To rank on Google, pick a search your buyers really type, study what the results page already shows for it, and publish a page that answers that search better than anything there. Get the on-page basics right, earn links from sites that vouch for you, and measure your position from where your buyers actually search, because a rank read from the wrong place can mislead you badly.
That's the method. The rest of this post explains each part, what it asks of a small team, and where people lose months doing the wrong work.
Read the results page before you write anything
Keyword research tells you what people type. The results page tells you what Google thinks they want, and that matters more. Thundertech's checklist of ranking tips puts this step first. Before writing, look at the page for your keyword at that moment and note which features appear on it. A featured snippet, a video carousel and a People Also Ask box each signal a different kind of answer, and your content should be shaped to win the one that's there.
If Google leads with a video, a page of text is fighting the format. Thundertech's advice for that case is blunt and correct. Either make an original video, or go back and choose a different keyword.
This is where most wasted effort starts. Teams pick a keyword from a tool, write a long article, and never look at what actually ranks. Do the reading by hand for every query you care about. It's slow, and it's the cheapest research you'll ever do.
Get the on-page basics right once
These are the parts of a page Google reads first. They won't make a weak page rank, but getting them wrong makes a good page harder to find and less likely to be clicked.
- Title tag. The bold headline on the results page. Thundertech recommends including your primary keyword and keeping it between 60 and 70 characters.
- Meta description. The text under the title. Aim for around 155 to 160 characters with the keyword included. It matters less for ranking than the title, but it shapes whether people click, and Google bolds the words that match the search.
- H1. The main heading on the page, and the clearest statement of its topic. Include the keyword, and vary it slightly from the title tag. A common pattern is a descriptive H1 with a "Keyword | Brand" title.
- Alt text. A short description of each image. Search engines can't read images without it, and people using screen readers depend on it too.
- URL. Short, simple, with the target keyword where it fits.
- Related terms. Words that naturally belong with your topic, used in the copy and in subheadings, without forcing them.
Check your robots.txt file as well. It tells crawlers which pages to fetch and which to skip, and a stray rule there can hide pages you want found.
If you own the site's code, generate these fields from data in the repository and fail the build when one is missing or too long. A checklist in a spreadsheet gets ignored within a month. A failing build doesn't.
Earn links and skip the shortcuts
Google is where the searches are. SEOSpace puts its share at 83.5% of global search traffic, and lists the signals it weighs, including the quality of your content, how relevant it is to the search, and the number of high-quality, relevant links pointing to your site. Links from authoritative sites act as endorsements. You earn them by publishing something other people want to cite, by building real relationships with sites in your field, and by sharing your work where your audience already is.
The shortcuts have a long and unhappy history. A practitioner on Hacker News noted back in 2014 that Google had massively discounted exact-match domains, so buying a domain that spells out your keyword stopped being a trick. The same comment pointed at a decade of shady link networks selling links by the million. Our position is simple: don't buy links, and don't hire anyone who sells them as a package. A link you paid for is one you'll eventually want gone.
Make every page earn its place
Service firms and software companies often want a page per city, industry or use case. That's a sound idea when each page says something the others don't, and a fast way to publish a pile of near-duplicates when it doesn't.
Heddle's method was built on one of our own sites, and that site shows both failure modes. Two of its city pages once shared 67% of their sentences. That means most of the copy was the same. Heddle now measures uniqueness on the built HTML against an 80% ceiling, so a page that's mostly a copy of its sibling fails before it ships.
The second problem is quieter. In one family on our own site, thirteen of twenty-five pages were never fetched by Google because every page linked the same nine siblings. The other pages had no path in. A page Google never fetches can't rank, however good it is. Heddle rotates sibling links so every page in a family gets linked from somewhere.
Those two checks sit among the 16 Heddle runs on built pages, alongside claims with evidence, a buyer test and agent readiness. Whatever tooling you use, the principle carries over. Check the page Google will actually receive, not the template or the draft.
Measure from where your buyers search
You can't improve what you can't see. SEOSpace recommends Google Search Console and a rank tracker to watch your positions over time and spot what needs fixing. That's sound, with one caution that catches a lot of local businesses.
Rankings change with location. A page on one of our own sites reported at #48 was #3 when read from the city it targets. If your tracker checks from a default location, you might rewrite a page that's already doing its job, or celebrate one that isn't. Set your tracker to the cities your buyers are in.
A note on answer engines. Heddle writes an llms.txt file and a Markdown twin of every page from the built HTML, for AI tools that read them. They don't help with Google, though. Google said on 15 May 2026 that Google Search ignores llms.txt and Markdown files. Treat them as work for AI answers, and judge your Google rankings on the pages themselves.
No honest person will promise you a position. What you can do is measure accurately and change one thing at a time, so you know what moved.
What it takes, and when to pick a different search
Ranking well asks for three things a small team often lacks: time to write pages that answer a search properly, someone willing to stand behind every claim on them, and the patience to measure before rewriting. Agencies sell the first. Tools sell dashboards. Neither gives you accountability on its own.
It doesn't fit every search. If the results page is all video and you won't make video, move on. If the top results are huge reference sites and your page can't add anything they lack, pick a narrower search your buyers also type, such as one with your service and your city in it. Winning a specific search that brings enquiries beats sitting on page three of a broad one.
Frequently asked questions
Does having my keyword in my domain name help me rank on Google?
Not in any meaningful way. Google heavily discounted exact-match domains years ago, so a clear brand name and a page that answers the search will do more for you.
How do I check where my website ranks on Google?
Use Google Search Console for your own site's queries and a rank tracker for specific keywords, set to the location your buyers search from. A rank checked from the wrong city can be off by dozens of places.
Does llms.txt help me rank on Google?
No, Google said on 15 May 2026 that Google Search ignores llms.txt and Markdown files. They're worth having for AI answer tools, but your Google ranking depends on the HTML pages themselves.
Can I make lots of city or service pages to rank in more places?
Yes, as long as each page carries material its siblings don't and every page is linked from somewhere. Near-duplicate pages and pages with no inbound links tend to go unranked or unfetched.