Keyword Opportunity Analysis: A Step-by-Step Scoring Framework for Prioritising Search Terms
Learn how to judge which keywords are genuinely worth a page and a budget line using a repeatable scoring model that balances demand, business intent, and realistic winnability.

Picture a quarterly planning meeting with forty candidate keywords and one budget line to defend. Without a repeatable scoring method, the answer defaults to whoever argues hardest or whichever term has the biggest volume number - and that is how content budgets get wasted. This guide walks through a keyword opportunity analysis you can defend: a weighted scoring framework built from Search Console evidence, honest treatment of volume data, and a cannibalisation check before anything is briefed.
Why Most Keyword Lists Fail Before Content Is Written
A keyword is only an opportunity if it clears three gates at once: a demand threshold, a business-intent threshold, and a realistic chance for your site to rank. Most exported lists fail at least one gate and nobody notices until three months after publication. A term can show 1,900 monthly searches and still be worthless to a B2B services firm if every ranking page is a government PDF or an AI Overview that answers the question without a click.
Keyword lists are inventories, not decisions. The score is what converts an inventory into a prioritised work queue.
Prerequisites: What You Need Before Scoring a Single Keyword
Define Your Business Goal and Conversion Path
Write down what a winning keyword produces: a demo request, a quote form, an add-to-cart, or a qualified pageview that feeds retargeting. If no page can convert that intent today, the keyword scores zero on business value regardless of volume. A South African IT managed-services firm can convert "outsourced IT support for Cape Town law firms" but has no path to convert "what is a firewall".
Connect Search Console and Assemble a Seed List
You need 12 months of Search Console query data, your sitemap, and a seed list of 40-100 candidate terms drawn from sales calls, support tickets, and competitor overlap. Categorise each seed by funnel stage and by which existing page (if any) targets it. This one column prevents most cannibalisation problems later.
Step 1 - Mine Search Console Before Touching Third-Party Tools
Your own impression data is the only volume signal tied to your actual site, market, and rankings. Filter queries where position sits between 11 and 20 with high impressions and a CTR under roughly 2%. These striking-distance terms are the most trustworthy opportunities you will find, because Google already considers you relevant - you simply are not being chosen.
Spotting Striking-Distance and High-Impression, Low-CTR Queries
Export the Queries tab, add a column for impressions multiplied by (expected CTR at your position minus actual CTR), and sort descending. A query at position 14 with 8,000 impressions and 0.6% CTR represents a far larger recoverable opportunity than a new term with 300 estimated searches. Use the Performance report filters for this; do not rely on memory.
Rule of thumb: any keyword already earning impressions on your domain outranks an identical-volume keyword you have never appeared for.
Step 2 - Respect How Unreliable Keyword Volume Really Is
Why Google Keyword Planner Shows Ranges and Tools Disagree
Keyword Planner buckets low-volume terms into ranges (for example 100-1,000) rather than precise counts, and its figures are modelled from ad auction data, not organic clicks. Third-party tools estimate difficulty from link and content signals that may not match your niche. Two tools can disagree by a factor of five on the same term. Treat every number as directional and band it: low (under 100), medium (100-1,000), high (1,000+).
Sanity-Checking Local Demand with Google Trends
Google Trends will not give you absolute volume, but it shows whether interest in "load shedding backup power" is rising or collapsing, and whether a term behaves differently in South Africa versus global results. Use it to catch dying terms before you publish five pages about them.
Step 3 - Classify Search Intent and Map It to a Page Archetype
Informational vs Commercial Investigation in an AI Overview Era
Read the actual SERP for each keyword. If an AI Overview, featured snippet, or People Also Ask block answers the query above the fold, informational clicks are heavily suppressed. Weight commercial-investigation queries - "cost of", "best", "vs", "for [industry]" - higher, because those searchers still need to visit a page to act. Match each surviving keyword to one archetype: service page, comparison page, pricing page, or guide. One keyword, one archetype, one URL.
| Intent signal | Likely archetype | AI Overview risk |
|---|---|---|
| "what is", "how does" | Guide / explainer | High - answer often absorbed |
| "cost of", "price of" | Pricing page | Moderate |
| "X vs Y", "best for" | Comparison page | Lower - needs judgement |
| "provider for [niche]" | Service landing page | Lowest |
Google's own keyword research fundamentals stress thinking in topics and reader needs rather than isolated strings - which is precisely what intent classification forces you to do.

