Google Ads Keyword Research

Pre-requisite

Read ~/.claude/skills/google-ads/knowledge.md in full before doing anything. Pay special attention to sections 2 (long-tail data), 3 (match types), 8 (4-question filter), and 9 (research methodology).

If the current date is 6+ months after the research date, re-fetch the long-tail keyword data URL and match type trends URL from section 10.

Gather inputs

If not already known from context, ask:

• •Business description and services offered
• •All locations served
• •Average booking value and profit margin (needed for max CPC calculation)
• •Existing keyword lists (if any, to build on)
• •Landing pages available (needed for the 4-question filter)

Phase 1: Generate seed keywords

For each service × location combination, produce:

Transactional intent seeds (highest priority)

Keywords where the searcher is ready to buy/book. Pattern: [action] + [service] + [location]

• •"hire saxophonist edinburgh"
• •"book wedding band glasgow"
• •"corporate entertainment edinburgh"

Commercial investigation seeds (second priority)

Keywords where the searcher is comparing options. Pattern: [best/top/reviews] + [service] + [location]

• •"best wedding saxophonist edinburgh"
• •"wedding musician reviews scotland"

Long-tail variants (high priority — reference section 2 data)

4–6 word phrases with strong intent. Pattern: [service] + [for] + [event type] + [location]

• •"live saxophonist for wedding reception edinburgh"
• •"saxophone player for corporate awards dinner"

Informational seeds (low priority — flag as top-of-funnel only)

• •"how much does a wedding saxophonist cost"
• •"what to expect from live saxophone at wedding"

Mark these clearly as informational. Only bid if there's a deliberate awareness strategy.

Phase 2: Apply the 4-question filter

For each keyword, evaluate:

1. •Would someone searching this actually book? → Keep only if yes
2. •Can we write an ad that directly answers this search? → If not, the keyword is too generic
3. •Does the landing page match? → Map each keyword to a specific landing page
4. •At estimated CPC and conversion rate, is cost per booking acceptable? → Calculate: estimated CPC / estimated CVR = estimated CPA. Compare to break-even CPA.

Remove any keyword that fails a filter. Flag borderline cases.

Phase 3: Negative keywords

Generate a comprehensive negative keyword list:

Universal negatives (apply to all campaigns)

• •Learning/education: lessons, tutorial, how to play, learn, course, beginner, school
• •Free content: free, download, mp3, spotify, youtube, backing track, karaoke
• •Employment: salary, job, vacancy, career, hiring
• •Equipment: reed, mouthpiece, fingering, sheet music
• •Budget signals: cheap, budget, DIY, free

Industry-specific negatives

Based on the business type, add negatives that would attract wrong-intent clicks.

Location negatives (if relevant)

Exclude locations outside the service area.

Aim for 100+ negative keywords minimum. Reference knowledge base section 7 — neglecting negatives is a top-5 mistake.

Phase 4: Organisation

Present keywords in a structured table:

Group by city campaign → ad group (service type).

Phase 5: Validation workflow

Give the user a step-by-step guide to validate these keywords using free tools:

Google Keyword Planner

1. •Go to Tools → Keyword Planner → "Discover new keywords"
2. •Enter the seed phrases from Phase 1
3. •Set location to [specific city]
4. •Sort results by: Avg. monthly searches, then Competition, then Suggested bid
5. •For each keyword, record: volume, competition level, suggested bid range
6. •Also try: "Start with a website" → enter 2–3 competitor URLs

Google Keyword Planner — per-city comparison

Run the same seeds with location set to each target city separately. Note differences in volume and CPC — these determine budget allocation.

Google autocomplete

Open incognito browser. Type each seed phrase partially. Screenshot/record all suggestions. These are real searches.

Google Trends

Compare term variants: e.g. "saxophonist" vs "sax player" vs "saxophone player". Check which is more common in each target region. Check seasonality.

Search term review (post-launch)

After 1–2 weeks of running ads: go to Keywords → Search Terms report. This shows actual queries. Add high-converters as exact match. Add irrelevant terms as negatives. This is the most valuable data source — review weekly.

Output format

Produce:

1. •Keyword lists per campaign (city) and ad group (service), with intent tier and recommended match type
2. •Negative keyword list (100+ terms)
3. •Keyword-to-landing-page mapping
4. •Validation workflow checklist
5. •Seasonal notes (if relevant — e.g. wedding booking peaks Jan–Mar)

Context: $ARGUMENTS