The Complete Keyword Research Framework

From a Single Seed Word to a Search Strategy Your Competitors Can’t Copy

Most keyword research isn’t research at all — it’s data collection. Someone opens a tool, types one word, exports five hundred rows into a spreadsheet, and calls it a strategy. That spreadsheet then sits untouched because nobody built a system to turn raw keyword data into decisions. Real keyword research is a discipline with layers: you start with one word, and you deliberately move through expansion, categorization, clustering, prioritization, and validation before a single piece of content gets written.

This post is a systematic walkthrough of that discipline — built for e-commerce and content teams who want a repeatable process rather than a one-off audit. We’ll go from the seed keyword all the way through Google Keyword Planner and its forecasting tools, then into the harder-to-teach parts: the observational skill and intuition that separates a strategist from a tool operator, and a practical method for reverse-engineering what your competitors are actually doing.

Part 1: The Six-Layer Framework, Starting From One Word

Before opening any tool, understand the shape of the process you’re about to run. Skipping a layer is why most keyword lists never turn into rankings.

Layer 1 — The Seed Keyword

A seed keyword is not something you target. It’s a doorway you walk through to find the keywords you’ll actually target. A good seed is the smallest, most neutral unit that describes what your business sells, solves, or is known for — a product category, a core problem, or a service name, stripped of modifiers.

The mistake most people make is inventing seeds from imagination. Real seeds come from evidence you already have sitting around the business:

• Your own product and category names, exactly as customers would say them — not internal naming conventions.
• Google Search Console’s Queries report — the phrases you already rank for, even at position 40, are seeds Google has already told you are relevant to your site.
• Customer support tickets, sales call transcripts, and live chat logs — this is unfiltered customer language, and it’s usually more honest than anything a keyword tool suggests first.
• Review platforms — Trustpilot, G2, Amazon reviews, App Store reviews — read how real buyers describe the problem you solve, in their words, not yours.
• Competitor homepage and navigation copy — not to copy it, but to see which category words an established player has already validated with traffic.

Write down 15–25 seeds before you touch a keyword tool. This single habit fixes more keyword strategies than any tool upgrade.

Layer 2 — Expansion

Expansion is the process of multiplying each seed into dozens or hundreds of real search phrases. Use several expansion methods together, because each one surfaces a different type of query:

• Google Autosuggest — type the seed and note every suggestion; then append a letter (seed + “a”, seed + “b”…) to force Google to reveal suggestions it hides behind the default list. This is the classic “alphabet soup” method.
• People Also Ask (PAA) — click into two or three boxes; PAA regenerates and reveals nested questions real users are asking around your seed.
• Related Searches at the bottom of the SERP — these are Google’s own machine-generated adjacent queries, and they’re an underused free source of clustering ideas.
• Question-based expansion — manually pair the seed with who / what / why / how / does / can / should / vs, since question intent is where AEO and GEO content increasingly gets surfaced in AI answers.
• Forums and communities — Reddit, Quora, and niche subreddits or forums for your industry. Search “seed + reddit” directly in Google to pull threads Google itself considers relevant enough to rank.
• Marketplace search bars — for e-commerce, the Amazon search bar autosuggest reflects actual purchase-intent phrasing, which is often different from Google’s informational-leaning suggestions.
• YouTube search bar — useful for identifying tutorial and comparison intent that written content sometimes misses.
• Keyword tool “Discover new keywords” features (Google Keyword Planner, Semrush, Ahrefs) — saved for last deliberately, because if you open the tool first, you anchor on whatever it gives you instead of what the market is actually saying.

By the end of Layer 2, one seed should have produced somewhere between 50 and 300 raw phrases, unsorted.

Layer 3 — Categorization by Intent

Every phrase from Layer 2 needs to be tagged against one of four intents before it means anything strategically:

• Informational — the searcher wants to understand something (“what is crawl budget”). Funnel stage: top.
• Navigational — the searcher wants a specific brand or site (“Alneeko case studies”). Funnel stage: brand-aware, any point.
• Commercial investigation — the searcher is comparing options before buying (“best technical SEO agency for Shopify”). Funnel stage: middle-to-bottom.
• Transactional — the searcher is ready to act (“hire technical SEO consultant”, “SEO audit pricing”). Funnel stage: bottom.

Intent tells you the content format before you write a word: informational intent wants an explainer or guide; commercial investigation wants a comparison, review, or case study; transactional wants a landing page, pricing page, or product page. Assigning the wrong format to the right keyword is one of the most common reasons content gets traffic but never converts.

Layer 4 — Clustering Into Topics

Keyword-by-keyword thinking is outdated. Modern search engines (and answer engines) reward topical depth, which means your unit of strategy should be the cluster, not the individual phrase.

The fastest reliable way to cluster manually: search each candidate keyword and look at the top 10 results. If the same URLs keep reappearing across several different phrases, Google is telling you those phrases share intent and belong on the same page or in the same content hub — regardless of how different the words look on the surface. If the top 10 results are completely different sets of pages for two phrases that look similar, treat them as separate topics even if they share words.

Once clusters are formed, name each one after its core intent (e.g., “Core Web Vitals — diagnostic,” “Core Web Vitals — Shopify-specific fixes”) and assign one owning page or pillar per cluster.

Layer 5 — Prioritization

With clusters formed, score them rather than picking by gut feeling alone. A simple weighted formula works well for most teams:

Priority Score = (Business Value × 3) + (Search Demand × 2) + (Achievability × 2) + (AI-Answer Visibility Potential × 1)

• Business Value — does a person searching this actually become a lead or a sale, or are they just curious?
• Search Demand — volume and trend, sourced from Keyword Planner and cross-checked with Search Console.
• Achievability — realistic assessment of whether your domain, given its current authority and content depth, can compete for this cluster in a reasonable timeframe.
• AI-Answer Visibility Potential — for GEO/AEO work specifically: does this query type currently trigger AI Overviews, Perplexity answers, or ChatGPT browsing citations? If yes, structuring content to be citation-worthy (clear definitions, source-graded claims, structured data) matters as much as ranking position.

Rank every cluster and work top-down. Resist the temptation to chase the highest-volume cluster first if its achievability score is low — that’s how teams burn six months on content that never moves.

Layer 6 — Validation

Before committing writer time or design time to a cluster, validate it once more:

• Re-check the live SERP for the primary keyword in the cluster — has it changed since you last looked? SERPs shift, especially with AI Overviews now appearing on more query types.
• Check Search Console for near-match queries you already get impressions for but aren’t ranking well on — these are often the fastest wins because Google already considers you relevant.
• Check for seasonality — a cluster that looks dead in July might spike every November; Keyword Planner’s historical trend view (covered next) will show this.

Part 2: Mastering Google Keyword Planner and Forecasting

Keyword Planner is a free tool built for advertisers, but it remains one of the most reliable sources of directional search demand data available to organic strategists — as long as you understand what it’s actually built to measure.

Getting Set Up

Keyword Planner sits inside Google Ads (ads.google.com/aw/keywordplanner). You don’t need an active, spending campaign to access it — a free Google Ads account is enough. Inside, there are two core tools worth mastering: “Discover new keywords” and “Get search volume and forecasts.”

Tool 1: Discover New Keywords

Enter one or more seed phrases, or a URL (your own, or a competitor’s), and Keyword Planner returns a list of related keyword ideas with the following columns:

• Average monthly searches — shown as a range on accounts without recent ad spend, and closer to an exact figure on accounts with active spend history.
• Competition — Low / Medium / High.
• Top of page bid (low and high range) — the estimated CPC range advertisers pay.

VERIFIED Keyword Planner’s “Competition” column measures ad auction competition among advertisers bidding on that term — it is not a measure of organic SEO ranking difficulty. A term can show “Low” ad competition and still be extremely hard to rank for organically, and vice versa. Treat it as a paid-media signal, not an organic difficulty score.

Filters worth using every time: location (narrow to your actual market, especially important for a Karachi-based team serving Frankfurt-facing or international clients), language, date range, and the option to exclude brand/adult terms so your seed list doesn’t get diluted with irrelevant results.

Tool 2: Get Search Volume and Forecasts

This is the more powerful tool for strategists, and the most underused. Paste your own list of keywords (up to 10,000 at once) instead of relying only on Google’s suggestions, and Keyword Planner will return two views:

• Historical metrics — monthly search volume trend over the past 12–24 months per keyword, letting you spot seasonality and growth or decline patterns at a glance.
• Forecast — a projected view of clicks, impressions, cost, CTR, and average CPC for the list, assuming a hypothetical ad campaign targeting those terms.

Even if you never intend to run ads, the Forecast tab is valuable to an organic strategist for one reason: it lets you compare relative demand and relative competitive intensity between two lists of keywords using the same methodology, on the same day, under the same settings — which is more reliable than eyeballing volume numbers from different tools that each use different data sources.

Reading the Forecast Tab, Step by Step

1. Set your date range — Keyword Planner defaults to the next 7 days, but you can extend this to 30, 90, or a custom range to smooth out short-term noise.
2. Set location and network targeting to match your real target market — forecasts are meaningless if they’re scoped to the wrong country.
3. Choose your match type — Broad, Phrase, or Exact. This single toggle can change the forecasted click and impression volume dramatically, because broad match pulls in a wider net of related queries. When comparing keyword sets, keep the match type identical across every list you test, or the comparison isn’t valid.
4. Read the Clicks vs. Cost curve — the chart plots how forecasted clicks change as hypothetical daily budget increases. The point where the curve starts to flatten tells you roughly where demand for that keyword set saturates — useful even for organic prioritization, because it’s an indirect signal of how large the addressable search demand actually is.
5. Compare like-for-like — run your top three clusters through the Forecast tool separately, using identical settings, and rank them by projected impressions. This gives you a demand-based tiebreaker when two clusters look similar on paper.

Forecasting Caveats Every Strategist Should Know

VENDOR CLAIM Google’s own Keyword Planner help documentation states that search volume is shown as a rounded range rather than an exact number for accounts without a recent, active spend history, and becomes more precise as an account’s ad spend increases.

DISPUTED How much this bucketing distorts real-world demand estimates — and whether third-party tools like Semrush or Ahrefs are meaningfully more accurate — is genuinely debated among practitioners. Some find third-party volume estimates closer to actual Search Console impression data for their own sites; others find all volume tools, including Keyword Planner, similarly rough as absolute numbers. Treat every volume figure, from any source, as directional rather than exact, and cross-check against your own Search Console impressions wherever the keyword already gets any visibility.

Practical takeaway: use Keyword Planner for keyword discovery, relative comparison between clusters, seasonality trends, and commercial-intent signals (a high top-of-page bid is a strong proxy that a keyword has real buyer value, even if you never run an ad against it). Use Search Console and, budget permitting, a paid tool like Semrush or Ahrefs, to sharpen the absolute numbers once a cluster has been shortlisted.

Part 3: Developing Observational Skill and Keyword Intuition

Tools give you data. Intuition tells you which data actually matters, and it is trained through repetition — not innate talent. The following five practices, done consistently, build that instinct.

Practice 1: Guess Before You Check

Before opening any tool for a given keyword, write down your own prediction: the intent, the content format that will rank (listicle, product page, comparison table, video, forum thread), and a rough volume guess. Then check the SERP and the tool. Score yourself. This single habit, repeated across 15–20 keywords a day for a few weeks, compresses years of passive experience into a deliberate feedback loop.

Practice 2: Live in Real Customer Language

Spend time in reviews, Reddit threads, support tickets, and sales call transcripts on a recurring basis — not as a one-time research task, but as an ongoing habit. Customers describe problems differently than marketers do, and the gap between the two is where under-served, low-competition keyword opportunities usually hide.

Practice 3: Keep a Language Journal

Maintain a running document of raw phrases pulled directly from customer-facing sources, tagged by date and source. Revisit it monthly as a seed bank — phrases that show up repeatedly across unrelated sources are a strong signal of unmet search demand before any tool confirms it.

Practice 4: Study SERPs Outside Your Own Niche

Deliberately look at search results in industries you don’t work in. Patterns repeat across niches — “how to” queries tend to pull a featured snippet plus a video pack; “best X” queries tend to pull listicles alongside a forum thread or two; “X vs Y” queries tend to pull comparison tables. Building a mental library of these patterns across many industries sharpens pattern recognition faster than only ever looking at your own vertical.

Practice 5: Calibrate Against Real Numbers, Repeatedly

The same discipline that builds a “feel” for where an SAP configuration will break after seeing enough process failures applies directly here: guess first, check the real number second, and note the size of the gap. Over weeks, the gap shrinks, and what started as a guess becomes a calibrated instinct you can trust under time pressure — which matters enormously when a client wants a prioritization call in a meeting, not a week later after a full tool-based audit.

Part 4: Reverse-Engineering a Competitor’s Keyword Strategy

Competitor research is not “look at what they rank for.” Done properly, it reconstructs the decisions behind their content — which tells you what to do next, not just what they already did.

Step 1: Separate SERP Competitors From Business Competitors

A SERP competitor is whoever occupies page one for your target keywords, regardless of industry. A business competitor sells the same thing to the same buyer. They overlap less often than people assume — a review blog or a Reddit thread can be your real SERP competitor for a given cluster even though it isn’t a business rival. Track both lists separately, because the strategy to outrank each is different.

Step 2: Crawl Their Site Structure

Run a Screaming Frog crawl of a competitor’s domain and export the URL list. Group URLs by folder (/blog/, /guides/, /vs/, /comparison/, /resources/) to infer how they’ve organized their content strategy. A dedicated /vs/ folder, for example, signals a deliberate commercial-investigation content strategy you can measure yourself against cluster by cluster.

Step 3: Run a Content Gap Analysis

Using a keyword gap report (Semrush or Ahrefs both offer this natively), compare your ranking keyword set against a competitor’s. The output that matters most is the list of keywords they rank for that you have no page targeting at all — not the keywords where you’re both already competing. Gaps are where the fastest net-new opportunity usually sits.

Step 4: Inspect On-Page Structure and Schema

View source, or use Screaming Frog’s custom extraction, to see what schema markup a competitor uses (Article, FAQPage, Product, HowTo), how they structure headings, and where they place internal links pointing toward commercial pages. This tells you which pages they consider most valuable internally — their own internal linking is effectively a vote of confidence you can observe from the outside.

Step 5: Track SERP Movement Around Their New Content

Set up simple manual or tool-based rank tracking for your priority clusters, and note when a competitor publishes new content and its rankings move. Treat every jump as market feedback: if a competitor’s new page climbs quickly, the market is telling you that cluster has real, current demand worth reacting to.

Step 6: Check Backlink and PR Patterns

A quick backlink report shows who links to a competitor and why — guest posts, PR mentions, directory listings, partnership pages. This isn’t keyword data directly, but it explains part of why their content outranks yours even on comparable on-page quality, and it often reveals acquisition channels worth copying independently of keywords.

Step 7: Check AI Answer Engine Citations (GEO/AEO Layer)

For shared queries, check whether a competitor’s content is being cited inside AI Overviews, Perplexity answers, or ChatGPT browsing responses. This is a newer, increasingly important competitive signal that sits alongside classic rank tracking — a competitor can be losing organic rank share while gaining AI-citation share, and vice versa, so track both rather than assuming they move together.

Part 5: A Repeatable Weekly Ritual

Systems beat one-off audits. A simple weekly cadence keeps every layer of this framework alive without demanding a full week of dedicated research time:

• Monday — seed review: scan Search Console, support tickets, and reviews for new raw language; add to the seed bank.
• Tuesday — Keyword Planner pass: run new seeds through Discover New Keywords and Get Search Volume and Forecasts; update the priority scoreboard.
• Wednesday — competitor crawl: rotate through one competitor’s site structure, gap analysis, and schema per week rather than trying to audit everyone at once.
• Thursday — clustering and prioritization: fold new keywords into existing clusters or spin up new ones; re-run the priority score.
• Friday — content brief handoff: turn the week’s top-scoring cluster into a content brief with intent, format, target schema, and source-grading requirements built in from day one.

Closing Thought

Search is no longer a single channel with one set of rules. The same keyword research discipline that used to feed Google rankings now also feeds what AI Overviews, Perplexity, and ChatGPT choose to cite as an answer. That raises the stakes on getting the fundamentals right, not lowers them: understanding real intent, building content around validated clusters, and developing the pattern recognition to know which data actually matters, all matter more in a world with more places to be found — not fewer.

Treat keyword research as a system you run every week, not a project you finish once. The seed word is where it starts. Everything after that is discipline.