The SEO Turtle journal
What we’re working on, what we’re learning, and the questions worth spending a little more time with.
Eighteen Perplexity rank trackers compared across pricing, accuracy and reporting. Every entry price re-checked against the vendor's own page on September 1, 2026, and a surprising number had moved. Includes which tools gate Perplexity behind a higher tier.
Perplexity rank tracking has gone from three tools and some hand-waving to a crowded SERP in eighteen months. We tested the tools, ran our own prompts, and dug into Perplexity's actual ranking signals. Here is what is real and what is dashboard theatre.
llms.txt has become the most adopted file in SEO that nobody can prove is being read. We have been running it across client sites for six months. Here is what we are seeing, what the search data says, and whether you should bother.
Google says AI Mode has passed a billion monthly users, and the signals behind its citations are not the ones classic SEO trained you on. Here is the GEO playbook we are using with clients right now.
Most SEO advice assumes you can publish 2,000 word guides. What about sites that genuinely don't have much to say? Here is how we got a single-page game ranking.
Most sites that block AI crawlers block the wrong ones. Here is every major AI bot user agent, checked against OpenAI, Anthropic, Google, Perplexity, Apple, Meta and Amazon's own documentation, sorted into training crawlers, search crawlers and user fetchers, with three robots.txt policies you can paste in.
Screaming Frog has let you run your own JavaScript on every rendered page since version 20. Most people still use it as a link checker. Here is how the seoSpider API actually works, the snippets we run on real crawls, and when the built-in AI tab is the better tool.
Google renders your JavaScript. ChatGPT, Claude and Perplexity do not. That one fact changes what a JavaScript SEO audit has to check in 2026. Here is how Google's pipeline works, how to see exactly what it rendered, and the fixes we keep making on React, Next.js, Vue and Angular sites.
Nearly half of US adults have used ChatGPT and a billion people a month use Google's AI Mode, yet trust in AI answers sits at 20 percent and most users go back to Google to check. Here is what the survey research says about why people use AI search, what they use it for, and what a business should change.
In one week of July 2026, four companies guided to roughly $720 to $745 billion of capital spending this year, most of it on the machines that answer search queries. Here is the money in one table, what it buys, why it means subsidised answers, ads inside them and agents that book for you, and how much of the bubble talk is worth your attention.
Google still takes more than 91% of search referrals, ChatGPT has 900 million weekly users, and Gemini just passed a billion. All three are true and none of them measure the same thing. Here are the 2026 figures with dates and sources, what market share means when the AI answer lives inside Google, and the number a business should actually be watching.
Legal is the most expensive category in Google Ads again this year, and most law firm accounts waste the budget on structure, not bids. Here is how we would build a law firm PPC account in 2026: which campaign types earn a place, what AI Max and ads in AI Mode change, and how to keep intake and compliance under control.
A dental practice competes in a three-mile radius, and most of that competition now happens inside Google Business Profile, the local pack and AI answers rather than on the website. Here is the working checklist we use for practices: categories, services, reviews that stay inside Google's and the GDC's rules, treatment and location pages, schema and tracking.
Most multi-location sites fail in one of two ways: one page for the whole brand, or fifty pages that say the same thing with the city name swapped. Here is how we structure Business Profiles, location pages, schema, reviews and tracking for a business with more than one address, and what changed now that Google's AI answers read the profile before the website.
Most backlink advice still reads like 2015: chase DA, get the link in the body, make sure it's followed. Here is how we actually judge a link in 2026, what Google has said about links mattering less, and a checklist you can run on any prospect in five minutes.
Most of the industry pays for links and most of it will not say so. Here is what buying backlinks looks like in 2026, what Google's systems actually do with a paid link, what the May, June and August updates targeted (and did not), and where we would put the same budget instead.
B2B keyword lists get pruned by volume before anyone checks who is searching. That was always a mistake, and now that AI Overviews and AI Mode absorb most of the research phase it is an expensive one. Here is how we plan B2B SEO in 2026: which pages still earn clicks, how to get cited when they do not, and how to measure it without pretending clicks are the whole story.
Google has said for nearly twenty years that a .edu link gets no special treatment, and that it ignores a lot of them because they get spammed so hard. Some university links are still worth real effort. Most of what is sold as edu backlinks is not. Here is how to tell the two apart and what the outreach actually looks like.