Keyword Strategy: How to Score, Assign, and Retire Keywords (A Practical Cycle)
A keyword strategy is a set of decisions, not a spreadsheet. Here is how to run the quarterly cycle that scores terms, assigns ownership, and kills what no longer earns its place.
The Spreadsheet Is Not the Problem: Why Mid-Sized Teams Stall on Keyword Decisions
Picture a mid-sized omnichannel retailer with three SEO and content people and one part-time analyst. They export a keyword sheet with roughly 4,000 rows, sort it by search volume, and find the top 300 terms are already owned by competitors or by their own homepage. Six weeks later, half of the mid-tail queries on that sheet are still unanswered, and two new blog posts are quietly competing for the same query.
Nothing here is a research failure. The team can find keywords. What they cannot answer is 'why this keyword, why now, and who is writing it?' That is a strategy failure, and it costs real money: at an illustrative estimated R8,000-R25,000 per in-house brief-write-review cycle (a range that varies widely with seniority and sector), every keyword sent to briefing without a decision is a budget commitment made by default.
A keyword list tells you what exists. A keyword strategy tells you what you will not do, who owns each query, when you re-decide, and what evidence retires a page.
What You Need in Place Before You Set a Single Keyword Rule
Strategy design collapses when the inputs are missing. Before you write a single rule, confirm five things:
- A named decision-maker with authority to say no to a keyword, not just yes.
- A crawl of your live URL inventory mapped to the queries each page currently targets, so you can spot internal competition before it starts.
- Access to two data sources minimum (for example, a keyword tool plus Search Console impressions) so volume is never a single-source claim.
- A defined publishing route with review capacity, whether in-house or via a scalable search growth platform.
- A measurement owner who will report on outcomes in business terms, not rankings alone.
A 20-question search quality framework is worth reading at this stage - Google's own people-first content guidance sets the quality bar your selection rules need to respect.
Step 1: Define the Business Outcome the Keyword Programme Serves
Scope is the first of three layers in a working keyword strategy:
| Layer | What it decides | Example artefact |
|---|---|---|
| Scope | Which business outcomes keywords must move | One-page brief: 'support 120 additional online conversions per quarter from non-brand queries' |
| Selection rules | How terms are scored, ranked and matched to content types | Priority formula plus intent taxonomy |
| Governance | Who owns, when to re-score, what retires a page | Keyword ledger with status and review dates |
Write the outcome as a sentence with a number and a channel. 'Improve SEO' is not a scope; 'increase non-brand organic enquiries for fitting-room services by 30% year on year' is.
Step 2: Build an Intent Taxonomy That Includes Comparison Queries
Most sheets label keywords 'informational' or 'transactional' and stop. That misses the class of query that now feeds AI answers: the evaluation query. Use five labels:
- Informational - 'how does X work', 'what is Y'.
- Commercial investigation - 'best X for Y', 'X reviews'.
- Comparison/evaluation - 'X vs Y', 'X compared to Y for Z'. These are frequently summarised directly in AI Overviews and chat responses.
- Transactional - 'buy X', 'X price', 'X near me'.
- Navigational - brand and product-name searches you should already own.
Label comparison queries separately. A comparison page is written, structured and measured differently from a transactional page, and folding both into one brief is one of the fastest routes to a page that satisfies neither.
Step 3: Score and Rank With a Formula, Not a Sort Column
Volume as the only sort input guarantees you chase the terms with the most competition. Use a weighted priority score instead. The formula below is an illustrative worked example, not a published industry standard - tune the weights to your business:
Priority = (Intent weight x Commercial value) / (Effort x Cannibalisation risk)
Intent weight: 1.0 informational ... 4.0 transactional
Commercial value: 1 (R0) ... 5 (high-margin / core service)
Effort: 1 (easy) ... 5 (needs new templates, data or integrations)
Cannibalisation: 1 (no existing page) ... 3 (two live URLs already compete)Applied to five hypothetical queries for an omnichannel furniture retailer:
| Query | Type | Intent | Value | Effort | Cannib. | Score |
|---|---|---|---|---|---|---|
| 'furniture' (head) | Informational | 1.0 | 3 | 5 | 3 | 0.20 |
| 'best sofa for small flats' | Commercial | 3.0 | 4 | 2 | 1 | 6.00 |
| 'leather vs fabric sofa' | Comparison | 2.5 | 3 | 1 | 1 | 7.50 |
| 'shop online' | Transactional | 4.0 | 2 | 3 | 3 | 0.89 |
| 'sofa delivery Cape Town' | Transactional | 4.0 | 4 | 1 | 1 | 16.00 |
The ranking flip is the point. The head term has the highest volume and the lowest score; the long, boring delivery query wins outright. If your sheet does not produce that flip, it is a research file, not a strategy.
The Three Layers of a Working Keyword Strategy (Scope, Selection Rules, Governance)
Keep the three layers visible in one document. Scope is the business outcome from Step 1. Selection rules are the intent taxonomy and priority formula from Steps 2 and 3, plus the content type each label maps to (comparison page, guide, category page, location page). Governance is the part most teams never build: ownership, re-score cadence, kill criteria and reporting. Without the third layer, a strategy degrades into a re-sorted spreadsheet within a quarter.
If your document cannot answer 'who owns this query and when do we re-score it?', you have a research file with a strategy title on it.