More from @mattdiggityseo

Oct 28

Kevin Indig just tested 35,000 backlinks across 1,000 sites to see what AI search engines actually care about.

What he found breaks every rule we learned about link building.

Here’s what actually works in 2025 and beyond:

Finding #1: You need to hit a minimum authority level first

According to Indig’s research in Search Engine Journal, having some authority helps you show up in AI results.

But here’s the thing: small gains don’t matter much.

You need to reach a certain level before AI starts mentioning you consistently.

Think of it like a video game. You need to level up to a certain point before you unlock the next stage.

And here’s what really matters: getting links from DIFFERENT websites.

This isn’t exactly breaking news, but it’s worth repeating because too many people still miss it.

One link each from 10 websites beats 100 links all from the same 2 websites.

AI cares about how many different places are talking about you, not just how many total links you have.

Finding #2: “Nofollow” links work just as well

You know those links that are supposed to be “less valuable”? The ones marked “nofollow”?

According to Search Engine Journal’s analysis, AI treats them almost exactly like regular links.

The data:

• Regular links: 0.504 correlation score
• Nofollow links: 0.509 correlation score

ChatGPT and Gemini actually prefer nofollow links slightly MORE than regular links.

This is huge because nofollow links are easier to get.

Guest posts? Check. Social media mentions? Check. Forum discussions? Check.

If you’ve been skipping these because they’re “nofollow,” you’ve been leaving wins on the table.

Finding #3: Links from images work better than text links

This one surprised me.

Indig’s research found that backlinks from images (like when someone shares your infographic) actually perform BETTER than text links.

The numbers:

• Image links: 0.538 correlation score
• Text links: 0.472 correlation score

Perplexity and SearchGPT especially love image links.

What does this mean for you?

Create shareable visuals. Make infographics. Design charts from your data.

When other sites embed your images, those links carry serious weight with AI.

Read 6 tweets

Oct 8

After running thousands of SEO audits for clients…

I’ve perfected a 10-step process that consistently uncovers traffic wins worth millions.

Here’s the exact framework (steal it and run it on your site today):

Step 1: Crawl Setup That Actually Works

Download Screaming Frog (worth every penny for serious SEO).

Critical settings most people miss:

• Uncheck resource links to speed up crawls • Enable near duplicates detection • Connect Google Analytics 4 and Search Console APIs • Add PageSpeed Insights API for performance data

This setup gives you actionable data instead of overwhelming noise.

Step 2: Internal Linking Quick Wins

Export your crawl to Google Sheets and filter for two killer opportunities:

• Pages deeper than 3 clicks (poor crawl depth) • Pages with fewer than 5 internal links

These pages are invisible to Google. Simple internal linking fixes can push them from page 10 to page 1.

Step 3: Speed Wins That Matter

Sort by performance score in your crawl data.

Anything below 80 gets flagged for optimization.

Why this matters: AI crawlers give you less than 3 seconds before bouncing.

Slow sites get skipped by both Google and ChatGPT.

Read 11 tweets

Oct 1

I’ve been doing SEO for 16 years, and I still see people making the same basic mistakes with on-page SEO.

Most think they need 500 backlinks to rank. Wrong.

You can rank with fewer links if you nail your on-page structure.

Here’s how to build sites that Google actually understands:

1. Start with your supporting pages first.

Stop thinking about your money pages first.

Start with your supporting pages. (Kyle Roof calls this the reverse silo approach)

Your blog posts and service pages do the heavy lifting, not your homepage.

Build from the ground up, not top down.

Your supporting pages should only exist for one main purpose: to send authority to your money page.

Don’t link these pages to multiple service areas or random other pages.

They link to each other and funnel everything to the page that makes money.

2. URL structure matters way more than you think.

The more folders you put between your target page and your domain, the harder Google has to work.

Skip the /blog/ and /page/ nonsense.

Put important pages closer to your root domain.

3. Focus on contextual terms, not word count

Word count isn’t about hitting some magic number.

It’s about including enough contextual terms for Google to understand what your page is about.

Use content optimization tools (like Surfer) to check if your content has the right mix of related keywords, not just the target term.

Most AI content fails because it doesn’t include the contextual terms Google needs.

LLMs write content that looks good to humans, but is missing the semantic signals Google looks for.

Always check your AI content with proper on-page analysis.

Read 8 tweets

Sep 2

You don’t need to chase every new AISEO “hack”.

The 80/20 of AI search comes from doubling down on SEO fundamentals.

Here’s what actually works for AI visibility in 2025:

1. Get mentioned across the web

AI picks up brand mentions, with or without backlinks. Focus on:

• Industry “best of” lists
• Customer reviews and case studies
• PR coverage and media mentions

2. Create content AI loves to cite

Ahrefs’ analysis shows these content types dominate AI citations:

• “Best” content: 7.06% of AI traffic (comprehensive buying guides work well)
• How-to guides: 6.35% of AI traffic (step-by-step tutorials with clear outcomes)
• “Top” lists: 5.5% of AI traffic (data-backed rankings and comparisons)
• Product comparisons: 4.88% of AI traffic (“vs” content addressing specific use cases)

3. Structure for AI comprehension

Make it dead simple for AI to parse and cite your content:

• Use clear H1, H2, H3 headings that mirror how people ask questions
• Add bullet points and short paragraphs (avoid text walls)
• Include specific facts and figures AI can cite directly
• Answer common questions explicitly with factual statements
• Add schema markup for FAQs and product information

Read 7 tweets

Aug 14

AI search doesn’t play by Google’s rules.

But it’s way easier to game.

Unlike traditional SEO, you can show up in AI search results in days… if you know what you’re doing.

The kicker?

Customers from AI recs are 4.4x more valuable than your average Google click (source: SEMRush).

Here’s how I’d approach AI SEO if I were starting fresh today: 🧵

1. Optimize for recognition

AI SEO is all about training LLMs to see you as “the best [fill in your niche].”

But these tools don’t crawl the web the way Google does.

They scrape for signals. Mentions. Testimonials. Listicles.

The good news? They’re way easier to influence.

You don’t need keyword stuffing. You need context.

LLMs think more like humans.

They don’t care if you used “best HVAC repair Chicago” 14 times (believe it or not, people still do this).

They rely on NLP to see if your content sounds like someone solving a real problem.

So talk like a human.

Do thorough research on your ideal customer and write to solve their biggest pains.

2. Don’t write as your business. Write ABOUT your business.

LLMs don’t want to hear from you. They want to hear about you.

So ditch the “we’re passionate about serving our community” fluff.

Instead, create short third-person blog posts that answer queries like:

“Who’s the best [service] in [location]?”

Make it sound like an unbiased expert wrote it. Keep it tight. Include a review, case study, or specific feature.

AI loves this stuff. Especially when it’s structured like a direct answer.

Read 9 tweets