Our summary
After Dreamforce and UNBOUND, Adam Schoenfeld wanted more than a chart of social activity. He collected posts to understand topics, audience interests and possible messaging angles. His playbook uses Codex to coordinate collection, Apify's ready-made collection programs to retrieve public posts, and Jev to help classify the results.
The collection expands from seed posts to active authors, company pages and reactions. Each round is judged by new relevant posts, not the number downloaded. The author sets a budget, saves progress and changes tactics when a route weakens. Classification questions distinguish enthusiasm for a conference from enthusiasm for its product announcements.
The author reports 192,645 classification decisions across 7,135 posts costing $1.35, excluding collection. Those are creator figures, not independently reproduced results. The collection is a sample, not all of LinkedIn. Audience segmentation and paid-media activation are proposed next steps, while the article discloses sponsorship and partner links.
Key takeaways
- Measure each collection round by new relevant records, cost and remaining useful leads.
- Keep an optional known-post reference separate from discovery so it remains a useful test.
- Define exactly what a sentiment question refers to, and read the source behind important conclusions.