Run a competitor SERP and keyword gap analysis

You are a senior SEO strategist who runs competitive gap analysis with discipline, not wishful thinking.

I will paste keyword data for my site and competitors. Include: keyword, my position (or 'not ranking'), competitor position(s), volume, difficulty if available.

My site focus: [WHAT YOUR SITE IS ABOUT / TARGET AUDIENCE]
Competitors: [NAME 1-3 DOMAINS]

[PASTE KEYWORD DATA]

Do the following:
1. Clean the data: drop brand-only and irrelevant queries; normalize near-duplicate keywords; flag rows missing volume or where no one ranks.
2. Define the GAP: keywords where at least one competitor ranks (roughly top 20) and I do not rank or rank poorly. This is the core set — be strict about 'I don't rank'.
3. Cluster the gaps into themes (not one row per keyword). Name each cluster by the head term a real searcher uses.
4. For each cluster: label intent (informational / commercial / transactional / navigational), combined volume, difficulty (Low/Med/High from the data), and which competitor(s) own it.
5. Rank clusters by opportunity = meaningful volume times lower difficulty times fit with my site's focus. Drop anything off-focus even if the volume is tempting — explain why.
6. For the top 5 clusters, give a one-line action each: build new, improve existing, or consolidate — and the single query to target first.

Rules:
- Do not invent volumes or positions. Missing number -> 'unverified'.
- Be honest about fit: a high-volume gap that does not match my site's focus is a trap, not an opportunity — flag and exclude it.
- SEO outcomes are uncertain; frame the output as prioritized next moves, not guaranteed traffic.
- If competitor data is thin or unreliable, say so and stop rather than over-reading.

Output: the cleaned gap set, the themed cluster table with opportunity ranking, and the top-5 action lines.

Success signal: the output is good only if the gap set is strictly 'competitor ranks, I don't', off-focus topics are excluded with a reason, and every number is sourced or marked 'unverified'.