Search is Chaos, But You Don't Have to Be :)

And once again, search is changing. Just as we’re still wrapping our heads around AI overviews, their impact, and their quirky citation habits, Google drops AI Mode in the U.S.

If you feel everything’s in flux — you’re not alone. But I think the real issue isn’t the change itself — it’s that we’re all waiting for the dust to settle… and it probably never will 😄

Also, we know that most of Google’s ranking signals are still “handcrafted”—crafted by engineers and algorithm designers — with RankBrain and DeepRank being the main ones powered by AI. This lets Google track, debug, and hopefully “fix” problems… although sometimes you do wonder if they really do 😏

In any case, in this blog I’m going to share what I think you should be focusing on as an SEO right now — outside our classic tactics — while we navigate these wild waters. I’ve gotten enough SEO entertainment this week to keep me thinking for months, so grab some popcorn 🍿 and let’s go!

TL;DR

• You can be penalized by HCU, and lost tons of traffic and still get your "unhelpful content" used in ai overviews. No, that's not contradiction at all from Google. Do they ever 😄
• AI overviews and AI mode reporting is not in a separated tab in GSC and it's very disappointing.
• Google is doing massive cleanup to it's database right now. It's like they're getting ready for an upcoming bigger change. I guess that explains all the recent bugs in search.
• Don't look at organic traffic metrics in isolation from conversion numbers (unless you're a publisher, that's a whole different story).
• Writing for ai search era is slightly different. You need to focus on the paragraph level, not the page level.
• It's a really good idea to allow different LLMs to access your site search in robots.txt
• and more interesting tips + free tool from Mike King below.

The Good enough to use in ai overviews, not good enough as helpful content

This tweet from Luc Wiesman, the founder of DMarge (Australia’s longest-running men’s lifestyle website), shows a screenshot where their content — in this case, even Luc’s arm — is used in AI overviews without giving DMarge any credit. Despite the website's popularity, credibility, and high‐quality content, DMarge has lost a ton of website traffic after the helpful content updates 🤷‍♀️

Tip: I recommend you spend some time analyzing the DMarge website. There's a lot to learn from it's performance history.

The report that's not really a report

I've been very vocal about the need for ai overviews and ai mode data in GSC and guess what? google just updated their documentation to include how data for ai mode is included in GSC.

But here's the twist, ai mode, same as ai overviews will not be reported on separately but rather part of the generic performance report 😏

Tip: The only way to go more granular right now is to combine keyword level data from third party tools, with data from GSC using VLookups or build your own SEO tools with Python.

The schema that doesn't matter, matters

Google said they will stop supporting some specific schema types to “simplify search.” If Google is getting rid of some schemas, doesn’t that mean the schemas they still support are still useful?

Jason Barnard, the knowledge graph dude also noticed some massive shakeups... Google is doing some deep cleaning!

Is this cleanup a subtle sign that google is getting ready for some big feature drop? or full rollout of ai mode? that would also explain all the recent bugs like the drop in Google indexing late May 2025.

Tip: Anyhow, my main takeaway from this is it's time we refresh our own schema codes as well, and use them if you're not already!

Traffic from AI, does it convert better?

To be or not to be, we have 2 interesting case studies on both sides of this argument.

A study by Dan Taylor came to the conclusion that "In most sectors, Organic KECVR (key event conversion rate) is higher than LLM KECVR, indicating that users arriving through organic search generally have a higher likelihood of engaging (converting) on the website.

However, remember Hubspot and their blog traffic drop situation? According to Mike King, the community was so focused on analyzing the relation between Google updates and the traffic drop, and totally ignored the impact of ai overviews!

And despite this drop.... their revenue is still climbing!!! FYI Organic traffic is THEIR ONLY REVENUE PRODUCING CHANNEL. What does this mean?

It means the same thing revealed by a recent Ahrefs study "AI Search Traffic Convert Better Than Traditional Search" at least for Ahrefs too!

It seems that traffic coming from organic after ai overviews may actually be converting better, at least for some websites/businesses. Your business may be one of those.

Tip: don't look at traffic numbers in isolation from your conversion rates and profitability from organic search. Always look at traffic and conversions hand in hand, unless you're a publisher not selling anything.... ahhhh that's a story for another day.

Query Fan Out

The buzzword circulating around, and we LOVE a new buzzword in SEO. Anyhow, if you're done debating is it SEO or GEO and looking to live the actual moment, maybe you should try out this tool built by Mike King that will help you understand and test how query fan out technique works.

Tip: obviously test drive the above tool!

Writing content for the new search

This section is built on Mike King's SMX deck and SEO Queen Aleyda Solis recommendations. I added some notes myself and made it easier to read. This could've been it's own blog, but here we are!

How to write content that's optimized for ai search?

• Optimize for chunk-level retrieval: ai search looks at chunk (paragraphs) on your page vs. the entire page. This means when someone searches for a query, ai search looks for the most relevant chunks not pages. So make sure every chunk on your page is specific, with a clear purpose. This actually aligns with google writing practices where the first sentence of a paragraph tells what a paragraph is about.
• Build your content around semantic tiples (subject-predicate-object relationships). You know who built a tool for this already? Steve Toth, check it out here.
• Use data in your sentences as LLMs often prioritize precise data points and statements of facts. Include verifiable data.
• Improve readability. Make sure your content is easily parsed and understood. You can run it by any LLM to for a readability check.
• Be specific and unique. It's vague I know, reminds me of the "create helpful content" tip by Google.
• Use topical clustering, consistent terminology and google internal linking.
• The search universe does not revolve around Google, optimize on other platforms like reddit, create a microsite, convert your content to YouTube videos and thus diversify your content formats. Organic traffic is going down, deal with it.
• Make sure your content is accessible for LLMs. This aligns pretty much with an interesting tip shared by Pedro Diaz a while ago. Pedro recommends allowing LLMs to access site search pages!

If you're wondering how, here's an example he attached with his post:

So,

Search is changing, things look messy right now because they are, the dust has not settled and I'm not sure if it ever will. But the way to go through this time is to keep learning, testing, stay open to new ideas and approaches and please avoid getting sucked into useless terminology wars (I'm a little guilty with that too).

I genuinely enjoyed writing this blog, I hope you did too. Thanks for reading so far and see you next newsletter.

To support me please consider one of the following:

• Sign up for my newsletter if you're not already.
• Share the newsletter and invite your friends to signup. Help me reach 2k signups by end of 2025 please 🙂
• Provide feedback on how I can make this newsletter better!!!
• Buy me coffee.
• If you're an SEO tool or an SEO service provider, consider sponsoring my newsletter. I'm also open to other partnership ideas as well.

Disclaimer: LLMs were used to assist in wording and phrasing this blog.