Keyword research did not die — it grew a second job. The classic craft still guides what to publish for Google rankings; the new craft maps the question spaces that AI retrieval and conversational queries run on. Keyword research in 2026 merges both into one workflow: seed terms to question maps to intent grids to prioritized briefs. This guide walks the full process, tools to judgment, within our SEO Fundamentals cluster.

What Changed (and What Did Not)
Volume-and-difficulty keyword picking still works for classic rankings; what changed is the unit of planning. AI surfaces decompose conversational questions into sub-queries, so coverage of a question space now outperforms targeting of individual strings.
Users type “best crm small business” into Google and ask an assistant “what CRM should a five-person consultancy use if we live in email?” — same intent, different retrieval. The classic string wins the first; question-space coverage wins the second, per the fan-out mechanics in Retrieval Optimization. One research workflow must now feed both.
The 2026 Workflow, Step by Step
- 1. Seed from business reality. Products, problems solved, audience vocabulary — plus sales calls and support tickets, the richest conversational-query source most teams ignore.
- 2. Expand classically. Keyword tools for volume, difficulty, and SERP features on head and mid-tail terms — this still allocates the ranking battles worth fighting.
- 3. Mine the question space. People Also Ask, forum threads, autocomplete variants, and LLM-assisted enumeration (the beginner/practitioner/skeptic prompt from Prompt Engineering for SEO). Group by intent, not phrasing.
- 4. Grid the coverage. Map questions onto the intent grid — definition, mechanism, comparison, execution, judgment — per topic (LLM Content Strategy). Blanks are your commissioning list.
- 5. Prioritize by value stack. Commercial relevance × attainable difficulty × question-space leverage (one piece answering many sub-queries beats one string).
- 6. Brief with both targets. Each brief names the classic term to rank for and the questions each H2 must answer directly — feeding the on-page checklist downstream.
Intent Is Still the Sorting Function
Every keyword and question sorts into the classic four — informational, commercial-investigation, transactional, navigational — and the split now also predicts surface: informational intent increasingly resolves in AI answers (optimize for citations, per AI Search vs Google Search), while transactional and local intent still clicks through (optimize classically). Misreading intent wastes content on the wrong surface — the most expensive research error, ahead of any volume misestimate. Google’s starter guidance quietly says the same: match what searchers actually want.
Tools, Data, and Judgment
Tool metrics are directional, not gospel: volumes are modeled, difficulty scores disagree, and conversational queries barely register in them. Calibrate with your own data — Search Console queries (including the long messy ones), site search logs, and monthly citation spot-checks on priority questions (AI Search Analytics). The durable skill is judgment: recognizing which questions your business can answer better than anyone, then building the cluster that proves it — the topical authority play. Strategy integration: our content strategy guide.
- The unit of planning moved from keyword strings to question spaces — one workflow now feeds rankings and retrieval.
- Six steps: seed from business reality, expand classically, mine questions, grid coverage, prioritize by value stack, dual-target briefs.
- Intent predicts surface: informational → AI citations, transactional/local → classic clicks.
- Tool metrics are directional; Search Console queries and citation spot-checks are your calibration data.
- Refresh question spaces quarterly — they drift faster than volumes, and first coverage is cheap authority.
Frequently Asked Questions
Are keyword volumes still worth checking in 2026?
Yes, for allocation: volumes still rank the classic battles by potential and expose seasonal patterns. Treat them as relative signals rather than precise counts, and remember conversational demand — often the growth segment — barely appears in them. Volume plus question-space coverage is the complete picture.
How do I research keywords for AI search specifically?
Mine questions rather than strings: ask assistants your seed topics and note the sub-queries their answers address, harvest People Also Ask trees, and enumerate the beginner-to-skeptic question list per topic. Then verify coverage monthly with citation spot-checks on the questions that matter commercially.
Does keyword difficulty apply to AI citations?
Only loosely. Citation contests are per-question and passage-level, so a difficult head term can hide easy citation wins on its sub-questions. Domain authority still helps selection, but focused clusters regularly out-cite stronger domains — difficulty scores were never built to predict that.
How many keywords should one article target?
One primary intent, expressed as a head term plus its question cluster. A well-structured article naturally ranks for dozens of related phrasings — forcing multiple distinct intents into one page fragments both rankings and chunks. When intents diverge, split into cluster siblings.
What is the best free keyword research setup?
Search Console (your actual queries), Google autocomplete and People Also Ask (demand phrasing), assistant question-mining (conversational space), and a spreadsheet grid. Paid tools add scale and competitor visibility, but the free stack plus judgment covers a focused site’s needs surprisingly well.
Conclusion
Research is still the highest-leverage hour in SEO — it just answers two questions now: what can we rank for, and which question space can we own? Run the workflow, grid the coverage, and let the briefs write themselves. Execution starts at the On-Page SEO Checklist.
Further reading & sources
- Creating helpful, reliable, people-first content — Google Search Central
- Schema.org vocabulary — Schema.org
See how your site actually shows up in AI search. An AI visibility audit maps where you’re cited, where you’re invisible, and what to fix first — in plain English.
Get your AI visibility auditTry the free SEO tools →
Prefer self-serve? The interactive checklists turn guides like this one into a working to-do list.
Keep reading in SEO Fundamentals
Get one email when something genuinely changes
AI search moves fast and most of it is noise. We send one short email when a real shift is worth your time. Unsubscribe anytime.
Published by Plain Intelligence — practical AI SEO, GEO, and technical SEO, documented in plain English. About Plain Intelligence →
↑ Back to SEO Fundamentals · Explore all articles · Free tools & resources · Glossary