How to Rank Better on Google: Tools, Steps and Timelines

Most articles about ranking better on Google give you a flat list of fifty things to do and no idea which three matter this week. This one is organized differently. I go through the levers in the order I would actually pull them, and for each one I tell you which tool I reach for, whether I have tested it myself, and whether you need to pay for it at all.

Two things shape my point of view. First, I use Semrush daily for in-house SEO work, so my opinions about it come from repetition rather than a trial account. Second, over the past few months I have been watching a new category form: SEO work driven through an AI coding agent, specifically Claude Code paired with community repositories of SEO skills and commands. I think that shift is going to put real pressure on the classic all-in-one platforms, and I say so plainly further down.

Quick answer: what actually moves rankings

If you only read one section, read this one.

1. Make sure Google can crawl and index the page. Nothing else matters if the page is not in the index.
2. Match the intent behind the query better than the pages already ranking. This is the single most common failure I see.
3. Fix the obvious on-page signals: one clear title, a specific H1, descriptive URL, headings that mirror the questions people ask.
4. Strengthen internal linking so your important pages are not buried four clicks deep with generic anchors pointing at them.
5. Earn links and mentions from sites that have nothing to do with you. Still the hardest lever, still one of the strongest.
6. Improve the pages that already rank between positions 5 and 20 before writing anything new. Fastest legitimate wins available.
7. Keep the winners fresh and prune or merge the pages that never earned an impression.

Effort versus impact, roughly: indexing checks and intent fixes are cheap and fast. On-page and internal links are cheap and medium speed. Links and topical authority are expensive and slow. Page speed work is medium effort and usually only matters once the rest is solid.

One promise nobody can make, including me: you cannot pay Google for organic rankings. Ads buy placement in the ad slots, and that is a separate auction. Anyone offering guaranteed position one in thirty days is selling you something else.

First decide which kind of ranking you are chasing

This is where a lot of effort gets wasted. There are at least four different games on a Google results page and they respond to different inputs.

Organic web results respond to content quality, relevance, internal structure and links. The local pack and Maps respond to your Google Business Profile, proximity to the searcher, and prominence signals including reviews. AI Overviews and AI Mode respond to whether your page is a clean, quotable answer to a specific sub-question, which overlaps with organic ranking but is not identical to it. Images and video have their own logic entirely.

The decision path is simple. If you are a local service business, start with your Business Profile and reviews, not with blog posts. If you are a SaaS, ecommerce store or publisher, start with organic on-page and content structure. If you run both a physical location and a content site, run the two playbooks in parallel but keep them mentally separate. Mixing them is the most common reason people work hard for three months and see nothing.

How Google decides rankings, in plain language

Google interprets the query, retrieves candidate pages, then ranks them using signals about relevance, quality, usability and context (location, language, device, some personalization). That is the honest summary. Every published list of “200 ranking factors” should be treated as directional folklore rather than a specification.

Some things I would stop worrying about immediately:

Keyword density is not a thing you should calculate. LSI keywords are not a real Google mechanism, whatever the plugin says. There is no ideal word count, and the widely repeated “2,100 to 2,400 words is optimal” figure is a correlation someone published years ago that has since been recycled into gospel. Length should follow the question, not a target.

What does hold up in practice: pages that answer the actual question early, sites that are structured so related pages reinforce each other, and domains that other people reference without being asked.

Step 1: confirm Google can find and index you

Free, five minutes, and it invalidates half the panicked emails I get.

Open Google Search Console, paste the URL into the URL Inspection tool, and read what it says. “URL is on Google” means you have a ranking problem. “Crawled, currently not indexed” or “Discovered, currently not indexed” means you have an entirely different problem and no amount of on-page tweaking will help.

The usual silent killers: a stray noindex left over from staging, a canonical tag pointing at another page, robots.txt blocking a directory, or content that only renders after JavaScript executes in a way Googlebot cannot resolve. Sitemaps help large or poorly linked sites get discovered. They do nothing for a fifty-page site that already has decent internal linking.

Step 2: find keywords you can realistically win

Start free. Search Console’s performance report already tells you which queries bring impressions but almost no clicks. Those are queries where Google thinks you are relevant and users disagree, which is the shortest path to more traffic you will ever find. Add Google autocomplete, related searches at the bottom of the results page, and the People Also Ask boxes for question phrasings.

Free research runs out of road when you need two numbers: how much search volume a term carries, and how hard the existing results are to displace. That is the moment a paid platform earns its keep, and it is the main reason I still pay for one.

Judging winnability is less about a difficulty score and more about looking at who ranks. If positions one through ten are all domains with ten years of history and hundreds of referring domains, and you launched in March, pick a longer-tail variant. Question keywords and specific comparisons are where new sites break through. Then build clusters around them rather than isolated posts, so each new page has somewhere to link from and to.

If you want the wider view of what a full stack costs, I broke down pricing and my own choices in my rundown of the best SaaS tools for SEO in 2026.

Step 3: match intent better than the pages already ranking

Open the top five results for your target query. Note the format of each: listicle, tutorial, comparison, product page, tool. Note what all five cover, because that is the baseline Google has decided the query requires. Then note what none of them cover, because that is your opening.

This is not theory. It is how this article got its outline. The pages currently ranking for “how to rank better on Google” almost never explain how to measure whether anything worked, and they say nothing about the tools you would use to do it. So that is where I spent my words.

Intent maps to format. Informational queries want a guide or an explainer. Commercial investigation wants a comparison with trade-offs and prices. Transactional wants a page where the action is one click away. If you publish a blog post against a results page made entirely of product pages, you will not win, no matter how good the writing is.

When an existing page is already ranking anywhere in the top three pages for a related query, upgrade it. Create a new page only when the intent is genuinely different, otherwise you are building your own cannibalization problem.

Step 4: on-page work that still matters

Titles carry weight. Make them unique, specific and led by the phrase people actually type, without stuffing. H1s should match the promise of the title, and subheadings should mirror the sub-questions users ask, because that is what gets pulled into featured snippets and AI Overview citations.

Meta descriptions do not rank you. They influence click-through, which is worth optimizing for its own sake. Descriptive URLs, compressed images with real alt text, a clean heading hierarchy: cheap, quick, still worth doing.

Two habits pay off disproportionately. First, answer the question in the first two sentences under the relevant heading, then expand. Second, add schema markup where you are genuinely eligible: Article, Product, LocalBusiness, Breadcrumb. Rich results widen your row in the results page, and a wider row gets more clicks at the same position.

Step 5: build topical authority with internal links

Group related pages into a hub with spokes. The hub targets the broad term, the spokes target specifics, and every spoke links back to the hub with an anchor that describes the destination. Anchors like “read more” waste a signal you are allowed to send.

Keep important pages within three clicks of the homepage. Audit for two pages competing for the same query and merge them, redirecting the weaker one. Handle genuine duplicates with canonicals rather than hoping Google guesses right.

Internal linking is the highest leverage work available to a small site because you control all of it. No outreach, no budget, no waiting.

Step 6: earn authority, links and mentions

Links from unrelated sites remain one of the strongest quality signals Google has, which is exactly why they are hard to get. The tactics that still work are unglamorous: publish original data nobody else has, get quoted as a source, reclaim mentions of your brand that were never linked, appear on podcasts, contribute genuine expertise to publications in your field. For local businesses, add consistent citations across directories with identical name, address and phone details.

What to avoid, per Google’s own spam policies: buying links, private blog networks, industrialized guest posting, and renting out subdirectories of your domain to a third party (site reputation abuse). These are all faster. They are also the reason people arrive with a manual action and no idea what happened.

Semrush: the platform I use every day

I use Semrush as my default. Keyword research, position tracking, site audits, competitor gap analysis, all in one place, which matters more than any single feature being best in class. My long-form take is in my Semrush review after a year of daily use, and if you are deciding between the two big platforms, I compared them directly in Semrush vs Ahrefs.

The short version of where it fits in this playbook: keyword volume and difficulty for step 2, competitor SERP analysis for step 3, on-page recommendations for step 4, internal link and technical issues for steps 5 and 7, and position tracking to know whether any of it worked. The price is the honest objection: $139.95 a month before add-ons and extra seats. For a solo site owner it is a lot of money for capabilities you will use twice a month.

Ahrefs: strongest for links, and there is a free door

Ahrefs is the one I reach for when the question is about backlinks: who links to a competitor, which of my mentions are unlinked, what a domain’s referring profile looks like over time. Its index has a reputation for link data quality that has held for years.

Before paying, note that there is a genuinely useful free tier. Ahrefs Webmaster Tools gives you site audits and backlink data for domains you can verify, and I went through exactly what you do and do not get in my Ahrefs Webmaster Tools review. For a site owner whose main gap is link visibility, that free plan plus Search Console covers a surprising amount of ground.

I have not run Ahrefs as my primary daily platform the way I have Semrush, so treat my comparisons as informed rather than exhaustive.

Claude Code plus Daniel Agrici’s SEO repo: the one I think changes the game

This is the part of the article I care most about, and it is also the least settled.

Claude Code is an AI coding agent that runs in your terminal, reads your files, executes commands and iterates. Daniel Agrici has published claude-seo, a GitHub repository that turns it into an SEO working environment: 25 skills and 18 sub-agents aimed at the kind of tasks you would otherwise pay a platform to do.

Why this matters more than it looks. A traditional SEO platform is three things bundled together: proprietary data (search volumes, link graphs), a computation layer (audits, scores, clustering), and an interface. An agent with the right instructions can rebuild most of the computation layer on demand, against your own data, in whatever shape you want. Instead of exporting a report and reformatting it, you describe the analysis and get it. Instead of paying for a feature you use twice a month, you write a command once.

My honest position: with Claude Code in the picture, the usual tools may well end up jeopardized. Not obsolete tomorrow, and not for everything. The proprietary data moat is real, because nobody is going to crawl the web’s link graph from a laptop. But the part of the subscription that is really just “a nice interface around calculations anyone could do” looks a lot more fragile than it did a year ago.

I have not yet run a full side-by-side against my Semrush workflow, so I am not going to pretend I have numbers. What I can say is which tasks I expect to migrate first: crawling and technical audits of my own sites, internal linking analysis, content gap analysis against a set of URLs I supply, bulk on-page checks, redirect mapping, schema generation, and Search Console data analysis at a level of specificity the interface does not offer. What I expect to keep paying for: keyword volume, keyword difficulty, and third-party backlink data.

If you are technical enough to be comfortable in a terminal, I would test this before renewing an annual plan. If you are not, the classic platforms remain the sensible choice, and that is not a consolation prize.

Ranking a business in the local pack and Maps

Different game, faster results. Google says local ranking comes down to relevance, distance and prominence. You cannot change distance. You can change the other two.

Claim and verify the profile, then fill in every field: precise address, hours including special hours, primary category chosen carefully plus secondary categories, attributes, services and products. Add real photos, not stock. Post occasionally. Answer the Q&A section yourself if nobody else does.

On your own site, back it up with a location page carrying consistent name, address and phone details, embedded map, and content that reflects the area you serve rather than a template with the city name swapped out.

Google reviews and star ratings, and why they are not rankings

There is persistent confusion here, so let me separate the two things. Your star rating is not an organic ranking factor for your website. Review volume, recency and rating do feed local prominence, and they heavily influence whether someone clicks you instead of the competitor listed above you. Different mechanisms, both worth attention.

Is 4.2 a good Google rating? It is acceptable and it is believable. In most service categories, 4.2 sits in a normal band: high enough that people will consider you, low enough that they trust the number. What hurts is not 4.2 versus 4.6, it is a 4.2 built from nine reviews, or a 4.2 where the three most recent reviews are unanswered complaints. Volume and recency do more work than the decimal.

How do you get to a 5.0? You mostly should not aim for it. A flat 5.0 with a meaningful number of reviews reads as suspicious to shoppers, and research from Northwestern’s Spiegel Research Center found that purchase likelihood peaks around 4.2 to 4.5 stars and dips as ratings approach a perfect 5.0. Beyond that, the compliant path is narrow: ask every customer, at the moment they are happiest, with a direct link. Do not offer incentives. Do not filter for happy customers before asking (review gating breaks Google’s policies). Respond to everything, briefly and without defensiveness, because the response is read by the next prospect, not the reviewer.

How to check your Google rankings

Search Console’s average position is free and honest, but understand what it measures: the average of the highest position your site appeared at, across every impression, blended over locations, devices and query variants. A move from 8.4 to 6.1 is meaningful. A single number for a single keyword is not a ranking check.

Manual checks work if you do them carefully: private window, correct country and language settings, and awareness that your result is not the result someone in another city gets. Do them for sanity checks, not for reporting.

Dedicated rank trackers, including the position tracking modules inside Semrush and Ahrefs, are what you use when you need daily keyword-level history, device split and a competitor set tracked alongside you. For local businesses, standard rank tracking is not enough, because your position changes street by street. Geo-grid tools that check a keyword from dozens of coordinates around your location give you the picture that matters.

Whatever you choose, build a weekly view with four things on it: indexed pages, impressions, average position, clicks. In that order, because that is the order they improve.

How to rank better on Google for free

A complete no-budget stack exists: Search Console for queries and indexing, Google Analytics for behavior, Business Profile for local, PageSpeed Insights for performance, Keyword Planner for rough volume, and autocomplete plus People Also Ask for phrasing. Add Ahrefs Webmaster Tools for backlink and audit data on your own domains.

A workable thirty-day plan with zero spend. Week one: fix indexing, confirm every important page is on Google, and set up Search Console properly if it is not. Week two: pull every query with impressions and few clicks, pick the ten best, and rewrite those pages’ titles and intros to match the query. Week three: internal linking pass, hub and spokes, better anchors, cannibalization merges. Week four: one genuinely useful new page targeting a long-tail question, plus a Business Profile completion sprint if you are local.

Free stops being enough at the point where you need to know whether a keyword is worth chasing before you write, and whose links are holding a competitor in position one. That is the cheapest justified upgrade, and it is where a single paid platform, or an agent-driven setup like the one above, starts paying for itself.

How fast can you actually rank?

Honest ranges, not promises. A brand-new domain going after competitive commercial terms: expect six to twelve months before anything resembling stable rankings, and expect the first six weeks to look like nothing is happening. An established site with existing authority publishing into a topic it already covers: days to a few weeks for long-tail terms.

The fast levers are specific. Pages already sitting between positions five and twenty respond quickest to intent and title improvements. Business Profile optimization can shift local visibility within days. Long-tail question pages on a healthy domain can rank almost immediately. Everything else is slower than you want it to be.

Watch leading indicators before rankings move: pages getting indexed, crawl frequency rising, impressions climbing while position is still poor. Those come first. Clicks follow.

Why your rankings dropped

Diagnose before you touch anything, and start with scope. Sitewide drop across all sections points at a core update, a technical regression, or a manual action. Section-level drop points at something you changed in a template or a directory. Single-page drop usually means a competitor improved their page, or you cannibalized yourself with something new.

The recurring self-inflicted causes I see: two pages targeting one query, thin content published at scale, over-optimized anchor text, a redesign that dropped headings or changed URLs without redirects, and migrations that shipped a noindex to production. Check Search Console’s coverage and manual actions reports first. Then check what changed on your side in the two weeks before the drop, because it is almost always that.

Your ranking checklist

Weekly: Search Console impressions and average position, any new indexing errors, one page improved from the positions five to twenty list.

Monthly: internal linking pass on new content, one refresh of a decaying page, one link or mention earned, Business Profile posts and review responses if local.

Quarterly: full crawl and technical audit, content pruning decisions, competitor gap analysis, and a review of whether your tool stack still earns what you pay for it.

Judge success in stages. Impressions first, then average position, then clicks, then conversions. Anyone measuring conversions in week two will conclude SEO does not work.

FAQ

How do I improve my Google ranking? In order: confirm the page is indexed, match search intent better than the current top five, fix titles and headings, strengthen internal links, earn external links, then measure with Search Console.

How to rank fast on Google? Improve pages already ranking between positions five and twenty, target long-tail question keywords, and if you are local, optimize your Business Profile. Those are the only genuinely fast legitimate levers.

Can I pay Google for better organic rankings? No. Ads buy ad placement. Organic positions are not for sale, and anyone claiming otherwise is misrepresenting what they do.

How long does SEO take to work? Weeks for easy wins on an established site, six to twelve months for competitive terms on a new domain.

Is 4.2 a good Google rating? Yes, in most categories. Volume and recency matter more than the decimal point.

Does my ranking differ by location and device? Yes, substantially, especially for anything with local intent. That is why a single manual check is not a measurement.

Do AI Overviews replace ranking number one? They change the value of it. Being cited in an AI Overview and holding the top blue link are related but separate outcomes, and answer-first formatting helps with the former.

Which tool should I start with? Search Console, because it is free and it is your data. Then Ahrefs Webmaster Tools for link and audit visibility on your own site. Then one paid platform when you need volume and difficulty data before you write. And if you are comfortable in a terminal, test Claude Code with an SEO repo before committing to an annual subscription, because I suspect that comparison is going to look different in a year.