Mani

Director at ➡ See your local SEO performance in one place

[@levelsio @levelsio]

Airbnb deletes negative reviews so everything is always 4.7

Google Maps removes negative reviews in many countries as defamation so everything is always a 4.7

Booking sites want you to book so will also happily remove negative reviews for their hotels so everything is always a 4.7

I made a site called https://t.co/4p4dzTDVN6 (or https://t.co/A1nzVE5AlD whatever you prefer, I have both!)

It gets ratings from all booking platforms, normalizes them to show the actual score (eg a booking site where most hotels are always between 4 and 4.7 means you can normalize that to a 0 and 5)

It also uses AI vision models to rate hotels based on how they look

And it browses the internet for actual experiences (think blogs and Reddit posts) to get the real opinion about them outside reviews

And it's starting to work, I just stayed in this beautiful boutique apartment hotel called Woo Suites in Athens for $400/night. Clean, nice minimalist interior, working cold AC, central, excellent staff, great healthy breakfast, and we found it through Hotelist!

It's free, I don't make any money on it (actually I lose thousands per month on it to collect data via APIs and web scrapers) and there's no affiliate money bs like other booking sites (they will increase ranking of hotels if they pay sites a higher commission!)

https://t.co/4p4dzTDVN6

Tell your friends please!!!!!! 🙏

[Shauvik Kumar @shauvikkumar]

👀 Today's Google Diff...

Google's canonicalization troubleshooting docs got a significant update!

New section on "technical canonicalization issues" highlights server misconfigurations & missing localization.

They're emphasizing technical signals over just content quality. Use URL Inspection to verify Google's chosen canonical & check for 'hreflang' or CMS errors.

#SEO #TechnicalSEO

[Mani @manibharij]

What browser extensions / tools are you using for researching with LLMs?

I’m looking for something where I can highlight a specific section of an article, paper, etc. and ask questions about that bit in context (rather than starting a new prompt and feeding the whole context back in).

Feels like there must be some good tools for this. What are you using?

[Denis Loginoff ⚡️ @DenLoginoff]

I'm having a hard time finding AI use cases outside of coding, for myself 🙈

- automating emails? I already only receive a dozen a day that pass my filters, and for those I want to make a decision whom and how to respond to myself anyway
- generating AI images/videos? What would I do that for? Plus, I find them looking so fake and cheesy anyway
- writing/scheduling posts? I want to write in my own voice, why would I use AI for that? That would be so fake
- automating what in my life exactly? The few things I genuinely spend a lot of time on, like laundry and other menial chores, AI can't do

I just don't understand why ppl on X are so obsessed with it. It's become tiring to see all of that on my timeline. It seems like more often than not ppl use AI for the goal of.. using AI? 🤔

[Sky News @SkyNews]

Sky's @rowlsmanthorpe shows how AI can hack humans, using a reconstruction from the AI Security Institute. Following incidents reported by OpenAI and Anthropic, new findings suggest the emerging AI "cyber wave" could be more serious than previously thought.

[WordPress @WordPress]

Ouch, telling people not to update... this wasn't a bug in 7.1. Perhaps consider your usage of @wp_rocket. This is also part of the risk of using plugins outside the official plugin directory, which has many alternatives that seemed fine on the 7.1 release day.

[Lily Ray 😏 @lilyraynyc]

A few SEO/GEO data providers have shared this week that Reddit citations are collapsing in ChatGPT with the launch of 5.6, and after reading several articles and perspectives about this, I'd like to share my own version of what I think might be happening.

First off, I highly recommend reading @suganthan’s newest article on this (link in first comment), in which he explains how ChatGPT’s search behavior appears to have changed with 5.6 - including his finding that Reddit is still heavily retrieved *behind the scenes,* even when it isn’t cited, potentially serving as a validation or sentiment layer for the final answer.

Some of his newest findings:

⭐ Reddit showed up heavily in ChatGPT’s retrieval set (84 out of 221 results) but none of those Reddit URLs were cited in the final answer.

⭐ ChatGPT appeared to use Reddit *after selecting brands*, searching for user opinions about them rather than relying on Reddit to discover the brands in the first place. (!)

⭐ ChatGPT searched much further back on Reddit than on many other sources, sometimes using years of historical discussions (assuming his interpretation of the numbers in the query string is correct). I think that part is interesting - as Suganthan mentioned, if ChatGPT is looking for older conversations, those will be less likely to be influenced by the recent influx of spam.

What I think might be happening here, and this is *just my own theory* as always:

I think OpenAI, perhaps in collaboration with Reddit since they are partners, may be finding ways to continue using Reddit’s valuable information to help formulate responses, such as evaluating brands it has already identified (which Suganthan observed in his article) , while reducing the commercial incentive for brands to spam Reddit conversations.

If ChatGPT is still retrieving Reddit heavily but linking to it less often in responses, that suggests OpenAI may recognize that visible citations themselves create an incentive for marketers to manipulate Reddit. Reducing those citations could make Reddit spam feel like a less attractive AI search/marketing strategy, even if Reddit continues to play an important role behind the scenes.

I’ve already seen people posting things like, “Well, so much for Reddit. I guess we should focus on optimizing our own company website now.” 👀 (Not gonna lie, I had to laugh when I read that one.)

But perhaps that is exactly the intended outcome: reduce the incentive to manipulate Reddit, shift marketers’ priorities elsewhere, and potentially help address the rampant spam problem on the platform.

Reddit CEO Steve Huffman has described the company as being in an “arms race” to detect and block AI-generated activity, and Reddit itself recently said that “spam, bot activity, and inauthentic content are top of mind.” (Link to source in 2nd comment)

Anyway, just my two cents as always. Curious what others think after reading Suganthan's latest work. 🫠

[News from Google @NewsFromGoogle]

First, we’re releasing a new interactive “Preferred Sources” button that publishers can embed on their pages. It lets readers add the site as a Preferred Source so they can more easily find their favorite publications in Top Stories, AI Overviews, and AI Mode. So far, people have already selected more than 600,000 unique sources.

[Mani @manibharij]

@RTNToronto Absolutely love this

[Mani @manibharij]

This Substack post is ranking #1 in the web index for this keyword, and the OP most likely never intended it to.

It's has all the right signals:
- original writing and photos
- first-hand experience clearly demonstrated
- highly engaged audience
- better than anything else ranking for the query

[NEXTA @nexta_tv]

🥲 The saddest thing you’ll see today: a Korean man visualized his life through his Google Maps location history

He mapped out a year of his travels — and there was almost nothing on it except one endless route: home → work → home → work.

Every now and then, the route changed. But the excitement didn’t last long: those rare detours turned out to be business trips.

The author didn’t even bother explaining anything and simply captioned the map: “Homeworkhomeworkhomework…”

[Cyrus SEO @CyrusShepard]

Before folks get worked up about AI watermarking, important to note that Google doesn't seem to care all that much

In this @ahrefs study, nearly 1/2 of top-ranking URLs contained at least 20-50% AI text

The very TOP results had slightly less AI text, so consider that...

[Marc Lou @marclou]

I made a mini-app that compares pesticide and longevity for 69 fruits, vegetables, and beans.

It uses 124,880 USDA + UK samples, EPA toxicity benchmarks, and Food Compass 2.0. Click a fruit to see more.

https://t.co/lqryeeYGY2

Take #1: The healthiest fruits/veggies have the most pesticides. Fuck.

Take #2: Beans are superfoods.

[David @dzhng]

This is a meme but I've actually been thinking about this a lot, maybe you can disagree with literally not reading any code, but it's clear that the amount of code being written today far outpaces our ability to review them.

I'm also willing to bet that the code being written will become less and less human readable over time. At the limit the AI should just generate machine code. If AI can emit the perfect assembly, why bother with typescript?

This is actually a huge interpretability problem - how do you trust software systems that becomes more and more of a black box. It's a very different problem from what AI code reviewers are trying to solve today.

I don't have a good solution. Curious what other tokenmaxxers think here.

[Aleyda Solis 🕊️ @aleyda]

.@TomekRudzki from @peec_ai sharing that Reddit and YouTube have dropped as ChatGPT citation sources since the GPT-5.6 rollout is yet another reason why we shouldn’t approach third-party citation optimization as a “one-trick pony.”

It’s also why AI search optimization should be aligned with community management, digital PR and broader digital marketing campaigns.

We can expect third-party platforms to gain or lose prominence as citation sources with every major model update…. these are out of our control as well.

What we can actually control is that our brand has a useful, relevant presence across the meaningful platforms and communities where our audience is actually active and where we can genuinely connect with them.

That’s what makes third-party optimization impactful regardless of citation volatility: building brand visibility and authority that’s aligned and can benefit it across search, AI platforms, communities and other referral channels, rather than optimizing for a single source because it’s highly cited now.

[Google Search Central @googlesearchc]

Today we released the August 2026 spam update. We'll update our ranking release history page when the rollout is complete: https://t.co/O9HQqBPAwY

[Vivek Panwar (Tech Dekhoji ❤️ Media) @TechDekhoji]

My YouTube channel Tech Dekhoji is gone. Just like that 💔

10 YEARS of my life.
1,700+ videos.
Countless days, nights, Sundays & holidays spent building this channel - deleted overnight without a single warning.

What makes this even harder to accept - in 10 years, we never received even ONE Community Guidelines strike. We’re a YouTube Partner and have always taken YouTube’s policies seriously while creating content around tech & gadgets

And yet, the entire channel simply vanished.

No warning.
No opportunity to explain.
No human reaching out before wiping out a decade of work.

This is exactly why creators should be worried about long term partnership & automated enforcement. One wrong decision can erase someone’s life’s work overnight.

I genuinely believe this is a mistake and urgently request @YouTube @YouTubeIndia @nealmohan @YTCreatorsIndia @YouTubeCreators to have a HUMAN manually review what happened to Tech Dekhoji.

To the tech community, creators and everyone I’ve connected with over these 15+ years - I need your support today 🙏

Please amplify this until someone at YouTube actually looks into it.

10 years of work deserves at least one human review before being erased.

[Cyrus SEO @CyrusShepard]

Google's biggest ranking "buckets", according to evidence collected at its antitrust trial, are:

1. Topicality: How relevant is the page to a topic/query. Heavily determined by the "ABC" ranking signals:
• Anchors - the links pointing to the page
• Body - the text on the page
• Clicks - user click signals to determine relevance

2. Navboost: a re-ranking system that uses 13 months of user click data to adjust search results (first/long/last clicks)

3. Quality (Q*): One of Google's most important signals, based on site authority + quality adjustments (e.g., Panda, Helpful Content, spam)

SEOs—including yours truly—often get distracted by shiny things. But focusing on these three ranking buckets will likely account for 80-90% of your Google visibility.

[Charly Wargnier @DataChaz]

every student trying to become an AI engineer after reading Andrew's post:

[Kirill @kirillk_web3]

A Chinese developer just explained the shift from Loop Engineering to Graph Engineering better than anyone.

most people are still building agents the way that's about to be obsolete.

> why single-agent loops break and go "goal blind"
> the 4 parts of a graph: nodes, edges, state, policy
> 3 topologies that run everything: diamond, supervisor, pipeline
> Anthropic's 5 official workflow patterns

the punchline: it's not how many agents you run. it's the determinism you build with verifiers, code fallbacks, and reality anchors.

I broke the same architecture down with Kimi K3. Full A-Z guide below.