Nathan Gotch has done SEO since 2011 and full-time since 2013, and he started working on AI search the month ChatGPT launched. In a recent video he explained how he would learn AI SEO if he started over in 2026. His answer is to skip the course and build a tiny business, then use AI search as its only way to find customers. Here is his plan step by step, and which steps you can hand off.
Step 1: Pick one tiny business to grow
Nathan's first rule is to build something and learn how to grow it. He calls the method just-in-time learning: you work on a real project, and every time you hit a roadblock, you find the answer, apply it and keep going. Each problem you solve becomes a skill you keep.
His recommended project is a micro SaaS, a very small software product that solves one specific problem. The goal is not to build a big company. The goal is a real business where AI SEO is the main way you get customers, so every lesson is tied to an outcome. If it makes money, even better.
How to choose it:
- Pick a category you understand.
- Pick one ideal customer inside it.
- Pick one problem that customer has.
His shortcut for finding the problem: look at an established software suite in the category and pick one feature they already sell. His example was the prompt research feature inside a big SEO suite. You then build a product that does only that one job, better than the suite does. The fact that a large company already sells it is your validation.
He also shared how he learned this the hard way. His first site was a baseball pitching blog. He spent his time tweaking the website and made no money, because the site was never the hard part. Getting customers was.
Milestone: one sentence that names the category, the customer and the single problem you solve.
Step 2: Sell it as a service before you build software
Most people build software for two years and then try to sell it. Nathan says to do it the other way around.
- Build a one-page website that sells the outcome as a service. For the prompt research example: a one-time price, and you deliver the prompt research yourself.
- Keep the price low. He gives $47 or $99 as examples.
- Deliver the work with whatever tooling you have. The customer never sees your software.
- Ask every customer for an honest review.
- Fix your tooling every time delivery exposes a weakness.
Nathan offers a full refund to customers who leave a review, and he stresses that he asks for an honest review, not a positive one. If you copy this, check the rules of the platform first. Google's review policy, for example, does not allow incentives for reviews on Google, so keep any rebate away from those.
His target is 50 to 100 delivered orders. That gives you cash flow, feedback, a stack of social proof that AI assistants can later find, and software that has been tested on real work before anyone logs into it.
He has done this himself recently. He built a one-page website with an AI coding agent in about 15 minutes, set up for crawling and indexing, and had revenue within 24 hours. He sold an AI SEO audit as a one-time service and it sold out almost immediately.
Milestone: a live one-page site with a price and a checkout, and a first paying customer.
Step 3: Learn enough vocabulary to name your problems
Nathan's third point is about language. If you do not know the words, you cannot even describe the problem you are stuck on, and you cannot search for the answer or ask an AI assistant for it.
He recommends reading a few foundational SEO books. Two he names that are not his own are The Art of SEO, which goes deep on technical SEO, and Product-Led SEO, which is written from the software side. Use them as references rather than reading them cover to cover before you start.
His method for the vocabulary: open the index at the back of a book, pick one term per day, and dig into that one idea. Then apply it on your project. You do not become an expert by memorizing terms. You become one by using them and seeing what happens.
Milestone: a running glossary of the terms you have actually used on your project, one line each.
Step 4: Set up your workspace and your data
Nathan does 99% of his work inside an AI coding agent. He uses Codex and says Claude is fine too. His setup:
- Create one project for the business. Everything related to it lives there.
- Write a context file first. Describe the product, the one problem, the customer and the fact that you sell it as a service. His rule: assume every agent knows nothing, every time.
- Use one chat per workstream, for example one for the product and one for the AI SEO work. When a chat gets too long and the answers get strange, start a new one and pass the old chat in as context.
- Give the AI real data. An AI without grounding makes things up. Connect an SEO data source, even a cheap pay-per-use API, so the answers come from real search data.
Then set up the free tools every AI SEO project needs:
- Google Search Console, for your Google data.
- Bing Webmaster Tools, because ChatGPT and other AI platforms use Bing as one of their search indexes.
- Google Analytics 4, plus Microsoft Clarity if you want session recordings.
He adds that AI agents are very good at technical setup and analytics configuration, especially for tools with a detailed public help center. The agent can read the documentation and learn the tool for you.
Milestone: Search Console, Bing Webmaster Tools and GA4 verified on your site, and a context file your agent reads before every task.
Step 5: Define the category and turn seeds into prompts
The most important strategic step, according to Nathan, is picking one focus. AI SEO has an endless list of tasks. Pick the one area that would change your business if you won it in the next 90 days, and give it all your attention. If you are a plastic surgeon, pick one procedure, not the whole practice.
He is testing this live on his own book launch. About two weeks before the video, ChatGPT did not mention his book for "best AI SEO books". At recording time it did, while the book was still on pre-order. He said he would share the full list of tactics after the launch.
The process:
- Research the topics, entities and keywords around your one problem.
- Turn your seed keywords into prompts, the way a buyer would ask ChatGPT.
- Draft a content plan for the first 90 days.
He also keeps an 80/20 split: 80% of content is built around the core topic, and 20% can be experimental.
Milestone: a list of 20 to 30 buyer prompts for your one focus area.
Step 6: Track prompts by hand and study three competitors
You do not need an AI visibility tracker at the start. Nathan runs the prompts manually:
- Open ChatGPT in an incognito window, logged out.
- Run each commercial prompt and save the answer and every citation.
- Ask your AI agent to extract the brands and entities from the answers.
- Ask it to crawl every cited page and report whether your brand is mentioned.
On a new site you will not be mentioned anywhere. That is fine. The point is to repeat the same process in six months and see you on two of the ten sources.
The answers also show you who you are up against. Nathan's advice is to pick your top three competitors and study them for a long time. Where do they get reviews? How much do they publish? Where do they distribute? Which creators talk about them? Most of the answer is off their website, in the sources that vouch for them. Match what they have, then do more.
When you prioritize, put reviews, links, brand mentions and your other owned channels first. Rewriting meta descriptions goes at the bottom of the list.
Milestone: a spreadsheet of your prompts, the brands each answer names, the pages it cites, and your three main competitors.
Step 7: Publish 10 assets in 90 days
Get the technical foundation right first. Nathan says this is where AI agents are strongest, so learn what the checks mean and let the agent do the work. Then publish ten core assets:
- 3 data-driven studies tied to what you sell. For a prompt research product, you could run the same prompts in ChatGPT, Perplexity and Claude and publish the differences. This is first-party data nobody else has.
- 2 free tools. They attract links, and building them teaches you a skill that keeps paying off.
- 5 commercial pages, starting in the middle of the funnel and moving down: a "best tools" listicle with your product included, alternatives pages for the big suites, and head-to-head comparison pages.
Then take the same topics to YouTube, LinkedIn and X in each platform's own format. When you publish a video, embed it on the matching page on your site. Nathan describes this as linking your content together into a small network of your own.
Milestone: ten assets live on the site, and each topic published on at least one other channel.
Step 8: Build visibility beyond your website
AI assistants repeat what other sources say about you. So the next step is to change what those sources say.
- Make sure every profile you own describes what you do accurately and in the same words.
- Keep collecting reviews.
- Reach out to the pages that AI answers cite for your prompts. Nathan's template is a short, plain email: you launched a tool that does X, and you wonder if they accept new submissions to their list.
- Answer questions where your buyers ask them. Nathan uses an agent to find and answer relevant Quora questions, leading with help and only mentioning his product when it fits.
He is clear that outreach is a volume game. Measure your output, meaning how many emails you sent today, rather than worrying about each reply.
Milestone: outreach sent to every page cited for your top prompts, and your profiles saying the same thing everywhere.
Step 9: Measure, fix, repeat
At the start, keep tracking manual. Set up dedicated accounts for tracking, turn memory off, and run your prompts several times a week, ideally daily. When you run several projects or a serious campaign, switch to a tool. We compared the main AI visibility tools if you get to that point.
Nathan's closing point is the one I agree with most. AI is good at execution and bad at strategy. You decide what to work on and why. The agent does the work.
Milestone: a weekly check of your prompt list, and one change made because of what you saw.
Automate it with Distribb
Nathan's plan is a good way to learn. It is also a lot of recurring work once you know how, and most of it repeats every week. Here is which steps Distribb's AI SEO agent takes off your plate, and how:
- Step 4, data and grounding. Distribb connects to your site and your Google Search Console. The agent answers from your own keywords, articles and Search Console data, not from guesses. If you already work in Claude or Codex like Nathan, you can install the Distribb skill there and ask for SEO work from the agent you already use.
- Step 5, research and planning. Keyword research runs on your market and competitors, finds buyer-intent keywords, groups them into clusters, and keeps a 30-day rolling content calendar full.
- Step 7, the assets. Articles are written in your brand voice, including comparison and alternatives content, and published to your CMS on schedule. Distribb can scrape real data and build a source-backed table into an article, which covers part of the first-party data idea. On Accelerator, your best articles are also turned into videos, carousels and posts for YouTube, Instagram, Medium and Quora.
- Step 8, visibility beyond the site. Distribb's backlink exchange earns contextual links from real businesses, and Link Outreach researches the lists you should be on and drafts the pitches.
- Step 9, measurement. AI visibility tracking covers five AI engines: which prompts mention you, which pages are cited, and how you compare with competitors. The agent can also find pages that are losing clicks and rewrite them while keeping the URL and internal links.
Steps 1 to 3 stay with you. Choosing the problem, selling the first service and learning the vocabulary are the parts that make you good at this.
Pro is $97 a month with 30 articles. Accelerator is $495 a month with 90 articles, done-for-you distribution and a human on our team who reviews every piece. Both have a 7-day free trial.
If your micro business is software, our SaaS SEO checklist covers the foundation, and our answer engine optimization guide goes deeper on getting cited.
Watch the full video
Nathan's video runs about 44 minutes and includes a screen share of his agent setup.
FAQ
What is the fastest way to learn AI SEO?
Build a small real project and grow it with AI search, learning each concept when you need it. Nathan Gotch recommends a micro SaaS that solves one problem, sold first as a low-priced service. Every roadblock you solve on a real business teaches more than a course you watch without applying.
Do I need an AI visibility tool to start?
No. Run your buyer prompts manually in ChatGPT in an incognito window, logged out, and save the answers and citations. Use dedicated accounts with memory turned off for repeat checks. A paid tool becomes worth it when you run several projects or need daily tracking across many prompts.
What content should a new site publish for AI search?
Nathan's 90-day plan is ten assets: three data studies tied to your product, two free tools, and five commercial pages such as a "best tools" listicle, alternatives pages and comparison pages. Then republish the same topics on YouTube, LinkedIn and X and embed the videos on the matching pages.
Why set up Bing Webmaster Tools for AI SEO?
ChatGPT and other AI platforms use Bing as one of their search indexes. Bing Webmaster Tools shows how Bing crawls and indexes your site, so you can fix problems that would keep your pages out of the results those assistants pull from. It is free, like Google Search Console.
Who is Nathan Gotch?
Nathan Gotch is an SEO who has worked in search since 2011 and full-time since 2013. He runs an AI SEO software company, teaches SEO on YouTube and wrote a book on AI SEO for the For Dummies series. He started focusing on AI search in November 2022, when ChatGPT launched.
Build the thing, let the SEO run
Nathan's best advice is to spend your time on the problem your business solves. Once you know the steps, you do not need to repeat them by hand every week. See how the Distribb AI SEO agent handles the research, publishing, links and tracking, then check Distribb pricing and start the free trial.