Keyword Research: A Complete Guide

The four-step keyword research process

Step 1: Seed keyword generation. Start with the problems your product solves and the job-to-be-done your buyers describe, not the features you ship. A SaaS marketer does not search for 'content orchestration platform'; they search for 'how to manage content across a team' or 'why our blog stopped getting traffic'. Interview three to five customers about how they found your product and what they searched before discovering it. Their language is your seed list. Supplement with Google's autocomplete, 'People Also Ask', and 'Related searches' for each seed term.

Step 2: Expand using a keyword tool. Feed your seed list into Ahrefs, Semrush, or Google's Keyword Planner. Pull keyword ideas, questions, and phrase-match variants. For each seed, export the top 100 results and filter immediately for two criteria: monthly search volume above 50 (below this, traffic is too variable to plan around), and keyword difficulty below 60 for a new or low-authority domain (below 40 is preferable in year one). This filter typically reduces a 5,000-row export to 200–400 actionable candidates.

Step 3: Classify by intent. For each remaining keyword, search it yourself and look at the top five results. Classify the intent: informational (question-answering), commercial investigation (comparison, 'best of', 'alternatives to'), or transactional (product pages, pricing, 'buy', 'get started'). This classification determines content type, not keyword difficulty. Informational keywords map to blog posts and guides. Commercial investigation maps to comparison pages and alternative-to pages. Transactional maps to product and pricing pages. A keyword research process that does not classify by intent produces a flat list that requires re-evaluation at the briefing stage.

Step 4: Cluster into topic groups. Keywords rarely stand alone — they belong to clusters. 'Content brief template', 'how to write a content brief', 'SEO content brief', and 'what to include in a content brief' are all variations of one topic. Assign every keyword to a parent topic cluster, then make one page the primary target for the whole cluster. Trying to create separate pages for each variation produces thin, competing content that fragments authority instead of concentrating it.

Evaluating keyword difficulty honestly

Keyword Difficulty (KD) scores from tools are directionally useful but mechanically imprecise. A KD of 45 in Ahrefs means the tool estimates that ranking in the top 10 requires a site with a domain rating roughly equivalent to the current top-10 average. What it does not capture is content gap — whether the current top 10 pages actually answer the query well. A high-KD keyword where the top results are thin or outdated is more winnable than the score suggests.

Evaluate actual ranking competition with the following four-point check. First, examine the domain authority of the top 5 ranking pages using Ahrefs' Domain Rating or Semrush's Authority Score. If the top 5 average DR is above 70 and your site is below 30, this is a long-horizon play, not a 90-day target. Second, count the number of ranking pages with fewer than 1,000 words of useful content — thin pages are vulnerable to displacement. Third, check the publication date of the ranking content — content more than two years old on a rapidly evolving topic is often outranked by fresher, more specific alternatives. Fourth, look at whether any ranking pages are exact-match on the specific question the keyword represents. If nobody has answered the query directly, the keyword's effective difficulty is lower than the score suggests.

A useful prioritization heuristic for sites under DR 40: favor keywords where at least one of the following is true — (a) the top 5 contain at least one result from a site with DR lower than yours, (b) the current top result is over 18 months old and the topic is time-sensitive, or (c) the query is a question and no current result has a direct answer in the first 150 words. These are the structural openings where a well-constructed piece can compete immediately.

Keyword prioritization: choosing which to target first

Discovery is not the hard part of keyword research. Prioritization is. Most keyword research exercises produce a list of 100–300 candidates; the constraint is that a team can produce meaningful content for 4–8 keywords per month. The prioritization framework needs to be explicit.

Score each keyword candidate on four dimensions: business value (does ranking for this keyword deliver traffic that converts to the business's actual goal — trial signups, demo bookings, qualified leads?), traffic potential (monthly search volume adjusted for realistic CTR at target position), keyword difficulty relative to your current domain authority, and content production cost (a 3,000-word original-research piece costs more than a 1,000-word how-to guide). Weight business value highest — the most common prioritization error is optimizing for traffic volume rather than traffic quality.

Cluster your prioritized list into three tiers. Tier 1: high business value, achievable KD, lower production cost — target these first. Tier 2: high business value, higher KD — begin these as longer-horizon investments, knowing they will take 3–6 months to rank. Tier 3: lower business value, speculative — park these and revisit quarterly. Allocate 60% of content capacity to Tier 1, 30% to Tier 2, and 10% to experimental Tier 3 plays. This ratio prevents the common trap of spending all capacity on traffic-volume keywords that do not convert.

Maintaining and refreshing keyword research

Keyword research is not a one-time event. Search landscapes shift, competitor content saturates existing clusters, and your site's domain authority changes the set of winnable keywords over time. A keyword research asset that is not refreshed quarterly becomes misleading — it directs content production toward opportunities that no longer exist or misses new gaps that have opened.

Quarterly refresh process: (1) Pull ranking positions for all URLs you have published in the past 12 months using Google Search Console's Performance report. Any page that has dropped more than five positions since last quarter deserves investigation — either a competitor has published a better piece, or the page needs updating. (2) For each active keyword cluster, check whether the top-ranking page has changed. If a new competitor has moved into position 1 with a page that directly copies your structure, flag it for a content update. (3) Run a 'keyword gap' report in Ahrefs or Semrush comparing your site to two or three close competitors — this reveals clusters they are targeting that you have not addressed. (4) Check your existing content for cannibalization: two pages on your own site targeting the same keyword cluster split authority and reduce ranking probability for both. When you find cannibalization, consolidate into one page with a 301 redirect.

For teams using autonomous marketing agents, keyword monitoring can be continuous rather than quarterly. An agent configured to watch ranking changes, competitor new-page publication, and keyword volume shifts can surface prioritization updates as they happen, rather than requiring a manual audit cycle. The output — a refreshed priority list with reasoning — feeds directly into the content brief queue.

RELATED