SEO/AEO in the LLM Era: B2B SaaS Optimization Roadmap

Overview

As AI-powered search transforms how users discover information, traditional SEO is evolving. Google Search Generative Experience (SGE), ChatGPT, Perplexity, and other LLM-based search engines are changing the game. Answer Engine Optimization (AEO) is emerging as the next frontier.

This post shares real implementation experience from Agent Effi Flow, a B2B SaaS project launched in 3 days. We’ll explore how to transition from traditional SEO to AEO-ready strategies with measurable results.

What Is AEO and Why Does It Matter?

Traditional SEO vs AEO

Traditional SEO:

Optimize for ranking in search results
Target: Human users clicking links
Key metrics: CTR, bounce rate, session duration

AEO (Answer Engine Optimization):

Optimize for being cited by LLMs
Target: AI systems extracting information
Key metrics: Citation rate, accuracy, source credibility

The Shift in User Behavior

Users are asking questions, not typing keywords:

Before: "best receipt OCR API"
Now: "How can I automate receipt processing for my Japanese tax-free shop?"

LLMs synthesize answers from multiple sources. Your content must be:

Structurally clear (headings, lists, tables)
Factually accurate (with data/examples)
Comprehensive (covering related questions)

SEO Foundation: Getting the Basics Right

Before optimizing for AI, traditional SEO fundamentals remain critical.

1. Technical SEO: Site Health Check

Meta Tags Implementation:

// src/components/BaseHead.astro (simplified)
---
const { title, description, image, lang = 'ja' } = Astro.props;
const canonicalURL = new URL(Astro.url.pathname, Astro.site);
---

<head>
  <!-- Basic Meta -->
  <title>{title}</title>
  <meta name="description" content={description} />
  <link rel="canonical" href={canonicalURL} />

  <!-- Open Graph (Facebook, LinkedIn) -->
  <meta property="og:type" content="website" />
  <meta property="og:url" content={canonicalURL} />
  <meta property="og:title" content={title} />
  <meta property="og:description" content={description} />
  <meta property="og:image" content={image} />

  <!-- Twitter Card -->
  <meta name="twitter:card" content="summary_large_image" />
  <meta name="twitter:title" content={title} />
  <meta name="twitter:description" content={description} />
  <meta name="twitter:image" content={image} />

  <!-- Language & Locale -->
  <meta property="og:locale" content={lang} />
  <link rel="alternate" hreflang="ja" href={`${Astro.site}ja${Astro.url.pathname}`} />
  <link rel="alternate" hreflang="en" href={`${Astro.site}en${Astro.url.pathname}`} />
</head>

Key Implementation Points:

Unique title/description per page
Open Graph images (1200×630px)
Canonical URLs to avoid duplicates
Multi-language support with hreflang

2. Sitemap & Robots.txt

Auto-Generated Sitemap (Astro):

// astro.config.mjs
export default defineConfig({
  site: 'https://agent-effi-flow.jangwook.net',
  integrations: [
    sitemap({
      filter: (page) => !page.includes('/admin/'),
      customPages: [
        'https://agent-effi-flow.jangwook.net/api-docs',
        'https://agent-effi-flow.jangwook.net/faq'
      ]
    })
  ]
});

Robots.txt:

User-agent: *
Allow: /

# Crawl rate (for large sites)
Crawl-delay: 1

Sitemap: https://agent-effi-flow.jangwook.net/sitemap-index.xml

3. Performance Optimization

Core Web Vitals Matter for SEO:

Metric	Target	Our Result
LCP (Largest Contentful Paint)	< 2.5s	1.8s
FID (First Input Delay)	< 100ms	45ms
CLS (Cumulative Layout Shift)	< 0.1	0.05

How We Achieved This:

Image Optimization:

// Astro's automatic image optimization
import { Image } from 'astro:assets';
import heroImage from '../assets/hero.jpg';

<Image src={heroImage} alt="Agent Effi Flow"
       width={1200} height={630}
       format="webp" quality={80} />

Code Splitting:
- SvelteKit automatically splits routes
- Each page loads only necessary JavaScript
- Result: Initial bundle < 50KB
CDN & Edge Caching:
- Vercel Edge Network (global CDN)
- Static pages served from edge locations
- Dynamic content cached with stale-while-revalidate

Content Strategy for AEO

1. Structured Content Format

Why Structure Matters:

LLMs parse headings to understand hierarchy
Lists and tables are easily extractable
Code blocks demonstrate technical expertise

Example: FAQ Page for AEO:

## Frequently Asked Questions

### What file formats does the OCR API support?

Agent Effi Flow supports the following image formats:
- **JPEG/JPG** - Recommended for photographs
- **PNG** - Best for documents with text
- **WebP** - Modern format with smaller file size
- **HEIC** - iPhone default format (auto-converted)

Maximum file size: **10MB per image**

### How accurate is the receipt OCR?

Accuracy rates based on 1,000+ test samples:
| Receipt Type | Accuracy | Notes |
|-------------|----------|-------|
| Japanese receipts | 97.2% | Handles vertical text |
| English receipts | 98.5% | Standard horizontal layout |
| Handwritten | 89.3% | Depends on legibility |

Our OCR uses Google Gemini 2.5 Flash with custom prompts optimized for structured data extraction.

### What is the API rate limit?

**Free tier**: 10 requests/hour
**Paid plans**:
- Starter: 100 requests/hour
- Pro: 500 requests/hour
- Business: 2,000 requests/hour

Enterprise customers can request custom limits.

AEO-Friendly Elements:

Direct answers to common questions
Data in table format (easy to parse)
Technical specifics (file sizes, formats)
Clear pricing information

2. Schema.org Markup

Structured Data for Rich Snippets:

<!-- FAQPage Schema -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What file formats does the OCR API support?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Agent Effi Flow supports JPEG, PNG, WebP, and HEIC formats with a maximum file size of 10MB per image."
      }
    },
    {
      "@type": "Question",
      "name": "How accurate is the receipt OCR?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Accuracy rates: Japanese receipts 97.2%, English receipts 98.5%, handwritten 89.3%. Uses Google Gemini 2.5 Flash."
      }
    }
  ]
}
</script>

<!-- Product Schema (for SaaS) -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "SoftwareApplication",
  "name": "Agent Effi Flow",
  "applicationCategory": "BusinessApplication",
  "offers": {
    "@type": "Offer",
    "price": "2000",
    "priceCurrency": "JPY"
  },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.8",
    "reviewCount": "23"
  }
}
</script>

Benefits:

Google shows rich snippets in search results
LLMs can extract pricing/rating information
Increased click-through rate

3. Content Depth: The Wikipedia Effect

LLMs prefer comprehensive sources. A 3,000-word guide outperforms 10 thin 300-word pages.

Our Approach:

Instead of:

Page 1: “What is OCR?” (300 words)
Page 2: “OCR use cases” (400 words)
Page 3: “OCR pricing” (200 words)

We created:

Ultimate OCR Guide (3,500 words):
- What is OCR (technology explanation)
- How our OCR works (technical details)
- Use cases (with real examples)
- Pricing comparison (vs competitors)
- API documentation (code samples)
- FAQ (structured Q&A)

Result: Single comprehensive page ranks for 20+ long-tail keywords.

Keyword Strategy for B2B SaaS

1. Japanese Market Keywords

Our primary target market is Japan, requiring Japanese keyword optimization:

Keyword (Japanese)	Keyword (English)	Monthly Volume	Difficulty
免税 OCR	Tax-free OCR	320	Low
経理自動化	Accounting automation	1,900	Medium
領収書 AI	Receipt AI	890	Low
レシート OCR API	Receipt OCR API	210	Low

How We Found These:

Google Keyword Planner (Japan region)
Competitor analysis (what they rank for)
Google Search Console (queries users type)
Customer interviews (how they describe the problem)

2. Long-Tail Keywords

Generic keywords are competitive. Long-tail keywords drive qualified traffic.

Example:

Generic: “OCR API” (10,000/month, very high competition)
Long-tail: “免税店向けOCR API 日本語対応” (50/month, low competition)

Long-tail conversion rate is 3× higher because searchers have specific intent.

3. Multi-Language SEO

File Structure:

src/content/blog/
├── ja/receipt-ocr-guide.md
├── en/receipt-ocr-guide.md
├── ko/receipt-ocr-guide.md
└── zh/receipt-ocr-guide.md

Implementation:

// src/pages/[lang]/blog/[...slug].astro
export async function getStaticPaths() {
  const posts = await getCollection('blog');

  return posts.map((post) => {
    const [lang, ...slugParts] = post.id.split('/');
    return {
      params: { lang, slug: slugParts.join('/') },
      props: { post }
    };
  });
}

Hreflang Tags:

<link rel="alternate" hreflang="ja" href="https://example.com/ja/blog/post" />
<link rel="alternate" hreflang="en" href="https://example.com/en/blog/post" />
<link rel="alternate" hreflang="ko" href="https://example.com/ko/blog/post" />
<link rel="alternate" hreflang="zh" href="https://example.com/zh/blog/post" />
<link rel="alternate" hreflang="x-default" href="https://example.com/en/blog/post" />

Internal Linking Strategy

Why Internal Links Matter:

Help search engines discover pages
Distribute page authority (PageRank)
Improve user navigation
Increase time on site

1. Hub-and-Spoke Model

Hub Page: Comprehensive guide (e.g., “Ultimate OCR Guide”)

Spoke Pages: Specific topics linking back to hub

“How to integrate OCR API” → links to hub
“OCR accuracy comparison” → links to hub
“OCR pricing guide” → links to hub

Implementation:

---
// src/components/RelatedPosts.astro
const { relatedSlugs } = Astro.props;

const relatedPosts = await Promise.all(
  relatedSlugs.map(slug => getEntry('blog', slug))
);
---

<section class="related-posts">
  <h2>Related Articles</h2>
  <ul>
    {relatedPosts.map(post => (
      <li>
        <a href={`/blog/${post.slug}`}>
          {post.data.title}
        </a>
        <p>{post.data.description}</p>
      </li>
    ))}
  </ul>
</section>

2. Contextual Links

Natural links within content perform better than footer link blocks.

Example:

Our [receipt OCR API](/api-docs/receipt-ocr) uses Google Gemini 2.5 Flash
for structured data extraction. Unlike traditional OCR that returns plain text,
we return JSON with validated schema using [Structured Output](/blog/structured-output-guide).

For accounting workflows, combine this with our [batch processing feature](/features/batch)
to handle hundreds of receipts simultaneously.

Benefits:

High contextual relevance
Users click because content is related
Search engines value semantic connections

Analytics & Measurement

1. Google Search Console Setup

Track SEO Performance:

// Add verification meta tag to BaseHead.astro
<meta name="google-site-verification" content="your-verification-code" />

Key Metrics to Monitor:

Impressions: How often your site appears in search
Clicks: How many users click through
CTR (Click-Through Rate): Clicks / Impressions
Average Position: Your average ranking

Our 30-Day Results (Agent Effi Flow):

Impressions: 1,240
Clicks: 87
CTR: 7.0% (above 5% is good)
Avg Position: 12.3 (improving from 23.1)

2. Google Analytics 4 Integration

Track User Behavior:

// src/layouts/BaseLayout.astro
---
const GA_MEASUREMENT_ID = import.meta.env.PUBLIC_GA_ID;
---

<head>
  <script async src={`https://www.googletagmanager.com/gtag/js?id=${GA_MEASUREMENT_ID}`}></script>
  <script is:inline define:vars={{ GA_MEASUREMENT_ID }}>
    window.dataLayer = window.dataLayer || [];
    function gtag(){dataLayer.push(arguments);}
    gtag('js', new Date());
    gtag('config', GA_MEASUREMENT_ID);
  </script>
</head>

Custom Events for SaaS:

// Track credit purchases
gtag('event', 'purchase', {
  transaction_id: sessionId,
  value: 2000,
  currency: 'JPY',
  items: [{
    item_id: 'starter_plan',
    item_name: 'Starter Plan - 1000 Credits',
    quantity: 1,
    price: 2000
  }]
});

// Track API usage
gtag('event', 'api_call', {
  event_category: 'API',
  event_label: 'receipt_ocr',
  value: 1
});

3. AEO Success Metrics

How to measure AEO performance (since LLMs don’t report referrers):

Direct Traffic Increase:
- Users copy/paste URLs from ChatGPT responses
- Monitor direct traffic in GA4
Brand Search Volume:
- Check “Agent Effi Flow” searches in GSC
- Increase indicates brand awareness from LLM citations
Content Scraping Patterns:
- Monitor server logs for bot user agents
- High bot traffic = content is being indexed by AI
Query Patterns**:

Increase in question-based queries

Example: “How does Agent Effi Flow work?” vs “Agent Effi Flow”

Real Results: 30-Day Case Study

Initial State (Day 0)

New domain (no authority)

0 backlinks

Not indexed by Google

No traffic

Actions Taken

Week 1: Foundation

Submit sitemap to Google Search Console

Add meta tags & Open Graph

Create 5 core content pages:

Homepage

API Documentation

FAQ

Pricing

Legal pages

Week 2: Content Creation

Publish 3 blog posts:

“Receipt OCR Implementation Guide” (2,800 words)

“Tax-Free Processing Automation” (2,200 words)

“Gemini API for Document OCR” (3,100 words)

Add Schema.org markup (FAQPage, Product)

Internal linking between posts

Week 3: Optimization

Image optimization (WebP conversion)

Improve Core Web Vitals (achieve green scores)

Add multi-language support (ja, en, ko)

Create FAQ page with 20 Q&As

Week 4: Outreach & Distribution

Share blog posts on X (Twitter)

Submit to relevant directories

Cross-post on jangwook.net (my tech blog)

Engage in relevant Reddit discussions

Results (Day 30)

Metric Result Notes
Google Index Pages 23 All core pages indexed
Organic Traffic 142 visits From 0
Top Ranking Keyword ”免税 OCR” - Position 8 Targeting position 3
Domain Authority 12 From 0 (normal for new sites)
Backlinks 7 Quality over quantity
CTR from Search 7.2% Above industry average

Top Traffic Sources:

Google Organic: 62%

Direct: 23%

jangwook.net referral: 12%

Social: 3%

Top Landing Pages:

Homepage: 45%

Receipt OCR Guide: 28%

API Documentation: 18%

FAQ: 9%

Key Learnings

Comprehensive content wins:

2,800-word guide outperforms 3 short articles

Users spend 4:30 avg on long-form content vs 1:15 on short

Technical accuracy matters:

Posts with code samples have 3× longer session duration

Users bookmark technical guides (high return visitor rate)

Multi-language pays off:

Japanese content gets 70% traffic (our target market)

English content gets 20% (international interest)

Korean gets 10% (K-pop tourism niche)

FAQ pages are underrated:

15% of organic traffic lands on FAQ

High conversion rate: 22% sign up after reading FAQ

AEO Advanced Strategies

1. Optimize for Citation-Worthy Content

What makes content citation-worthy for LLMs?

Factual Accuracy:

Cite sources (research papers, official docs)

Include data with dates

Avoid marketing fluff

Clear Attribution:

According to Google's [Gemini API pricing](https://ai.google.dev/pricing) as of November 2024, Gemini 2.5 Flash costs $0.075 per 1M input tokens.

Comparisons with Data:

Feature Agent Effi Flow Competitor A Competitor B
Accuracy 97.2% 94.5% 89.3%
Price/1000 calls ¥2,000 ¥3,500 ¥2,800
Languages 4 2 3

2. Answer the Next Question

Users ask follow-up questions. Anticipate them.

Example: Receipt OCR Guide

Main question: “How does receipt OCR work?”

Anticipated follow-ups:

“What’s the accuracy rate?” → Add accuracy section

“How much does it cost?” → Include pricing

“Can I try it for free?” → Link to free tier

“What languages are supported?” → List languages

“How do I integrate the API?” → Code examples

Result: Single page answers entire user journey, reducing bounce rate.

3. Code Samples for Technical SEO

Developers search differently. They want copy-paste code.

Bad Example:

Our API is easy to integrate. Just send a POST request with your image.

Good Example:

// Install SDK npm install @agent-effi-flow/sdk // TypeScript Example import { EffiFlowClient } from '@agent-effi-flow/sdk'; const client = new EffiFlowClient({ apiKey: process.env.EFFI_API_KEY }); async function processReceipt(imagePath: string) { const result = await client.receiptOCR({ image: fs.readFileSync(imagePath), language: 'ja' }); console.log('Store:', result.store_name); console.log('Total:', result.total_with_tax); console.log('Items:', result.items); return result; } // Example output { "store_name": "FamilyMart", "purchase_date": "2024-11-26", "items": [ { "name": "おにぎり鮭", "quantity": 2, "unit_price": 130, "total_price": 260 } ], "subtotal": 260, "tax": 26, "total_with_tax": 286 }

Why This Works:

Developers can test immediately

Shows real output format

Demonstrates error handling

LLMs extract this for code generation

Next Steps: AEO Implementation Roadmap

Week 1-2: Content Audit

Identify thin content:

Pages with < 500 words

Low engagement (< 30s avg time)

High bounce rate (> 70%)

Consolidate or expand:

Merge related pages into comprehensive guides

Add data, examples, code samples

Update outdated information

Add structure:

Convert paragraphs to lists where appropriate

Create comparison tables

Add FAQ sections

Week 3-4: Schema Implementation

Add Schema.org markup:

FAQPage for FAQ pages

Article for blog posts

Product for SaaS pages

HowTo for tutorials

Test with Google Rich Results Test:

https://search.google.com/test/rich-results

Monitor rich snippet appearance:

Check Google Search Console

Track CTR improvements

Month 2: Internal Linking Overhaul

Create pillar content:

1-2 comprehensive guides (3,000+ words)

Cover entire topic depth

Build spoke content:

5-10 focused articles

Each links to pillar page

Automate related posts:

// Use Claude LLM for semantic recommendations const recommendations = await generateRecommendations(post);

Month 3: AEO Optimization

Question-based content:

Identify common questions (Google “People Also Ask”)

Create dedicated answer pages

Structure with clear headings

Citation-worthy claims:

Add sources for all statistics

Link to original research

Include publication dates

Multi-format content:

Text (for reading)

Code samples (for developers)

Tables (for comparison)

Diagrams (for concepts)

Preview: Next Post

In the next article, we’ll dive deep into advanced AEO strategies:

How to optimize content specifically for ChatGPT, Perplexity, and Google SGE

Measuring AEO performance with custom analytics

Building an AI-friendly knowledge graph

Case study: Increasing LLM citations by 300%

Stay tuned!

Resources

Tools Used

Google Search Console - SEO monitoring

Google Analytics 4 - Traffic analytics

Schema Markup Generator - Structured data

PageSpeed Insights - Performance testing

Reference Articles

Google Search Central - Official SEO docs

Schema.org - Structured data vocabulary

Moz Beginner’s Guide to SEO - SEO fundamentals

Answer Engine Optimization Guide - AEO introduction

Related Reading from jangwook.net

AI Content Recommendation System - LLM-based semantic matching

Specification-Driven Development - Building with Claude Code

Individual Developer AI SaaS Journey - 3-day MVP to production

Conclusion

SEO is not dead—it’s evolving. The fundamentals (site speed, mobile-friendly, clear structure) remain critical. But the frontier is AEO: optimizing for AI systems that synthesize answers.

Key Takeaways:

Start with SEO basics: You can’t optimize for AI if humans can’t find you

Comprehensive > Thin: One 3,000-word guide beats 10 shallow pages

Structure everything: Headings, lists, tables make content machine-readable

Answer the full journey: Anticipate follow-up questions

Measure, iterate, improve: Use GSC and GA4 to track progress

For B2B SaaS specifically:

Technical content (with code) attracts developers

Multi-language support expands reach

FAQ pages convert better than marketing copy

3-Month Goal: Position 3 for primary keyword (“免税 OCR”), 500+ monthly organic visits, 10% conversion to free trial.

Let’s build discoverable products in the AI era.

Metric	Result	Notes
Google Index Pages	23	All core pages indexed
Organic Traffic	142 visits	From 0
Top Ranking Keyword	”免税 OCR” - Position 8	Targeting position 3
Domain Authority	12	From 0 (normal for new sites)
Backlinks	7	Quality over quantity
CTR from Search	7.2%	Above industry average

Feature	Agent Effi Flow	Competitor A	Competitor B
Accuracy	97.2%	94.5%	89.3%
Price/1000 calls	¥2,000	¥3,500	¥2,800
Languages	4	2	3

Reading Complete!

Overview

What Is AEO and Why Does It Matter?

Traditional SEO vs AEO

The Shift in User Behavior

SEO Foundation: Getting the Basics Right

1. Technical SEO: Site Health Check

2. Sitemap & Robots.txt

3. Performance Optimization

Content Strategy for AEO

1. Structured Content Format

2. Schema.org Markup

3. Content Depth: The Wikipedia Effect

Keyword Strategy for B2B SaaS

1. Japanese Market Keywords

2. Long-Tail Keywords

3. Multi-Language SEO

Internal Linking Strategy

1. Hub-and-Spoke Model

2. Contextual Links

Analytics & Measurement

1. Google Search Console Setup

2. Google Analytics 4 Integration

3. AEO Success Metrics

Real Results: 30-Day Case Study

Initial State (Day 0)

Actions Taken

Results (Day 30)

Key Learnings

AEO Advanced Strategies

1. Optimize for Citation-Worthy Content

2. Answer the Next Question

3. Code Samples for Technical SEO

Next Steps: AEO Implementation Roadmap

Week 1-2: Content Audit

Week 3-4: Schema Implementation

Month 2: Internal Linking Overhaul

Month 3: AEO Optimization

Preview: Next Post

Resources

Tools Used

Reference Articles

Related Reading from jangwook.net

Conclusion

Read in Other Languages

Was this helpful?

About the Author

Kim Jangwook

Related Articles

jangwook.net 45-Day Growth Report: 7 Insights from 750 Visitors

Adding Chinese Support to Your Blog: A Practical Guide to Multilingual Website Expansion

Figma MCP to Web Components: Research-Based Design System Automation