Skip to content

How we do SEO, in our own words.

Everything in AwesomeSEO starts from four working documents: a recorded talk about the day-to-day work, the tool that opens every keyword’s top 20, the content-enrichment list and the full working list. Here they are, as written.

The talk.

From a recorded talk (August 2026) about the day-to-day work. Filler words and false starts are removed, names are spelled correctly and client names are left out; nothing is reworded. The label next to each theme is its evidence tier in our playbook.
“Everything that encompasses my day-to-day work, explicitly said and written down, with the aim of giving all this information to an AI that will automate my job.”
  1. Actions and a scheduled crawl

    practitioner

    “The first sheet is named Actions, and it’s literally just a formatted list of actions, with the description of the action and who is designated to be responsible for the action.”

    “Then in that sheet, there should be a scheduled crawl through Screaming Frog that automatically exports, by running headless, to Google Drive.”

    “I know that can be automated. It is partially automated today. It just needs to be standardized and sort of productized internally.”

  2. Keyword research and the niche come first

    documented

    “After that there should be a keyword research, which I think should be done manually. On top of that, there should be also a competitive scouring, like the landscape of the niche. That should be somewhat done manually as well, as that is both foundational stuff.”

  3. Every SERP, the top 20, for years

    practitioner

    “The URLs and the names of the competitors and every single SERP, and that is the search engine result page of every single keyword that one deems viable for one’s business, should be mapped out.”

    “It’s a SERP that for the coming years we will constantly have a look at and study, and see what our competition looks like. What is the information gain that we can offer to the SERP?”

    “Every such keyword, for every website, should have a constant scheduled AI that constantly studies the search engine result page for that keyword.”

    “We only want to study top 20, because we cannot really be sure that Google is not randomizing the rankings.”

  4. Point out where the competition is wrong

    documented

    “That is mapping out the discrepancies and inaccuracies in the competitors, that we can point out and do better than the competitors, and say: this is what X, Y and Z said, but they are wrong, because of X, Y and Z.”

  5. A better page than the other 19

    measured

    “Every content piece produced, that is a page type, that is a template that is done to the highest possible quality.”

    “Having a better above the fold than every single competitor, or the 19 competitors that are in the top 20.”

    “Having a faster website than all the competitors, having a more intuitive mobile website, making the website a PWA, a progressive web app.”

    “Every single page obviously should have some form of intention, a goal that we want the user to take, be that just reading another article, or click an internal link, or downloading a PDF.”

    “Having facts with real exact numbers that are more exact than our competitors.”

    “Having better Core Web Vitals than the competition, obviously.”

    “Having a bottom bar menu on mobile, giving it a mobile app native feel.”

  6. Keep every page alive

    documented

    “At least for the most important pages we want to update it like once a month, depending on the niche.”

    “No page should go ever more than like 3 months without being updated somewhat. And that is at least, say, 10, 15 % of it should be updated.”

    “While we never forget the content that has already been produced, that needs to be kept up.”

  7. One keyword, many page types

    practitioner

    “Having different page types in the form of product page, a blog page, guide page or service page, literally e-com product page, and software SaaS solution page. Every keyword can be framed in so many different ways, and this is the reason to attack it in different ways.”

  8. Publish in clusters, expand step by step

    practitioner

    “Publish always in clusters. So that you map out the entire cluster, that should be around maybe from 5 to 20 pages, depending on the cluster. And then at once, you publish everything together.”

    “You have mapped out all the internal linking. You have integrated that cluster into the bigger cluster that is the website, and make it fit in nicely.”

    “If you write about AI tools, it’s logical that the next cluster would be integration of AI software in businesses or enterprises. That’s a small but logical and reasonable stepping stone, instead of straight up starting to write about AI killer robots.”

  9. Richer language, never forced

    measured

    “Every single page should have a higher degree of human language quality. Somewhat information gain, but also the vocabulary should be more semantically rich, without it being forced.”

    “Having richer and more concise and concentrated language.”

  10. Win the time between core updates

    speculative

    “Between two core updates, no bigger jumps can be done. But what you can do is you can prove yourself to the Google algorithm and outperform your closest competitor.”

    “What you want to do is, during a period of time, outperform your competition, and doing that by publishing high quality, more unique and amazing content, and doing interviews and having quotes, and get the information and data that nobody else has.”

  11. Information nobody else has

    practitioner

    “Finding that data by literally going to Google Scholar or something and saying, hey, this is interesting, or just drawing your own conclusions that you write out.”

    “You can have like a chronicle as a page type that you write every single week.”

    “Making transcriptions of content that is on different modalities, be that a podcast, a YouTube video, a PowerPoint or just images with text. Or just doing an insightful comment, and that creates information gain.”

  12. Every variation Google bolds

    measured

    “Every single variation of the keyword that would trigger a bolded version of that word in the meta descriptions of the results would be considered a variation, as Google literally sees that as the same query.”

    “The plural version of a singular keyword, or the other way around. The cat and the cats and a cat. It shows that you’re covering the entire entity that is a cat, and obviously all the attributes of the entity.”

  13. Agree with the authorities, then add something new

    documented

    “Having, let’s say, 60, 70 percent of the common must-have attributes of the entity, and then once again using information gain and offering new attributes that obviously are valuable, but that you yourself identify or come up with, and that actually offers something new.”

    “What the authority in that niche says is sort of the truth, and you should be mostly aligned with that and not go against it too much. But you want to make some place and go against that and offer something new.”

  14. What other languages already know

    practitioner

    “There are some important attributes that are not covered here for cats, for example that they don’t like dogs. And that is something that the Google Knowledge Graph knows, because in the Polish and the Italian Knowledge Graph it has been clearly stated. It’s well known in other languages that cats do not like dogs, so Google knows this. But it’s never been stated in Swedish. So we are the first ones to do that, so we’re offering something new.”

  15. Top three on many metrics, the best in total

    practitioner

    “If there are 20 results, you don’t want to be the number one on that, you want to be in the top three for as many metrics as possible. So that’s like a compounding effect. In the end, you will be, in total, the best one, but you’re not actively standing out and being suspiciously good on anything special.”

    “If you implement the top 100 most common ones, let’s say having keyword in H1, using variations, using semantically relevant terms, having a fast website, maybe even using schema markup, there’s a statistically relevant correlation between that and high rankings, which definitely proves that SEO actually works.”

  16. A knowledge graph and a source context, from the start

    documented

    “The entire website should in the beginning be mapped out on a knowledge graph, and the connection between the entire subject. Preferably, you should have a really overarching topic. With the example of AI tools, the overarching topic would be artificial intelligence.”

    “Source context should also be established in the beginning: how do you frame every single article? The source context is what you are.”

    “Every single article, be that the product we sell, open data, a blog post or an interview, the way that we present it is framed through the lens of our source context. If we are a community, when we sell a product we would say: this is a product that our community loves.”

  17. An ID for every entity, and the connections between them

    confirmed

    “An ID for each entity. So in the schema markup on the entire website, you don’t have to reference the main entity of the page with the Wikidata ID or Wikipedia URL. Instead, you just reference the ID, and obviously have the sameAs.”

    “Making the connection between the attributes of the entities. We have an entity type that is a person, in AI, let’s say Dario Amodei. He has the attribute that is a title, and his title is CEO of the other entity type that is a company, and that is Anthropic.”

    “We should make connections where possible, especially where there is an information gain.”

  18. Every entity page lists everything about it

    practitioner

    “Every single entity page should have those different formats listed. Like on Sam Altman, we would have all the related blog posts mentioning Sam Altman listed on his entity page. Also, we should have listed his publications, if he has published on Google Scholar. His history should be found there.”

  19. A simple model of how ranking works

    practitioner

    “It is my strong belief that just to be in the top 100, you do on-page SEO. Then, depending on the competition, to be in the top 20, let’s say you have some links. And then, when you’re in the top 10, or let’s say top five, it’s just user signals and historical data.”

    “I don’t have any evidence for that, but that’s my personal take and mental model or framework. It just simplifies a lot.”

  20. One entity, every platform

    practitioner

    “For each page you have a canonical version, and a LinkedIn page for that, and let’s say even a Blogger page, a Medium page, an Instagram post. The page type doesn’t have to be the same, but the entity is the same.”

    “YouTube and LinkedIn should be the priorities.”

    “You should have a podcast also, that can just be the audio version of the article that you have on the website.”

  21. Declare how AI is used

    confirmed

    “You should declare how you use AI. Always with human in the loop, of course. Be very transparent with that.”

  22. Schema that matches the page

    confirmed

    “Never using content in schema markup that is not represented on the page; that has a negative correlation with ranking. But having the things that are present on the page in schema markup as well has a positive correlation, as far as I know.”

  23. Be inside Google’s ecosystem

    speculative

    “You should apply to be a publisher. You should apply to have a Merchant Center. You should try to get in on Google Discover and Google News, and have some Google Ads at least running, just to get in some data.”

  24. Search Console is the best feedback

    confirmed

    “We should obviously check Search Console to see what keywords we are ranking for on a page that are not properly covered. Actually, the best feedback possible we can actually get is going on the real Search Console data and just acting based on that.”

    “Having a server response time that is less than 200 milliseconds.”

  25. Branded searches back the links

    documented

    “The links work better if you can back it up with branded searches. But if nobody’s searching for you, but you have a lot of links and the discrepancy is too big, then you will most probably get punished.”

The method as a tool.

The Crawl Tree Explorer opens a site the way the talk describes it: folder, then URL, then its keywords, then the top 20 in Google for each. The site and the numbers here are generated example data; they say nothing about the real sites named.
Crawl Tree ExplorerExample data

Open it full screen · The same view is built from a real crawl in the Crawler and in the Google Sheets export.

What a page can carry.

Innehållsberikning: 496 points to choose, develop or research for a given page, in 3 parts, verbatim in Swedish. Sammanförd lista över möjliga innehållsberikningar från dina anteckningar och bilagor. Varje punkt beskriver ett element att välja, utveckla eller undersöka för en viss sida.

The full list.

The working list behind every project: 1,483 rows in 11 chapters, in Swedish, as written. Chapter titles are ours; everything inside is verbatim.

What is left out, on purpose

  • Personal to-dos, client and employer names, and plans written for one specific site.
  • Tactics that break Google’s spam policies: link networks, paid or tiered links, bought expired domains, click manipulation and social engineering. We never automate them (our guardrails).
  • The last third of the sheet: notes and tips saved from other people, not our own method.

Where it lives in the product

Every idea in the talk and every row kept from the list is tracked to a playbook practice with its evidence tier, or to the screen that measures it. See the method and the ranking factor library.

See where each idea lives in the products.

The six products put this method to work, step by step.