I Went From 0 to 800+ AI Citations in 4 Months. Here’s What I Actually Did.
Four months ago, the app I was working on had essentially zero visibility across the AI platforms I was tracking.
Ask ChatGPT about the category. We weren’t there.
Perplexity? Nothing.
Gemini? Nothing.
Google AI Mode? Nothing.
Grok? Nothing.
Then I started treating AI visibility as a distribution problem rather than a traditional backlink problem.
Within roughly four months, my tracking showed 800+ citations and mentions across ChatGPT, Perplexity, Gemini, Google AI Mode, and Grok.
This post is the exact process I followed, what appeared to move the needle, what underperformed, and what I’d change if I were launching another product from scratch.
One caveat before we start: I can measure the correlation between this campaign and growing AI visibility, but I can’t prove that one specific Medium post caused one specific ChatGPT answer. AI retrieval systems change constantly. So I treat this as a repeatable experiment, not a magic formula.
TL;Dr
My distribution over the first four months looked roughly like this
• 10–15 strong research-led articles on the main website
• 50 unique Medium articles
• 100 LinkedIn posts
• 40 Facebook/community posts
• Relevant Reddit/community participation and mentions
• Consistent category and competitor positioning
• AI visibility tracking from the beginning
The strongest-performing content types were specific FAQs and comparison/listicle content.
The weakest? Generic promotional posts
The Mental Model: AI Visibility Is a Distribution Problem
I started with the wrong mental model.
Like most founders who have done SEO, I was thinking:
Create pages → build links → improve rankings → wait for traffic.
But the question I actually needed to answer was:
“If someone asks an AI assistant, ‘What’s a good product for X?’, does the web contain enough credible context for that AI system to know my product belongs in the answer?”
That changed the strategy.
Instead of focusing only on my own site, I wanted the product to appear naturally in the places where a person researching the category would expect to find it:
• Reviews
• Comparisons
• Alternatives
• Listicles
• FAQs
• Founder content
• Community discussions
• Third-party mentions
The goal wasn’t to repeat the brand name everywhere.
The goal was to create consistent context.
If multiple independent-looking sources describe a product as “an invoicing tool for freelancers,” compare it with other invoicing tools, and answer questions about when it is useful, the product becomes easier to understand in relation to the category.
What I Actually Published
1. Specific FAQ Content
This was the most interesting format in my experiment.
Instead of chasing broad terms such as:
“best CRM”
I created content around narrower questions:
“What’s the best CRM for a three-person recruiting agency?”
“What’s the easiest invoicing app for a freelancer?”
“What’s a good alternative to [competitor] for small teams?”
“What software is best for [specific workflow]?”
Why did this appear to work?
My theory is that the structure closely matches how people use AI assistants: question in, answer out
A focused FAQ gives an answer engine a clear problem, a clear category, a set of options, and a useful conclusion.
2. Listicles and Comparison Posts
The next-best format was the classic:
“Top 10 [tools] for [use case]”
But I didn’t write these as disguised ads.
I included established competitors that genuinely belonged in the article, then positioned the newer product where it made sense.
For example:
1. Established tool
2. Established tool
3. Newer product
4. Established alternative
5. Specialist option
I deliberately avoided putting an unknown product at #1 with no justification.
Humans can smell that instantly, and it makes the whole page feel like affiliate spam.
A better article explains where each tool fits.
3. Category Roundups
Roundups such as:
“Best [category] apps in 2026”
also worked, especially when the article ended with a real verdict by use case.
For example
Best for freelancers: Product A
Best for larger teams: Product B
Best for beginners: Product C
Best for a specific niche: Product D
That’s more useful than pretending one product is universally “the best.”
The Distribution Numbers
This is the part that turned the strategy from content marketing into a production system.
In under four months, I pushed roughly:
50 Medium articles
Every article was unique. Different topic, angle, wording, and structure.
100 LinkedIn posts
This was the highest-volume channel in the campaign. I mixed list-style posts, founder observations, category education, and problem/solution content.
40 Facebook and community posts
These produced less obvious citation impact than LinkedIn in my tracking, but they added distribution diversity.
Reddit and community mentions
I looked for real discussions already happening around the problem and category.
The important part: don’t manufacture fake conversations, coordinate fake comments, or pretend unrelated accounts “discovered” your product.
Participate where you actually have something useful to add. Reddit communities are especially good at detecting marketing behavior.
10–15 foundation articles on the product website
I consider these essential.
Distributed content works much better when it points back to a site that clearly explains:
• What the product does
• Who it’s for
• How it works
• What it costs
• How it compares
• Which problems it solves
A weak website gives all of your third-party distribution nowhere credible to lead.
What I Learned From Each Channel
If I were doing this again, LinkedIn would remain a major part of the strategy.
But I’d lean even harder into founder-led posts rather than generic company content.
For example:
“I analyzed 50 prompts buyers use when researching [category]. Here are the five questions that kept appearing.”
That creates something people may actually want to read while still establishing category context.
Medium
Medium worked well as a place to publish longer, focused articles quickly.
The mistake would be posting the same article 50 times with minor rewrites.
I wanted topic diversity, not duplicate pages.
Instead of producing 10 versions of “Best Accounting Software,” I’d use angles like:
• Best accounting software for freelancers
• QuickBooks alternatives for consultants
• How solo founders should track expenses
• Best bookkeeping tools for contractors
• Product A vs Product B
• What should accounting software cost?
Same market. Different intent.
Reddit is not a content syndication platform.
It is a community.
The best approach is to find relevant questions and contribute useful experience.
If your own product genuinely solves the problem, say so transparently where the subreddit rules allow it.
A sentence like:
“I built X after running into this exact problem, so take my recommendation with that bias…”
is more credible than trying to manufacture social proof.
What Didn’t Work
Generic promotion performed poorly.
Posts whose entire message was basically:
“Here’s our product. It’s great. Check it out.”
had almost no measurable citation impact compared with FAQs and comparisons.
I also wouldn’t repeat exact-match messaging across dozens of sites.
The product/category relationship should stay consistent, but the language should vary naturally.
And I would never pay for fake discussion or fake endorsements.
The objective is to make the product easier to discover and understand, not to fabricate consensus.
How I Outsourced the Work
There was no realistic way I was going to personally write nearly 200 pieces of content while also building the product.
So I turned it into a briefing and production system.
The brief I gave writers was intentionally small:
1. One specific title
2. The product URL
3. At least three legitimate competitors or alternatives to consider
4. A target length, usually around 800–1,200 words for articles
That was usually enough.
I wanted different writers to sound like different writers.
Over-templating everything would defeat the point.
One brief. One angle. One unique piece.
If You’d Rather Buy the Distribution Than Coordinate It Yourself
Once you understand the strategy, execution becomes a production problem: research topics, source writers, publish unique pieces, place community mentions, coordinate Medium and LinkedIn distribution, and track what happens afterward.
For founders who don’t want to build that operation themselves, LLMentioned packages this kind of AI-visibility distribution, including Reddit/community mentions plus LinkedIn and Medium content.
LLMentioned AI Visibility: https://www.1stpage.agency/llmentioned-ai-visibility/
I’d still keep strategy close to the founder
No external service knows your customer, positioning, competitive advantages, and product truth as well as you do.
Outsource production if it makes sense. Don’t outsource the thinking.
Disclosure note before publishing: if you own, work with, receive commission from, or otherwise have a commercial relationship with LLMentioned / 1stPage.agency, state that relationship clearly in this section. Indie Hackers readers are generally much more receptive to transparent promotion than hidden promotion.
How I Tracked the Results
I tracked AI visibility rather than relying only on referral traffic.
The questions I cared about were:
• Are AI platforms mentioning the product?
• Which prompts trigger mentions?
• Which sources are being cited?
• Which competitors appear more often?
• Is visibility trending up over several weeks?
• Are AI referrals turning into signups or revenue?
I also learned not to obsess over daily fluctuations.
At one point, ChatGPT visibility climbed much higher and then settled lower.
That didn’t automatically mean the strategy had stopped working.
AI search results change. Retrieval changes. Sources rotate. Prompts vary.
The multi-week trend is more useful than yesterday’s number.
What Appeared to Drive the Most Citations
Based on my own tracking, the rough order was:
1. Specific FAQs
2. Comparison/listicle content
3. Category roundups with a clear conclusion
4. Useful founder/category posts
5. Plain promotional posts
The common denominator was usefulness plus context.
The better the piece answered a real question, the more valuable it seemed to be.
If I Were Launching a New SaaS Tomorrow
I’d repeat the strategy in five phases.
Phase 1: Build the foundation
Publish 10–15 genuinely useful pages on the product site.
Cover the main problem, use cases, comparisons, alternatives, FAQs, pricing questions, and terminology buyers actually use.
Phase 2: Publish 30–50 third-party articles
Start with Medium and other relevant publishing opportunities.
Mix FAQs, comparisons, alternatives, roundups, and how-to content.
Every piece should target a different useful angle.
Phase 3: Build LinkedIn distribution
Publish consistently for six to eight weeks.
Mix founder lessons, research, problem breakdowns, product-category insights, and comparisons.
Don’t make every post a pitch.
Phase 4: Participate in communities
Find where people are already discussing the problem.
Reddit, Indie Hackers, niche forums, Facebook groups, Slack communities, and other relevant spaces can all matter depending on the market.
Contribute first. Promote only when relevant and permitted.
Phase 5: Measure from day one
Establish a baseline before you start.
Then track AI mentions, citations, prompts, competitors, referral traffic, signups, and—ultimately—revenue.
A Smaller Version for Indie Hackers Who Don’t Have a 190-Post Budget
You do not need to start with 190 pieces.
If I were a solo founder testing this with limited time and budget, my first 30 days would look like:
Week 1:
Publish three strong FAQ/comparison pages on the main site.
Week 2:
Publish five unique Medium articles targeting narrow customer questions.
Week 3:
Publish three to five genuinely useful LinkedIn posts and participate in five relevant community discussions.
Week 4:
Check which prompts and pages are showing signs of visibility, then double down on the angles that are working.
The point is to prove the model before scaling the production.
Final Takeaway
The biggest change for me was stopping thinking about AI visibility as “SEO, but for ChatGPT.”
It’s closer to building a web-wide body of evidence around your product:
What is it?
Who is it for?
What problem does it solve?
Which category does it belong to?
How does it compare with known alternatives?
When should somebody choose it?
Your own website should answer those questions.
Then useful third-party content and genuine community participation should reinforce them.
That’s the system I used to go from essentially zero tracked AI visibility to 800+ citations and mentions in roughly four months.
I wouldn’t copy the exact numbers blindly.
I’d copy the process:
Build a strong source of truth.
Create useful category context.
Distribute it widely.
Keep the content genuinely unique.
Participate in communities transparently.
Track what the AI platforms actually surface.
Then scale the formats that work.
That’s what I’d do again.
