Keyword research is not about finding the biggest numbers — it is about finding demand you can actually capture and monetize. You do not get to invent demand; you rank for searches people already type, and your job is to find the ones you can realistically win.
Three filters decide whether a keyword is worth your time: can your site plausibly rank for it, is it attached to real buying or decision intent, and does it tie to something you actually earn on. Get those right and content becomes a repeatable revenue engine; get them wrong and you pour months into beautiful pages that rank for terms nobody searches with a wallet open. This is the 2026 version of the process — written for the reality of AI Overviews eating informational clicks, keyword-difficulty scores that lie with a straight face, and a search landscape where most Google searches never send a click to the open web. New terms are defined in the SEO glossary; the wider picture is in SEO fundamentals.
Before volume, before difficulty, ask what the searcher actually wants. There are four intent types, and matching intent matters more than any number for an affiliate. Informational queries want to learn ("how does a VPN work"). Navigational queries want a specific site ("Coinbase login"). Commercial-investigation queries compare before buying ("best VPN for streaming," "NordVPN vs ExpressVPN") — and this is where most affiliate value lives. Transactional queries are ready to act ("buy Ledger wallet," "Notion pricing"). A high-volume informational term can be nearly worthless to a monetized page because the searcher is not in a buying mindset, while a lower-volume commercial term converts far better. You read intent off the live SERP: the content type, format and angle that already rank tell you what Google has decided satisfies that query. If the results are all stores, a guide will not rank there.
| Intent | Example query | Funnel | Best content type for affiliates |
|---|---|---|---|
| Informational | "how does a VPN work" | Top | Guide / explainer — builds authority, feeds links (often AI-Overview-eaten) |
| Commercial investigation | "best VPN for streaming", "X vs Y" | Middle | Comparisons, "best X" lists, reviews — the money zone |
| Transactional | "NordVPN coupon", "Notion pricing" | Bottom | Deal / pricing / where-to-buy — most AI-resilient |
| Navigational | "Coinbase login" | Outside | Only worth targeting your own brand terms |
Seed keywords are your broad one-or-two-word starting points — the roots of the tree — and the quality of everything downstream depends on them. Brainstorm from three angles: the products you promote, the problems your audience has, and competitor brands. Then expand. Keyword tools (Ahrefs Keywords Explorer, Semrush Keyword Magic) turn seeds into thousands of variants and surface related terms the top pages also rank for. Google autocomplete, People Also Ask and related searches are free idea engines straight from Google — expand one question and it spawns more. AnswerThePublic maps question and comparison phrasings, which are almost always long-tail. Competitor gap analysis shows terms rivals rank for that you do not, but pick comparable competitors, not the giants. And if you already have a site, Google Search Console is gold: the Queries report reveals terms you rank for but never deliberately targeted. Layer on modifiers — best, top, review, vs, alternative, cheap, pricing, for small business — to multiply commercial intent variants.
Every tool number is an estimate, and reading them honestly is what separates operators from dashboard-watchers. Search volume is a trailing-12-month average, so it hides seasonality and is not traffic — the average, the two-thirds of searches that end without a click, and shared SERP real estate all erode it. Keyword difficulty is directional, not absolute: Ahrefs computes it mostly from backlinks to the top pages, Semrush uses a multi-factor formula, and the same keyword gets different scores in different tools. Treat KD as a triage filter for big lists, never a verdict — always confirm with a manual look at the SERP. CPC is a useful proxy for buyer intent, because advertisers only bid high on terms that convert ("how does a VPN work" is cheap; "best VPN deal" is not). The most important reframe is traffic potential over single-keyword volume: a top page ranks for hundreds of keywords, so judge a topic by the total traffic the current number-one page actually gets, and sort by that — never by raw volume. No third-party tool has ground-truth numbers except Google, and even Google's tools hedge, so compare relative, not absolute, and never trust one tool as truth.
Long-tail keywords are longer, more specific, lower in volume and competition, and usually higher in intent — and the vast majority of all searches are long-tail. For a new or low-authority site they are the entry point, because head terms are brutally competitive and long-tails are the first terms you can actually win. They also convert better: "keto diet" is someone learning, "best keto supplement for beginners" is someone near a purchase. The strategic move is to stop treating keywords as a flat list and start grouping them into topics — a broad pillar page owning the head term, surrounded by cluster pages each owning a specific long-tail. That interlinked structure is how you signal depth and rank for many related terms at once rather than chasing exact matches one at a time. This is the heart of content clusters and topical authority, and your long-tail research is the raw material that fills them out.
Map each keyword to intent and funnel stage, then prioritize with a simple filter: real intent you can serve, low enough difficulty to rank, and genuine business value. Anchor your money pages on commercial and transactional intent ("best," "review," "vs," "alternative," "pricing"), build informational content around them to establish authority, and internally link the informational pages toward the money pages — a bottom-funnel page with a few qualified visitors can out-earn a blog post with thousands of casual readers. Business value is decisive: a keyword you cannot monetize is worth zero to you no matter its volume. Gate ambition by site authority — a new site starts on low-difficulty long-tail, a growing site owns the cluster then attacks the head term. And do not overlook striking-distance wins: pages already sitting on the second page of Google (positions 11–30 in Search Console) often rank faster from a refresh than a brand-new term ranks from scratch.
The live SERP is the honest answer key, so read it before writing a word. Check the authority of the top ten against yours — if they far outrank you, look elsewhere for now. Check whether you can realistically out-create the incumbent; a "best air purifier" page built on eight years of testing is expensive to beat. Match the format Google is already rewarding: if the results are all comparison tables, an essay will not rank; if the SERP is all free tools, you cannot write your way in. Watch for Reddit, Quora and forum results, which now appear on roughly a quarter of commercial searches — a signal Google wants first-hand experience for that query, and a warning that a thin affiliate page may struggle. And note how much of the SERP is consumed by features and AI answers before any classic result appears.
Two shifts define keyword research now. Zero-click search is the majority: leading 2026 clickstream data puts roughly two-thirds of US Google searches ending without a click to the open web, up from prior years — though the exact figure swings by methodology. AI Overviews appear on a meaningful and volatile share of searches (conservatively 20%+ of all queries, higher on tracked commercial sets), and where they appear, click-through to the top organic result drops sharply. The practical consequence is not "SEO is dead" — it is a reweighting. Simple informational and definitional queries ("what is X," "X definition") are the biggest losers, satisfied on-page without a click. Commercial-investigation and transactional queries stay valuable for affiliates — transactional AI Overviews have actually receded — because people still click to verify, compare and buy, and because queries needing current pricing, first-hand testing or trusted sources resist AI synthesis. Google disputes the click-loss narrative; independent data contradicts it. Either way, the operator move is the same: weight effort toward the middle and bottom of the funnel, win on original testing and clear authorship, and measure earnings per click, not raw traffic. The trade-off with paid channels is covered in SEO vs paid traffic.
One primary keyword — in the title, H1 and meta — plus five to ten supporting terms woven naturally into the body and subheadings. A well-covered page ranks for dozens of variations, so you target a cluster of related terms, not a single string. Never assign the same primary intent to two pages, or they cannibalize each other.
Low-competition long-tail keywords with clear commercial or transactional intent. Skip the high-volume head terms you cannot rank for yet, build depth around one topic to establish authority, and read each SERP first — if the top ten are all high-authority brands or free tools, move on. Expect a few months before long-tails start ranking.
Selectively. Simple "what is" and "how does" queries are increasingly answered by AI Overviews without a click, so they are weak traffic plays. But informational content still earns its place when it builds topical authority and links internally to your money pages, or when the topic needs genuine first-hand experience and current data AI cannot authentically synthesize.
Neither is right in absolute terms; they use different data and formulas, and only Google has the real numbers. What matters is that they are rank-correlated — they agree on which keywords are relatively bigger or harder. Pick one tool, use it consistently for comparison, and validate the shortlist against the live SERP.
The SEO foundations affiliates need: how search works, what ranks, and how to build content that earns durable, high-intent organic traffic in the...
Core · 13 min readOn-page SEO makes you eligible to rank, not guaranteed.
Core · 12 min readPublishing more posts is not a strategy.
