini or ChatGPT? Go for it!
25. Step 1: Set up a Claude Project
26. Add a new project
27. Add instructions & “opportunity” guidelines
28. Instructions
29. Guidelines
30. Step 2: Get an OpenAI API key
31. platform.openai.com/docs/quickstart
32. Step 3: Set up Zapier MCP
33. A great AI workflow is a collaborator that uses real data and is trainable
34. Model Context Protocol
35. MCP servers give AI tools like Claude easy access to all kinds of data & tools.
36. MCP servers give AI tools like Claude easy access to all kinds of data and tools.
37. Why I use Zapier
38. Zapier.com
39. mcp.zapier.com
40. Add your tools
41. Copy and paste into Claude
42. You should see this
43. And something like this
44. Create a folder Zapier can access
45. Step 4: Run Screaming Frog
46. Screaming Frog will aggregate page- level analytics data for you.
47. In Screaming Frog, connect: Google Analytics Search Console OpenAI (using your API key)
48. In Screaming Frog, set up: Embeddings
49. Run your crawl
50. Export your crawl to the Google Drive folder you shared with Zapier (use the columns on the right)
51. The result
52. I have multiple clients, so I export to a clearly-named subfolder and point Claude accordingly
53. Step 5: Export one Google Analytics report
54. Google Analytics 4 exploration Active users Engaged Sessions Views Last 90 days vs. previous 90
55. The result
56. Quick check-in
57. You’ve got
58. You’ve got
59. You’ve got
60. All this great food for the AI
61. Food you collected using ScreamingFrog & integrations
62. That’s now accessible to Claude using Zapier MCP
63. Step 6: Put Claude to work
65. Data For SEO is in my instructions. Claude will grab that data as needed, using Zapier MCP.
66. Step 7: Review
67. Keep editing!
70. This workflow is not an SEO expert. AI is not an SEO expert. You are the expert.
71. A great AI workflow is a collaborator that uses real data and is trainable
72. Step 8: Refine
73. The AI will get it wrong.
74. A great AI workflow is a collaborator that uses real data and is trainable
75. Keep editing!
76. Keep editing!
77. Add more details
78. Now what?
80. Agents. Someday.
82. You’ve got an SEO AI workflow that is a collaborator, uses real data, and is trainable
83. This presentation was lo-carb Questions? ianlurie.com https://ianlurie.me/ai-seo-workflow