Keyword Research for 2026: Targeting Intent in an AI Search World

In 2019, keyword research meant finding a high-volume, low-competition term and building content optimised for that exact phrase. In 2026, that approach doesn't just underperform — it fundamentally misunderstands how both Google and AI search systems work. AI search queries average 23 words compared to 4 words on traditional Google. Users describe full situations, contexts, and problems. "Accountant Auckland" has become "What should I look for when choosing an accountant for a small business in Auckland that handles both personal and company tax?" This structural shift renders short-tail keyword hunting nearly irrelevant for the fastest-growing category of search queries.

The good news: the intent-first, topic-cluster approach that 2026 demands is actually more durable and defensible than keyword-density optimisation ever was. A business that deeply understands what its customers are trying to accomplish — and creates comprehensive, well-structured content that serves those intentions — builds search visibility that doesn't evaporate with algorithm updates. This guide covers the complete evolution of keyword research practice for 2026 and the practical methodology to execute it. For the broader strategy context, see our complete SEO & AIO strategy guide.

How AI Has Changed the Keyword Research Discipline

The AI search revolution has created several structural shifts in keyword research that practitioners must internalise before any tactical work makes sense.

Queries are longer and more contextual. According to Semrush's 2026 AI Search Trends report, the shift to longer, complex queries is accelerating — users entering full questions like "What's the best CRM for a 50-person marketing agency with Salesforce integration that costs under $150 per user monthly?" rather than fragments like "CRM software pricing." Searches beginning with "tell me about..." jumped 70% year-over-year in 2025 (Google data). Traditional keyword research tools — which measure search volume for specific query strings — struggle to capture this long-tail, conversational demand accurately.

AI systems prioritise topics, not keywords. Google's NLP capabilities have advanced to the point where keyword density targeting is counterproductive. Google and AI platforms now understand context and relationships between concepts, prioritising comprehensive topic coverage over keyword matching. As Exploding Topics noted in March 2026, "traditional keyword research focuses on individual search queries, but AI search platforms like Google's AI Overviews, ChatGPT, and Perplexity understand topics, context, and relationships between concepts — creating content around individual keywords just doesn't work anymore." The practical implication: keyword research should produce a topic map, not a keyword list.

Zero-click is the new normal for informational queries. 60% of all searches now end without a click, rising to 83% when AI Overviews appear. For purely informational keywords with straightforward factual answers, even ranking #1 may produce minimal traffic as AI Overviews answer the query directly on the SERP. The keyword research implication: prioritise keywords where users need to take action or make decisions — commercial, transactional, and decision-support queries — over pure information queries where zero-click rates are highest.

AI citation potential is a new keyword value dimension. Beyond search volume and competition, keywords that generate AI-cited responses represent a form of visibility that doesn't require a click. Appearing in an AI Overview or ChatGPT response builds brand recognition and trust even when no click occurs, and drives eventual direct searches and conversions. Identifying keywords with high AI Overview appearance rates — primarily informational queries with complex, multi-faceted answers — is now a legitimate research objective alongside traditional volume/competition analysis. For the GEO-specific approach to earning AI citations, see our SEO & GEO strategy guide.

Intent-First Classification: The Foundation of Modern Keyword Research

Intent classification — understanding what a user is fundamentally trying to accomplish with a search query — is not new. What's new in 2026 is that intent-first thinking must happen before any volume or competition analysis, not after. Building a keyword list and then assigning intent labels is backwards; the right approach is to map the intentions of your customers first and then identify keywords that represent those intentions.

The traditional four-category intent framework (informational, navigational, commercial, transactional) remains valid but needs nuancing for 2026. Research from SE Ranking shows that 70% of searches have informational intent, 22% commercial intent, 7% navigational intent, and 1% transactional intent — but these proportions shift dramatically by industry, audience, and stage in the buyer journey. For B2B professional services, commercial intent queries (comparing options, evaluating providers) dominate the high-value traffic. For ecommerce, transactional queries drive direct revenue. Understanding the intent distribution specific to your business is more valuable than applying generic benchmarks.

The extended intent categories that matter in 2026:

Informational intent — the user wants to learn something. Examples: "What is topical authority?" "How does Google's local algorithm work?" These queries trigger AI Overviews most heavily (over 88% of AI Overview appearances are informational). High zero-click risk, but excellent for brand awareness, topical authority building, and AI citation potential. Content format: comprehensive explainer articles, guides, definition pages.

Commercial investigation intent — the user is actively researching options before making a decision. Examples: "Best SEO agencies Auckland," "Local SEO tools comparison 2026," "Google Business Profile vs Bing Places." High commercial value, moderate AI Overview appearance. Content format: comparison articles, roundup guides, best-of lists, feature comparison tables. These are the highest-value informational keywords for service businesses.

Decision-stage intent — the user is close to making a decision and seeking validation or finalising criteria. Examples: "Involve Digital reviews," "Is [agency name] worth it?" "[Service] pricing NZ." Very high commercial value, primarily navigational/transactional. Content format: pricing pages, testimonials, case studies, detailed service pages.

Transactional intent — the user is ready to take immediate action. Examples: "Hire SEO consultant Auckland," "Book SEO audit," "Get local SEO quote NZ." Highest conversion intent, lowest volume. Content format: service pages with clear CTAs, landing pages, contact pages. These keywords are the primary targets for conversion-focused SEO programs.

AI-research intent (emerging in 2026) — the user is using AI tools as their primary research mechanism rather than traditional search. These queries are longer, more context-rich, and often don't produce measurable search volume in traditional tools because they happen entirely within ChatGPT, Perplexity, or Claude. Capturing this intent requires understanding what questions your customers ask AI tools, not just what they type into Google — a gap that traditional keyword research tools don't fill.

Question Keyword Mining: The Highest-Value Research Activity in 2026

If there is a single keyword research activity that delivers the highest return in 2026 across both traditional SEO and AI search visibility, it is question keyword mining — identifying and targeting the specific questions your target audience asks about your topic area.

Question keywords are valuable for multiple reasons. They trigger People Also Ask (PAA) appearances in Google SERPs — and the Related Searches / PAA feature appears in 85% of SERPs (SE Ranking 2025). They match the natural language structure that AI systems receive as user queries, making content that directly answers them more likely to be cited. They correspond directly to FAQ schema content, which is among the most AI-citable structured data types. And they represent clear intent — a question keyword tells you exactly what the user wants to know, eliminating the guesswork of inferring intent from short, ambiguous terms.

The best sources for question keyword mining:

People Also Ask (PAA) box. Google's PAA box is a live feed of related questions users are actually asking. Search your primary keywords and harvest every PAA question. Note that PAA is recursive — clicking one question reveals more related questions below it. A single seed keyword can generate 50-100 related question keywords through recursive PAA expansion. Tools like AlsoAsked.com automate this mapping process, visualising the question hierarchy that Google's PAA data implies.

Google Autocomplete. Type your seed term into Google and note the autocomplete suggestions. Then modify the query with question words: "how to [seed term]," "why is [seed term]," "what is [seed term]," "when should [seed term]." Each produces a different set of autocomplete suggestions that represent actual user queries. Google Autocomplete is the most accurate available signal of what people are actually typing — it's drawn directly from real user query data.

Answer Socrates and AnswerThePublic. These tools specifically visualise question-format keywords around a seed term, organised by question word (what, why, how, who, when, where). They provide a rapid view of the question landscape for any topic. Answer Socrates provides free, real-time data; AnswerThePublic provides broader visualisations but charges for full access.

Reddit and online communities. NZ-specific subreddits, Facebook groups, and community forums reveal the actual language your customers use and the questions they ask in natural conversation. This is the most underutilised keyword research source for 2026 — and it's free. The questions asked in community forums are precisely the questions that AI systems receive as conversational queries. Building content that answers these questions positions you as the reference for AI systems handling similar queries.

Customer service and sales interactions. Every question your team receives from prospects and customers is a keyword research data point. Systematically collecting and categorising these questions reveals a keyword set that no tool can replicate: the exact language your specific customer profile uses, about the specific problems your product or service solves. For businesses with a sales team, the questions asked in initial conversations are the highest-value keyword research source available.

Topic Cluster Mapping: Building the Keyword Architecture

Topic cluster mapping is the process of organising identified keywords into a hierarchical structure that guides content creation. It replaces the flat keyword list with an architectural view of which content pieces should exist, how they relate to each other, and which keywords each piece should target. For both Google topical authority and AI search coverage, the map is the strategy.

The mapping process starts with identifying pillar topics — the broadest, most comprehensive subjects your business has authority in. A pillar topic should correspond to a broad, high-volume search category that encompasses dozens of related subtopics. For an Auckland SEO agency, "SEO for NZ businesses" is a pillar topic. For a Wellington accountancy firm, "tax planning for small businesses NZ" is a pillar topic.

Below each pillar topic sit cluster topics — the specific subtopics, questions, and angles that compose the pillar. Each cluster topic becomes one piece of content. The cluster topics for "SEO for NZ businesses" might include: local SEO for NZ businesses, technical SEO foundations, keyword research for NZ businesses, link building in NZ, Google Business Profile optimisation, AI search visibility for NZ businesses — and so on. Each of these cluster topics is targeted at a set of related keywords, linked to the pillar page, and interlinked with sibling cluster articles on related topics.

Identifying cluster topic gaps is as important as identifying covered topics. Topical coverage completeness — having content that addresses every significant question in your space — is a core signal of topical authority. Use tools like Ahrefs' Content Gap analysis or Semrush's Topic Research to identify subtopics your competitors rank for that you don't cover. These gaps represent content opportunities where you can claim authority with new cluster articles.

NLP-Based Competitor Gap Analysis: Finding What's Missing

Competitor keyword gap analysis — identifying keywords your competitors rank for that you don't — is one of the highest-value activities in a mature keyword research program. It identifies proven demand (if competitors rank for it, the keyword has traffic potential) in areas where you can potentially displace them.

Traditional gap analysis compares your ranking keyword set against competitors' using tools like Ahrefs' Content Gap, Semrush's Keyword Gap, or manual comparison of organic keyword reports. The 2026 layer adds NLP-based semantic gap analysis: rather than just comparing exact keyword matches, tools now identify topic areas and concept clusters that competitors cover comprehensively where you have thin or absent coverage. Semrush's Topic Research and Ahrefs' content tools both offer versions of this semantic clustering.

The practical competitor gap analysis process: identify your top 3-5 organic competitors (they may differ from your business competitors — they're the businesses ranking for the same keywords you want to rank for, not necessarily the same services you offer). Export their top keywords from Ahrefs or Semrush. Filter for keywords in positions 1-10 (high ranking, therefore proven authority). Identify keywords where all or most competitors rank, but you don't appear — these are the highest-priority gap opportunities. For each gap, ask: is this a topic we have authority in? Is it genuinely relevant to our audience? Can we create better content than what currently ranks? If yes to all three, it becomes a content priority.

The AI search gap analysis layer is newer and requires manual work: test 10-20 high-priority queries in ChatGPT and Perplexity. Note which businesses and sources are cited in AI responses. If competitors are regularly cited and your business never appears, this represents an AI visibility gap that requires entity building (more third-party mentions, stronger citation signals) alongside content improvement.

AI Citation Potential: The New Keyword Value Dimension

Beyond traditional volume/competition/intent analysis, 2026 keyword research requires evaluating a fourth dimension: AI citation potential — the likelihood that content targeting this keyword will be cited in AI-generated responses.

Keywords with high AI citation potential share several characteristics: they're informational with complex, multi-faceted answers (AI Overviews appear for 88%+ of informational queries). They have question format or can be naturally answered in question-answer structure. They represent topics where authoritative, comprehensive content is more valuable than thin, fast answers. They correspond to areas where your business has genuine expertise and can create definitively authoritative content rather than generic summaries.

The AI citation potential framework can be applied to any keyword: evaluate whether the query typically generates an AI Overview in Google Search Console (check AI Overview appearance rate in Search Appearance filters). Test the keyword in ChatGPT and Perplexity to see what sources are currently cited. Assess whether your content can provide a more comprehensive, well-structured answer than current AI sources. If your content can be the definitive source — not just a good source — the AI citation potential is high.

Brands are 6.5 times more likely to be cited through third-party sources than their own domains (Position Digital, 2026). This means that for keywords with high AI citation potential, the off-page strategy (earning mentions in industry publications, building review presence, generating original data) is as important as the on-page content strategy. AI citation is an ecosystem play, not a single-page optimisation.

Keyword Research Tools in 2026: The Essential Stack

The keyword research tool landscape has evolved significantly in 2026, with AI capabilities now integrated into most major platforms. Here is the practical tool stack we recommend, with the specific use case for each:

Ahrefs Keywords Explorer — 28.7 billion keyword database, the most honest traffic estimates in the industry (using "traffic potential" to show total traffic achievable from a page, not just one keyword). Best for competitor keyword analysis, backlink gap identification, and discovering new keyword opportunities. Ahrefs added an AI content helper that scans top-ranking pages and identifies missing topics in your content — valuable for content gap analysis. Pricing from $129/month.

Semrush Keyword Magic Tool — feature-rich with strong intent classification, clustering capabilities, and personalized keyword difficulty (shows difficulty specific to your site's authority, not a generic score). The Semrush One platform adds AI Visibility tracking across ChatGPT, Perplexity, and Google AI Overviews — making it the strongest choice for businesses that want unified traditional SEO and AIO measurement. Pricing from $199/month (Semrush One).

Google Search Console — free, and provides the most accurate real-user query data for your own site. The Performance report shows exactly which queries are generating impressions and clicks for each page. The AI Overview appearance filter (in Search Appearance) shows which of your keywords are triggering AI Overviews — critical data for identifying zero-click risk and AI citation opportunities. Non-negotiable for any keyword research program.

AlsoAsked.com — specifically visualises the People Also Ask relationship map for any seed keyword, showing which questions are related and how they connect. Essential for question keyword mining and FAQ content planning. Free tier available; paid plans for higher volume research.

Answer Socrates — free, real-time question keyword generation around any seed term. Good complement to AlsoAsked for expanding question keyword sets without cost.

Surfer SEO — content optimisation tool that analyses top-ranking pages for a target keyword and identifies the semantic topics, NLP terms, and content structure patterns that correlate with ranking. The most useful tool for translating keyword research into content briefs. Strong AI content writer integration for those creating AI-assisted content.

Google Trends — free, essential for understanding keyword seasonality, identifying emerging topics before they peak in search volume tools, and comparing relative query volume across geographic regions (useful for NZ vs. global comparisons).

ChatGPT and Perplexity — direct AI testing as keyword research. Enter your target queries and observe: what sources are cited? What format do AI responses take? What related questions are generated? This qualitative research provides insight that no tool database captures — the actual AI search experience your customers are having. For the keyword research angle on technical SEO implementation, see our technical SEO foundations guide.

Keyword Cannibalism: The Most Common Implementation Error

Keyword cannibalism occurs when multiple pages on the same website target the same keyword or keyword cluster, causing Google to be uncertain which page to rank for that query — often resulting in neither ranking well. It's the most common and most easily avoidable keyword research implementation error, yet it's endemic in websites that have been publishing content for several years without a deliberate keyword architecture.

The symptoms: Google Search Console showing multiple URLs switching positions for the same keyword (the ranking page changes between months). Pages from the same site appearing multiple times in the same SERP for a query (a temporary signal that often precedes one page being deranked). Rankings improving then declining without any external explanation (Google settling on which page to rank, then re-evaluating). Traffic distributed thinly across multiple similar pages rather than concentrated on one authoritative piece.

The diagnosis: use Semrush's Position Tracking or Ahrefs' Rank Tracker, filter by URL, and identify keywords where 2+ pages from your site appear. Alternatively, use Google Search Console and filter by URL to see which keywords are shared across multiple pages with similar positions.

The fix options, in order of preference: Consolidation — merge the competing pages into one comprehensive piece, redirecting the absorbed page to the consolidated URL. This concentrates all link equity and content quality on one page. Differentiation — if both pages genuinely target different aspects of the same topic (different stages of buyer journey, different audience), make the differentiation explicit in content, intent signals, and internal linking. Noindex — if one page is genuinely thin and redundant, noindex it and focus link equity on the primary page. Consolidation is almost always the right answer for content cluster pages targeting overlapping keywords.

Keyword Research for AI Search: The Distinct Methodology

Traditional keyword research optimises for what people type into Google. AI keyword research must additionally optimise for what people type into ChatGPT, Perplexity, and ask Google AI Overviews. The methodologies overlap significantly but have important distinctions.

Traditional keyword research tools don't measure AI query volume — there's no "search volume" for ChatGPT queries in Ahrefs or Semrush. AI keyword research is therefore more qualitative: identifying the types of queries your audience likely asks AI tools, based on observed search behaviour, customer conversations, and direct testing in AI systems.

The key insight from Semrush's 2026 AI Search Trends research: queries are getting longer and more scenario-specific. Instead of "CRM software pricing," users ask "What's the best CRM for a 50-person marketing agency with Salesforce integration that costs under $150 per user monthly?" Content that answers the specific scenario question — not just the general category question — is what gets cited in these AI responses. The implication for keyword research: look beyond head terms and even traditional long-tail to the scenario-specific, situational questions that describe your buyers' exact contexts.

Practical AI keyword research approach: interview your best customers about what questions they've asked ChatGPT or Perplexity in their buying process. Test 20 scenario-specific queries in ChatGPT and Perplexity — note which competitors appear and what content they're using. Review your sales team's call notes for the complex, contextual questions prospects ask. Mine community forums (Reddit, Quora, Facebook groups) for the long-form questions your audience posts. These sources reveal the AI-era keyword landscape that no tool database captures. For how this connects to GEO and AI citation strategy, see our GEO for B2B high-consideration services guide and our piece on why AI-referred leads convert better.

Building Your 2026 Keyword Research Workflow

Bringing together all of the above into a practical, repeatable workflow:

Step 1: Audience and intent mapping (Week 1). Document the key buyer personas: who buys from you, what problems do they have, what are they trying to accomplish? Map their journey: how do they first become aware of their problem? How do they research solutions? What information do they need to make a decision? This becomes the intent map that directs all keyword research.

Step 2: Seed keyword identification (Week 1). Brainstorm 20-50 seed terms that represent the core topics of your business — service names, industry terms, category descriptions. These are starting points, not targets. Enter each into Google Search Console to see which you already rank for, and Ahrefs/Semrush to understand the volume and competition landscape.

Step 3: Topic cluster mapping (Week 2). Organise seed keywords into topic clusters. For each cluster, identify the pillar topic (the comprehensive overview page) and the cluster topics (specific subtopics, questions, and angles). This produces your content architecture map.

Step 4: Question keyword expansion (Week 2). For each cluster, mine question keywords using AlsoAsked, Answer Socrates, and Google Autocomplete. Expand each pillar and cluster with the 5-10 most relevant question keywords. These become your FAQ sections and PAA/AI citation targets.

Step 5: Competitor gap analysis (Week 3). Run Ahrefs Content Gap or Semrush Keyword Gap analysis. Identify the top 20-30 keywords your competitors rank for that you don't. Evaluate each for relevance and content creation priority. Add the highest-value gaps to your content roadmap.

Step 6: Prioritisation (Week 3-4). For each identified keyword or keyword cluster, score it on: business value (how much does ranking for this drive conversions?), AI citation potential (high or low), competition level (can we realistically rank?), and content creation difficulty (do we have the expertise and resources?). Build a prioritised content roadmap based on this scoring.

Step 7: Content brief creation and execution. For each priority keyword cluster, create a detailed content brief: target keywords, user intent, competitor analysis (what currently ranks and why), suggested structure (headings, FAQ topics), internal linking plan, and schema requirements. Assign to writers and execute in priority order.

For the complete picture of how keyword research connects to on-page implementation, topical authority, and the technical foundations that make content discoverable, see our technical SEO guide and the strategic SEO & AIO complete guide.

Keyword research done well is the foundation that determines whether your content investment drives business results. Our Growth Plan Generator analyses your current keyword positioning, identifies the highest-value opportunities in your market, and generates a prioritised content roadmap. Build your personalised growth plan with Involve Digital.

Get Started Using The Form Below

This keyword research guide is part of Involve Digital's complete SEO and AIO strategy series. For how keywords translate into content architecture, see our main SEO & AIO complete guide. For the technical foundations that help your keyword-targeted content get indexed and cited, see our technical SEO foundations guide. For the local SEO keyword angle specific to NZ businesses, see our local SEO NZ guide.

‍