From Zero to Cited: How a Business Appeared in AI Answers in 90 Days

In January 2026, we ran an experiment. We asked ChatGPT, Perplexity, and Google Gemini the same question: "Which AI automation agency in the Netherlands is suitable for SMBs?"

Mindsora appeared nowhere. Not in the top 5. Not in the top 10. Not at all.

That was confronting. We sell AI discoverability to clients, but our own website was invisible to the very tools we reference. It was time to take our own medicine.

Ninety days later, Mindsora appeared in AI answers for relevant queries. Not occasionally — consistently. This is the exact story of how that happened.

Day 0: The Baseline

We started with a thorough audit of our own website. The results were sobering.

Area	Score (1-10)	Status
llms.txt	0	Not present
Schema markup	3	Only basic Organization
Open Graph	4	Present but incomplete
Semantic HTML	7	Good, but not perfect
Loading speed	8	< 2 seconds
robots.txt	6	No blocks, but not optimized
External authority	2	Barely any mentions
Review signals	1	No Google reviews

Total score: 3.9 out of 10.

We had a fast, well-built website. But AI agents couldn't understand what we did, for whom, and why they should recommend us.

Week 1-2: The Technical Foundation

Creating llms.txt

First, we created an llms.txt file. Not just a list — a structured document that tells AI agents exactly:

• Who we are
• What we do
• Which sectors we serve
• What the primary actions are on our website
• What we explicitly don't do (no implementation without intake call)

This took two hours. The effect turned out to be disproportionately large.

Expanding Schema.org markup

We went from basic Organization markup to a complete set:

• Organization with all business details
• ServiceCatalog with every service, price, and description
• FAQ schema on the homepage and services page
• BreadcrumbList on all pages

Time investment: four hours. No visual impact on the website — purely semantic enrichment.

Perfecting Open Graph

Every page received a unique og:title, og:description, and og:image. We wrote the descriptions as if they'd be read aloud by an AI agent to a user. Short, concrete, with a clear value proposition.

Week 3-4: Content Optimization

Blog content with AI agents in mind

We published four blog posts, specifically written for the questions AI agents answer:

• "AI automation for accountants" (our core market)
• "What does AI automation cost for SMBs?"
• "GoHighLevel for Dutch businesses"
• "AI automation for real estate agents"

Each article: 800-1200 words, concrete numbers, comparison tables, clear structure with H2/H3 headings. Exactly the format that AI agents can parse and cite effectively.

Expanding FAQ sections

We added FAQ sections to the homepage and services page. Not decorative — strategic. Every question was a real client question we'd heard. Every answer was concrete, with numbers.

AI agents use FAQ sections as direct sources for answers. If someone asks "What does AI automation cost?" and your FAQ answers that with specific amounts, you get cited.

Week 5-8: Authority Building

This was the hardest part. Technical optimization you can do in a weekend. Authority takes months to build.

Collecting Google Reviews

We asked our first client and two project partners for a Google review. Within three weeks, we had five reviews with an average of 4.8 stars.

Not spectacular in volume. But AI agents value the presence of reviews disproportionately — it signals that the business actually exists and has real clients.

LinkedIn thought leadership

We published three times a week on LinkedIn. No generic AI content, but concrete results and insights:

• "Our client saves 12 hours per week with these three automations"
• "Why 90% of Dutch SMB websites are invisible to AI"
• "The technical audit every business owner should do"

LinkedIn posts get indexed by AI models. They build a pattern: this company consistently publishes about this topic with concrete expertise.

External mentions

We got our profile created on three platforms:

• Clutch.co (business services directory)
• A guest blog on a Dutch marketing platform
• Listing in an industry directory for automation service providers

Every external source linking to Mindsora strengthens the signal to AI agents: this is a real business with relevant expertise.

Week 9-12: Measuring and Adjusting

The first citation

In week 7, Mindsora appeared for the first time in a Perplexity answer to the query "AI automation agency Netherlands SMB." We weren't mentioned first, but we were in the answer.

In week 9, we were cited by ChatGPT for the query "Which company helps accountants with AI automation?" The answer mentioned us as an option, with our service description taken almost verbatim from our llms.txt.

The numbers after 90 days

Metric	Day 0	Day 90	Change
AI discoverability score	3.9/10	7.8/10	+100%
ChatGPT citations (per month)	0	8-12	new
Perplexity citations (per month)	0	5-8	new
Google reviews	0	7	+7
External mentions	2	9	+350%
Organic website visitors	340/mo	580/mo	+71%
Leads via website	4/mo	9/mo	+125%

The most striking metric: leads. We went from four to nine leads per month. Not all directly attributable to AI citations, but the correlation is strong. Two leads explicitly mentioned they found us via ChatGPT.

What Made the Biggest Difference?

If we had to pick the three actions with the most impact:

1. llms.txt (Week 1) Disproportionately effective. The first AI citations referenced text almost verbatim from our llms.txt. This file is the most direct communication channel with AI agents.

2. FAQ schema with concrete answers (Week 3-4) AI agents look for answers to specific questions. FAQ sections with schema markup deliver those answers on a silver platter.

3. Consistent LinkedIn publishing (Week 5-12) Not a viral post. Not a brilliant article. Just three times a week, every week, sharing concrete expertise. AI models recognize patterns of consistent authority.

What It Wasn't

Let's be honest about what didn't work or wasn't necessary:

• Paid advertising had no impact on AI discoverability
• Backlink outreach to hundreds of sites wasn't necessary — quality over quantity
• Publishing daily wasn't better than three times a week
• Technical optimization alone wasn't enough without content and authority

The Lesson for Your Business

AI discoverability isn't a magic trick. It's a systematic process:

1. Lay the technical foundation (Week 1-2): llms.txt, schema markup, Open Graph
2. Optimize content (Week 3-4): FAQ, blog posts, concrete answers
3. Build authority (Week 5-12): Reviews, publications, external mentions
4. Measure and adjust (ongoing): Which queries do you appear in? Where not?

Ninety days. No secrets. No shortcuts. Just discipline and the right technical knowledge.

---

Want to know where your website stands on AI discoverability? Request a free AI Readiness Scan — we'll show you your score and which steps deliver the most value.