We publish a lot of SEO content (I am founder of one of those SEO automation tools, LLaMaRush), but this time we decided to write an article specifically with AI search in mind.

We published it 3 days ago.

Today:

• It's showing up in Google's AI Overview.
• It's ranking #2 on Google for the target keyword.

Obviously, it's way too early to say why. It could be freshness, low competition, topical authority, the way we structured the content, or a combination of everything.

Still, I wasn't expecting this kind of movement after only 3 days.

Going forward I'm planning to document these experiments instead of treating GEO like some black box. It'll be interesting to see which tactics actually hold up over time and which ones are just coincidence.

My plan is, if I can successfully reproduce these results for few more blogs, I will change my content engine accordingly and add AEO and GEO content to my tool as well.

Spent the last 4 months trying to sell LLaMaRush through demo calls.

In that entire time I got 3 customers.

I kept thinking people needed to understand the product first, so everyone had to book a call before they could use it.

3 days ago I got rid of the demo requirement and added a free trial instead.

Since then:

• 8 people have signed up
• 1 already upgraded to the highest plan

Nothing else changed.

I guess I massively underestimated how much people hate scheduling calls. If someone already has intent, they usually just want to try the product and make up their own mind.

Wish I'd done this months ago.

One exercise completely changed how I think about AI search visibility.

Most SEOs approach Perplexity the same way they approach Google.

They search:

• best CRM software
• email marketing platform
• project management tool

And then they check whether their client appears.

I think that's the wrong approach.

Your customers aren't opening Perplexity and typing keyword fragments.

They're having conversations.

They're asking:

"We're a 20-person SaaS company managing leads in spreadsheets. We need LinkedIn integration and can't spend more than $100/month. What CRM should we use?"

Or:

"We currently use WhatsApp and Excel to manage deliveries. Is there a better system that doesn't require hiring an IT team?"

Those queries produce completely different answers than traditional SEO keywords.

Over the last few weeks I've been manually testing industries and documenting citations, recommendations, and source patterns.

A few observations stood out:

1. The competition is bigger than websites

When I first started checking citations, I expected to find competing company websites.

Instead I found:

• Reddit threads
• YouTube transcripts
• Documentation pages
• Industry forums
• Review platforms
• News articles
• Product comparison sites

Sometimes the source influencing a recommendation wasn't a competitor's homepage at all.

It was a Reddit discussion from months ago.

Or a detailed comparison article.

Or a review page.

If you're only tracking SERP competitors, you're missing a large part of the ecosystem AI systems actually use.

2. Direct answers outperform beautiful introductions

This one surprised me.

Many websites still follow the traditional content formula:

Long introduction → background → context → answer.

AI systems seem to prefer:

Answer → explanation → supporting details.

For example:

"What is Perplexity SEO?"

Article A:

"Artificial intelligence has transformed information retrieval..."

Article B:

"Perplexity SEO is the practice of making content easier for AI systems to extract, verify, and cite."

Which answer is easier for an AI system to use?

The difference becomes obvious once you start reading citations closely.

3. Being recommended and being cited are different things

A lot of people only look for citations.

I think recommendations matter more.

I've seen cases where a company isn't directly cited but is repeatedly recommended.

I've also seen companies cited frequently but rarely recommended.

Those are different visibility layers.

One measures source usage.

The other measures commercial influence.

4. Trust signals appear everywhere

Many discussions focus exclusively on content.

But when you inspect sources, you keep finding:

• Reviews
• Third-party mentions
• Expert authors
• Industry publications
• Documentation
• Community discussions

It feels less like traditional ranking and more like building a web of evidence that your company is credible.

The experiment I'd recommend

Open Perplexity.

Forget keywords.

Write down 10 actual customer questions.

Not search terms.

Questions.

Run every query.

For each answer record:

• Which brands were recommended?
• Which domains were cited?
• Which sources appeared repeatedly?
• Did Reddit appear?
• Did review sites appear?
• Did documentation appear?

After doing this, you'll probably learn more about AI visibility in your niche than from reading 20 GEO blog posts.

Because you'll stop guessing and start seeing where the model is actually getting information.

Curious what everyone else is finding.

What has moved the needle most for you:

• Better content structure?
• Off-site mentions?
• Reviews?
• PR?
• Community discussions?

Or are we all still collectively reverse-engineering this thing?

I built ToolsForImage, an open-source image toolkit for compression, resizing, format conversion, cropping, watermarking, blur tools, and batch image workflows.

It’s privacy-focused: processing happens in the browser, so images don’t need to be sent to a backend for processing.

Live demo:
https://toolsforimage.com/

Source:
https://github.com/mlbrothers/ToolsForImage-OSS

I’d love feedback from self-hosters

Recently started working with a founder of ChipVersus about a month ago, I am working on SEO for him part-time.

When I joined, Google had indexed only 1 page from the whole site. Most pages had extremely thin content with almost no useful information.

Over the last month, we:

• Rewrote/rebuilt almost every important page
• Added substantial unique content
• Improved structure/internal linking
• Cleaned up technical SEO issues
• Resubmitted pages in GSC

Now the weird part:
It’s been ~20 days since resubmission, but most pages are still stuck in “Discovered / Crawled - currently not indexed” type states and indexing barely moves.

Domain is not new, but previous content quality was poor.

Wanted to ask:

• Is this normal after a large-scale content rebuild?
• Does Google take longer to trust sites that previously had thin content?
• Any specific things you’d check in this situation beyond standard technical SEO?

Would appreciate insights from anyone who has recovered sites from thin-content situations before.

Just a note: Site has almost no backlinks, so will directory submit help here?