Strategy
Part of How do you build a keyword research strategy in three workstreams?
Running a keyword research strategy framework: owner, log, review
Run a keyword research strategy framework after the build with one owner, one cluster log, a fixed review cadence and a clear rule for retiring clusters.
What to take away
- Name one owner for the frameworkusually a content lead, or whoever writes the briefs in a small team.
- Keep a single source of truth for cluster definitions, with a one-line rationale for every cluster.
- Keep the framework stable for a year and change only the inputs, so comparisons stay fair.
- Review on a fixed cadence and retire a cluster that stops serving the objective, rather than letting it drift.
- A Manchester retailer planning 2027 content should be able to explain every cluster's role in one sentence.
One owner for the framework
Name one owner for the framework: usually a content lead, or whoever writes the briefs in a small team. One owner means one person who can say why a cluster exists, who may change it and when it is next reviewed.
The owner does not do every task. They hold the definitions, run the review and sign off changes, so writers and analysts work from the same sheet. Without that role, each person quietly keeps a private version of the plan and the versions drift apart.
One log for every cluster
Keep a single source of truth for cluster definitions, because separate spreadsheets produce diverging priorities within a quarter. Each row records objective, intent, rationale, owner and next review date.
| Field | What goes in it | Example |
|---|---|---|
| Objective | One sentence naming audience, action and time frame | A Bristol accountancy wants 40 qualified enquiries from small limited companies by March 2027 |
| Search intent | Informational, commercial or transactional | Commercial |
| Rationale | One line on why the cluster was included | It leads straight to an enquiry |
| Owner | The person who writes the briefs | A named writer |
| Next review | The date of the next scheduled check | The end of the quarter |
A one-line rationale makes the next review far quicker. GOV.UK's content guidance states that every piece of published content should meet a valid user need, and the rationale column applies the same test to each cluster. The planning template shows what else a plan may need to record.
Keep the structure stable and change the inputs
Keep the framework stable for a year and change only the inputs. The five steps stay in order: objective, seed set, intent grouping, prioritisation and review cadence. Only the seeds, scores and objective change between reviews.
Five Steps in Order
- State the objective
- Build a seed set
- Group by intent
- Prioritise
- Set a review cadence
Revisit seeds every six months, since seasonal demand shifts and so do the words customers use. Seeds come from sales call notes, support tickets, site search logs and competitor category pages, so ask the people who hold them what has changed.
Sorting seeds into informational, commercial or transactional groups stays the same each time. The guide to intent grouping covers the method, and the log records the result.
The review and the retirement rule
Decide now who reviews cluster performance and when. Quarterly is common. A review compares each cluster with the objective it was built for. GOV.UK's own tools flag content that is overdue for a review, based on a review date someone set when they last updated it.
- Compare each cluster with the objective it serves.
- Check that the one-line rationale still holds.
- Refresh the seed set if six months have passed.
- Keep or retire each cluster, with a reason written in the log.
- Set the next review date.
A cluster that stops serving the stated objective after two full review cycles should be retired. Retire it deliberately rather than letting it drift, and record why, so the next reviewer does not rebuild it from a fresh tool export. If you are building the first quarter from scratch, the ninety day plan sets out a workable sequence.
Tools and the shared glossary
Tools supply volume, difficulty and related terms. They do not supply your margin structure, sales cycle or service capacity. Treat them as measurement instruments, and note in the log which tool and which export date each score came from.
A short glossary stops the team arguing about words. Agree it once and keep it beside the log:
- Seed keyword
- an initial phrase used to generate related terms.
- Cluster
- a group of keywords sharing one search intent.
- Search intent
- what the searcher is trying to achieve.
- Prioritisation score
- a simple ranking of clusters by objective fit, effort and demand.
- Review cadence
- the fixed schedule for checking cluster performance.
Common questions
How many clusters should a small team run?
Three to six is workable for a team of one or two writers. More than that usually means none get finished properly, and the log becomes too long for anyone to keep current.
Does the framework change for Scotland, Wales or Northern Ireland?
No. The structure is the same. Only the seed set changes, because local terminology and competitors differ, and the log records which seeds belong to which place.
When should a cluster be retired?
When it stops serving the stated objective after two full review cycles. Retire it deliberately rather than letting it drift, and write the reason in the log.
How does this connect to a full annual plan?
This framework is the structure; the annual plan is the schedule built on top. The strategy and planning guide for 2027 shows how the two fit together.
