How We Built a 21,000 Impression SEO Engine in 90 Days with Zero Agency Spend

The Starting Point: Zero Organic Traffic

In December 2025, Aira had no organic search presence. Zero blog content. No indexed pages beyond the homepage. Competing against established companies like Smith.ai, Ruby, and Dialzara, each with years of content and thousands of backlinks.

The typical advice would be to hire an SEO agency. Spend $3,000 to $10,000 per month. Sign a 12 month contract. Wait 6 to 12 months for results. Instead, we decided to use our own product: an AI SEO agent that runs the entire workflow autonomously.

This case study documents exactly what happened over the next 90 days, with real numbers from Google Search Console, GA4, and our internal tracking.

The Strategy: Let the AI Agent Drive

The approach was simple: let the AI SEO agent handle the entire strategy and execution. The agent followed its 7 phase cycle (review, audit, research, strategy, create, publish, measure) and ran 11 complete optimization cycles in 90 days.

Here is what the agent decided to do:

• Target service business verticals (dental, HVAC, legal, real estate, salons, plumbing, roofing, pest control, landscaping, and more) because each vertical has specific keyword clusters with commercial intent
• Build comparison and alternatives pages against every major competitor (Smith.ai, Ruby, Rosie, Dialzara, PATLive, AnswerConnect) because these target bottom-of-funnel buyers
• Create data-driven content (statistics, cost guides) because these attract citations from AI platforms and other websites
• Optimize every article for AI search with TL;DR blocks, entity embedding, and comprehensive schema markup
• Submit every page for instant indexing via Google Indexing API and IndexNow to accelerate crawling

Month 1: Building the Content Foundation

The agent's first priority was building a comprehensive content base. In 30 days, it created and published 101 articles across 15 content categories:

Every article was created with proper schema markup (Article, FAQ, BreadcrumbList), internal linking to related pages, entity optimization, and AI-ready TL;DR blocks. The agent handled all of this automatically as part of its content creation pipeline.

An agency producing this volume at 4 to 8 articles per month would take 12 to 25 months to match the same output.

Month 2: Technical Optimization and Indexing

With the content foundation in place, the agent focused on technical optimization and getting Google to discover and index the pages.

Technical fixes deployed

• Created robots.txt for proper crawl directives
• Generated and submitted a sitemap with 211 URLs
• Resolved a canonical issue: the site had conflicting www and non-www canonicals for 3 months (December 2025 to February 2026). The agent identified the problem, implemented 301 redirects, set canonical tags to www, and resubmitted affected URLs.
• Deployed IndexNow for instant notifications to Bing, DuckDuckGo, Yandex, and Ecosia on every publish
• Submitted 88 unindexed URLs via the Google Indexing API
• Optimized meta titles for 6 high-impression, low-CTR pages
• Added blog lead capture (inline CTAs and sticky bar)

Indexing progress

Google indexed 20 out of 101 pages within 90 days. This is typical for a new domain. The agent continued to resubmit URLs via the Indexing API and monitor crawl status in Google Search Console.

Month 3: Results and Growth

By March 2026, 90 days into the experiment, here are the actual numbers from Google Search Console and GA4:

The impression growth trajectory was strong: 91% increase from the previous measurement snapshot. Clicks were up 58.3% week over week. These are early-stage numbers for a new domain, and SEO compounds over time. The important signal is the direction, not the absolute numbers.

Perhaps the most interesting finding: AI referral traffic was already a meaningful channel. 15 sessions per week from ChatGPT, Perplexity, Copilot, and Gemini. One signup was directly attributed to ChatGPT. AI referral visitors spent 3x longer on the site (3:44 average session duration vs 1:14 for Google organic visitors).

What Worked Best

Comparison articles drove the most impressions

Articles comparing AI receptionists (best-ai-receptionist, best-virtual-receptionist, best-ai-answering-service) generated the highest impression volume. Three comparison pages were among the first 20 to get indexed by Google. These target bottom-of-funnel buyers who are actively evaluating solutions.

Data-driven content attracted AI citations

The "missed business calls statistics" article was cited by 3 different AI platforms (ChatGPT, Copilot, Perplexity). Data-heavy, statistic-rich content with clear sourcing is exactly what AI models prefer to cite. This is the single most effective format for AI visibility.

Schema markup accelerated indexing

Every article included Article, FAQ, and BreadcrumbList schema markup. Pages with comprehensive schema were indexed faster than those with minimal markup. In cycle 11, we added SoftwareApplication schema to comparison pages after discovering that every competitor cited in AI Overviews had this schema.

Instant indexing via API made a measurable difference

Pages submitted via the Google Indexing API and IndexNow were discovered and crawled significantly faster than pages left for natural discovery. We submitted 43 URLs in one batch and saw indexing movement within days.

Internal linking created topical authority signals

The agent automatically built internal links between related articles: industry pages linked to the main "best AI receptionist" comparison, alternatives pages linked to head-to-head comparisons, and cost guides linked to comparison and industry pages. This created a web of topically connected content that Google interpreted as authority.

What We Would Do Differently

Fix canonical issues before launching content

The www vs non-www canonical confusion took 3 months to resolve. During that time, Google was splitting authority between two versions of the same pages. This delayed indexing and diluted ranking signals. Lesson: resolve all canonical, redirect, and domain configuration issues before publishing any content.

Submit for indexing immediately after every publish

Early in the project, we published batches of articles and submitted them for indexing days later. This created unnecessary delays. The agent now submits every URL via both Google Indexing API and IndexNow within minutes of publishing.

Prioritize fewer, higher-quality articles initially

Publishing 101 articles in the first month was ambitious. In hindsight, starting with 20 to 30 high-quality articles targeting the highest-impact keywords, then expanding, would have produced faster ranking signals with less total effort.

Start AI visibility optimization from day one

We added TL;DR blocks and entity optimization in later cycles. The articles that had these features from the start were cited by AI platforms sooner. Going forward, every article gets AI optimization as a standard part of the creation pipeline.

The Numbers: AI Agent vs Agency

The cost per article comparison is striking: $500 to $1,000 per article from an agency versus $8.88 per article from the AI agent. Both produce content that Google ranks on the same criteria. For the full pricing breakdown, see our complete SEO agency cost comparison.

Key Takeaways for Business Owners

1. SEO results take time regardless of method. Whether you use an agency or an AI agent, Google needs 6 to 12 months to fully evaluate and rank your content. The advantage of AI is not faster rankings. It is faster execution and lower cost.
2. Content volume matters. More quality content means more surface area for Google to rank. 101 articles give you 101 chances to appear in search results versus 12 to 24 from an agency in the same period.
3. AI search is a real channel now. ChatGPT, Perplexity, and Gemini are sending real traffic that converts. Optimizing for AI citations is no longer optional. It is a competitive advantage most businesses are ignoring.
4. Technical SEO foundations matter. Our canonical issue cost us weeks of ranking potential. Get your technical house in order before investing heavily in content.
5. The risk profile is fundamentally different. If your SEO strategy does not work, you have lost $900 with an AI agent versus $12,000+ with an agency. No contracts means you can adjust or stop at any time.

Frequently Asked Questions