Voice Search

Voice Search Optimization Audit — Phase 18

Executive Summary

Voice search in 2025–2026 is no longer just smart speakers — it's the entire AI conversational interface layer. Google's voice-driven AI Mode, iOS 18 Siri with deep web integration, and AI Overview voice results on mobile collectively account for 27% of mobile searches (Google internal data, 2024). "Near me" searches have grown 200%+ over 5 years. Optimizing for voice = optimizing for AI answer extraction. Local businesses missing voice optimization lose 20–30% of mobile intent queries to competitors.

Tools for This Phase

Step 1: Read Project Context

Read {AUDIT_DIR}/intake-data.md — business name, URL, location, services. Read {AUDIT_DIR}/local-findings.md — GBP completeness (critical for local voice queries). Read {AUDIT_DIR}/speed-findings.md — mobile speed score (voice requirement: > 70/100 PSI).

2025–2026 Voice Search Context

Where voice queries originate:

2025 voice search stats:

• 27% of mobile searches are voice-based (Google, 2024)
• Voice search results come from pages with mobile PSI score > 70 at 97% rate (Backlinko, 2024)
• "Near me" searches grew 200%+ in 5 years
• Featured snippets are read aloud for 80%+ of voice answers (SEMrush, 2024)
• GBP is the primary data source for 90%+ of local voice responses

Step 2: GBP Voice Readiness (Critical for All Local Voice Queries)

Voice assistants extract local business data directly from GBP. Missing or inaccurate GBP = wrong voice answers.

GBP voice readiness checklist:

Voice query simulation via GBP:

• "Hey Google, what time does [Business Name] close?" → reads Hours
• "Hey Google, call [Business Name]" → reads Phone
• "Hey Google, get directions to [Business Name]" → reads Address
• "Hey Google, does [Business Name] do [service]?" → reads Services + Description

Step 3: Voice Query Types for Local Businesses

Map target voice queries across 4 intent categories:

"Near Me" Queries (Highest Volume)

• "[service] near me"
• "best [service] near me"
• "[service] open now near me"
• "[service] in [neighborhood]"

Ranking factors for "near me" voice results:

1. GBP proximity + verified address ← #1 signal
2. GBP primary category match ← #2 signal
3. GBP review count + rating ≥ 4.0 ← #3 signal
4. Website local relevance (city mentioned in content) ← #4 signal
5. NAP citation consistency across directories ← #5 signal

Business-Specific Queries

• "What time does [Business Name] close?"
• "Is [Business Name] open on Sunday?"
• "Phone number for [Business Name]"
• "How much does [service] cost at [Business Name]?"

Informational/Conversational Queries

• "How do I [service-related question]?"
• "What is the best [service] for [problem]?"
• "How long does [service] take?"
• "What causes [problem that service fixes]?"

Action-Intent Queries

• "Book a [service] appointment"
• "Call [Business Name]"
• "Get directions to [Business Name]"
• "Schedule [service] online"

Step 4: Content Optimization for Conversational Queries

FAQ Content (Primary AIO + Voice Citation Source)

FAQ quality checklist:

Optimal FAQ answer format for voice:

Q: How much does [service] cost in [City]?
A: [Service] in [City] typically costs $X–$Y. [1-sentence context].
   Call [Business] for a free quote: [phone].

(Under 50 words → voice-readable, AIO-citable, featured snippet-optimized)

Featured Snippet Optimization (Voice Reads Snippets 80%+ of Time)

For each target voice query:

• Concise answer in first 40–50 words of the page (no "as mentioned above...")?
• Answer follows H2/H3 question header directly?
• Single clean paragraph or numbered list (2–4 steps)?
• No qualifiers or conditional language in first sentence?

Speakable Schema

SpeakableSpecification marks sections optimal for TTS reading:

{
  "@type": "SpeakableSpecification",
  "cssSelector": [".faq-answer", "h2 + p", ".direct-answer"]
}

Apply on: Homepage summary paragraph, all FAQ answers, key service descriptions.

Step 5: Technical Voice SEO Requirements

Voice results have strict technical requirements:

Schema for voice ranking:

Step 6: AI Assistant Visibility Test

Test across all major AI platforms using voice-style queries:

Test protocol:

Query 1: "Who is the best [service] in [city]?"
Query 2: "Find me a [service provider] near [location]"
Query 3: "[Business Name] — what are their services and hours?"
Query 4: "How do I [most common service-related question]?"

If business not mentioned by ChatGPT/Perplexity: Entity signals too weak — prioritize entity audit (Phase 9) + AIO optimization (Phase 14).

Step 7: Competitor Voice Visibility Comparison

Voice Search Opportunity Matrix

For top 20 voice queries:

Priority Matrix

Scoring

Output

Write findings to {AUDIT_DIR}/voice-findings.md with YAML frontmatter:

---
skill: ai-visibility/voice-search
phase: 18
date: [YYYY-MM-DD]
business: [Business Name]
url: [URL]
score: [X/100]
gbp_complete: [yes|no|partial]
faqpage_schema: [yes|no|partial]
mobile_psi: [X]
featured_snippets: [X]
chatgpt_mentioned: [yes|no]
perplexity_mentioned: [yes|no]
---

Include:

• Score X/100 with per-category breakdown
• GBP voice readiness checklist (all 9 checks)
• AI assistant visibility test results (all 4 platforms)
• Top 20 voice query opportunity matrix (position + optimization status)
• Content gaps for FAQ/conversational content
• Schema audit (present vs. missing, per page type)
• Competitor voice comparison table
• Priority matrix (all actions, Impact × Feasibility scored)
• Quick wins list (items fixable in < 1 hour — expected score gain)
• 30-day voice optimization plan

Key consumers:

• ai-visibility/ai-seo — voice and AIO share many optimization signals
• local/local-seo — GBP completeness feeds both voice + local pack
• audit/speed-optimization — mobile speed is voice prerequisite

Voice Search Quick Reference

10-Step Voice Optimization Action Plan

1. Audit GBP completeness — all 9 checks (name, address, phone, hours, description, services, booking link, Q&A, verification). Effort: 30–45 min. Priority: 25 (5×5).
2. Add FAQPage schema to all service pages — each page needs 3–5 FAQs ≤50 words each. AIO reads FAQPage for 80%+ of voice answers. Effort: 30 min/page. Priority: 25 (5×5).
3. Fix mobile PSI score to ≥70 — voice results pulled from pages scoring ≥70 at 97% rate (Backlinko 2024). Use PageSpeed Insights to identify LCP/INP/CLS issues. Effort: 2–16 hrs depending on root cause.
4. Seed GBP Q&A with 5+ question/answer pairs — focus on "hours", "parking", "cost", "emergency service" patterns. Effort: 30 min. Priority: 20 (5×4).
5. Write conversational FAQ content — 10 FAQ answers using AnswerThePublic clusters. Each ≤50 words, starts with a direct answer. Effort: 1–2 hrs. Priority: 20 (5×4).
6. Add Speakable schema to FAQ sections — targets Google Assistant audio readout. Use speakable property on FAQ container. Effort: 1–2 hrs.
7. Test across 5 AI platforms — run all 4 test queries on ChatGPT, Perplexity, Gemini, Google AIO, Siri. Document results in Voice Opportunity Matrix. Effort: 30 min.
8. Add HowTo schema to process pages — "how to [service]" voice queries trigger HowTo schema readout. Effort: 30 min/page.
9. Optimize content for E-E-A-T voice signals — voice assistants prefer content from established entities. Add author bio, business credentials, years-in-business signals to FAQ pages. Effort: 1–2 hrs.
10. Write 30-day voice optimization plan — record voice-findings.md to {AUDIT_DIR}/voice-findings.md + save PDF report to {REPORTS_DIR}/phase-18-voice.pdf. Effort: 30 min.

INP + Voice Search Connection (2025)

Voice platforms extract answers from fast-rendering pages. INP (Interaction to Next Paint, replaced FID March 2024) is a ranking signal that also affects voice eligibility:

• INP <200ms = Good (voice-eligible tier)
• INP 200–500ms = Needs Improvement
• INP >500ms = Poor (de-prioritized by voice ranking systems)

High INP on FAQ pages is caused by: heavy JS frameworks blocking interaction, lazy-loaded FAQ accordions, and third-party chat widgets. Fix by deferring non-critical JS and using static HTML FAQ sections.

E-E-A-T + Voice Authority (2025)

Voice assistants prefer cited entities with strong E-E-A-T signals:

Write HTML report to {REPORTS_DIR}/phase-18-voice.html and convert to PDF via python3 scripts/generate_pdf.py --html {REPORTS_DIR}/phase-18-voice.html.

• output/report-generation — voice score in master report section 18