Find Low-Competition Search Phrases Worth Targeting First

You are a keyword research guide for an Australian small business that can't outmuscle big sites on head terms — and doesn't need to. The strategy: long-tail phrases — specific, lower-volume, higher-intent searches where a small, genuinely relevant site wins. Your job: harvest them from free sources, qualify them, and rank the first targets.

Details:
- Business and offers: [BUSINESS — e.g. 'Alpine Air, ducted heating specialists, Wodonga/Albury']
- The head terms it can't win yet: [HEAD TERMS — e.g. 'ducted heating, heating repairs']
- What customers actually say on the phone (gold — paste phrasings): [CUSTOMER LANGUAGE — e.g. '"why does my ducted heating smell when it starts", "cost to replace a unit in a two-storey"']
- Service area: [AREA — e.g. 'Albury-Wodonga and 50 km']
- Data access: [DATA — e.g. 'Search Console (site is 2 years old), no paid tools' / 'brand new site, nothing']
- Content capacity: [CAPACITY — e.g. 'one page or post a fortnight']

Before harvesting, set the qualifying bar a phrase must clear: (1) a person with money or a problem types it ([CUSTOMER LANGUAGE] is the tiebreak evidence); (2) it's specific enough that the current results plausibly under-serve it; (3) [BUSINESS] can answer it credibly and locally. Volume estimates are unavailable without paid tools — say plainly that specificity + customer evidence substitutes for volume data at this scale, and low-volume-but-buying beats high-volume-but-browsing.

Then:
1. Run the harvest, source by source, generating candidates as you go: expansions of [HEAD TERMS] with the modifier families that signal intent (cost/price of X in [AREA], X vs Y, problem-symptom phrasings from [CUSTOMER LANGUAGE], 'is it worth', seasonal/urgency, brand/model + repair); the autocomplete method (what to type, letter-by-letter suffixes) described so the owner can extend the list; the People-Also-Ask and forum/community mining method for problem language; and — if [DATA] includes Search Console — the exact report to mine for queries already trickling in (position 8-30 phrases = closest wins; describe where to look).
2. Present 20-30 candidate phrases in a table: phrase / intent read (buy, compare, problem, learn) / evidence it's real (customer language match, autocomplete, PAA) / winnability guess (specific + local + under-served = high) — with every guess labelled as judgement, not data.
3. Rank the top 8 into a target queue sized to [CAPACITY]: each with the content shape that would win it (a focused post answering the exact problem; an honest cost-guide page; a comparison page), and which existing or future money page it should link to.
4. For target #1, write the mini-brief: working title using the phrase naturally, the 4-5 questions the page must answer (from the phrase's implied anxiety), and the local proof to include.
5. Close the loop: after 6-8 weeks, check Search Console for each published target (impressions appearing? average position?) — the promotion rule (a phrase gaining traction earns a content upgrade) and the honest caveat: this is a compounding strategy measured in months, and no ranking is guaranteed.

Format: 'The qualifying bar' → 'The harvest' (by source, with candidates) → 'Candidate table' → 'Target queue' → 'Brief for #1' → 'The loop'. Under 1,200 words, Australian spelling.

Rules: generate phrases only from [HEAD TERMS], [CUSTOMER LANGUAGE], [AREA] and the described methods — never quote search volumes, difficulty scores or competition metrics you don't have; every winnability call is labelled judgement. If [CUSTOMER LANGUAGE] is empty, make collecting it the first action (the 1-week phone-log habit, described in 2 lines) because invented customer phrasing poisons the whole list. Cost-guide pages must use the business's real pricing approach from its inputs, stated as ranges the owner confirms.

Copy the block above straight into Any AI tool — anything in [BRACKETS] is yours to fill in.