The 2015 keyword research workflow — export volume, sort by KD, pick the winners — is only half a job in 2026. LLMs pull from entities and canonical facts, not blue-link rankings. Miss that layer and you'll rank #3 in Google and be invisible in ChatGPT.
TL;DR
- Add an entity column to your keyword sheet.
- Cluster by question (who/what/why/how) — LLMs love questions.
- Track KD and 'answer difficulty' separately.
- Prioritize keywords where classic SERPs and AI Overviews overlap.
The new columns your keyword sheet needs
| Column | Why it matters | Source |
|---|---|---|
| Primary entity | LLMs retrieve by entity, not string | Wikipedia / Google KG |
| Question form | Wins PAA + AI Overviews | SERP scrape |
| Canonical fact | The one-line answer to steal | Manual |
| AI Overview presence | Predicts GEO potential | Manual SERP check |
Cluster by intent, not just topic
- Informational — the LLM playground. Optimize hard for GEO/AEO.
- Commercial investigation — where Google still owns the click.
- Transactional — protect classic SERP rankings first.
- Navigational — brand terms; make sure your schema is clean.
The 60/30/10 rule
Spend 60% of research time on entities, 30% on question phrasing, 10% on classic volume. It inverts the old ratio for a reason.
Do I still need a paid keyword tool?
Yes — but pair it with a manual SERP + AI Overviews scan for the top 20 terms per silo.
How do I find canonical facts?
Read the top-cited pages in Perplexity for the query. Their bolded lines are your fact list.
Ship this workflow inside WordPress
WBP Omni SEO Pro turns every playbook on this blog into an approvable, reversible diff.
Get WBP Omni SEO ProAffiliate — this link goes to the official WBP Omni SEO Pro product page.
T
The WBP Editorial Team
WP Bulk Publishing
