SEO that compounds from day 2

1. 1Open Google Keyword Planner and enter a few keywords around your product
2. 2Sort by traffic, filter out high-competition keywords
3. 3Pick the top 10 and place them natively in your homepage copy and meta tags
4. 4Add one blog article every week, manually or with an AI blogging tool

Keywords EverywhereBrowser extension that shows search volume, CPC and competition inline on every Google search, faster than tabbing back to Keyword Planner.

A keyword's value is roughly its monthly search volume times its cost per click: that's what advertisers collectively pay to own it. "AI video generator" at ~60,000 searches and ~$2 CPC is a ~$120,000/month keyword. Reading the two numbers together tells you what kind of market you're looking at: high volume + high CPC is a strong market, low volume + high CPC is still worth it when intent is strong, and high volume + $0 CPC is probably informational traffic that never converts. Before building a page (or a product) around a keyword, ask four questions:

1. 1Would this searcher pay for the thing I sell?
2. 2Are advertisers already paying for this click?
3. 3Can one page satisfy this intent?
4. 4Can I realistically compete with what already ranks?

AhrefsThe fourth question is the one you can't eyeball. Ahrefs answers it with keyword difficulty, who ranks today with how many backlinks, and which keywords competitors quietly earn traffic from. Free Webmaster Tools tier covers a young site.

Four yeses means build it. A no on the first question means the traffic is a vanity metric no matter how big the volume looks.

You don't have to guess what people search: Google will tell you. Type the problem the clumsy way a real user would describe it, then read the titles Google chooses to rank. The pages on page one have already been optimized for the real keyword, so their titles reveal it.

1. 1Search the long, awkward version of the problem, the way a customer would type it
2. 2Read the titles of the pages Google ranks
3. 3Extract the shorter, repeated keyword those titles are built around

Example: searching "create a profile picture from a group of photos with AI for my LinkedIn" surfaces pages titled around "LinkedIn profile picture" and "AI LinkedIn profile picture". That's the keyword to build for, extracted from the SERP itself.

Your competitors have already spent years finding out which pages earn search traffic. Their top pages are a pre-validated keyword list:

1. 1Open a competitor that ranks in your SERPs
2. 2Use Keywords Everywhere or Ahrefs to view their top pages by traffic
3. 3Save the pages with traffic that's relevant to your product
4. 4Expand the keywords each of those pages ranks for
5. 5Group the keywords by URL, not individually: one page's keyword set is one content brief

Then let AI do the clustering. Export the data and prompt: "Here are competitor top pages and their ranking keywords. Cluster these by search intent. For each cluster return: recommended page slug, main keyword, related keywords, search intent, why this page is worth building, and whether it should be a landing page, blog post, tool, listicle, or social play."

Inside Ahrefs, lean on Top Pages instead of Organic Keywords. A single page can rank for hundreds of keyword variants, so the keyword view makes you think you need 500 pages when the top-pages view shows the real content structure: one page per search intent cluster. Settings that keep the data honest:

1. 1Set keyword data to United States for English or global SaaS markets
2. 2Use "Best links" when checking backlinks
3. 3Read Top Pages first, Organic Keywords only to expand a page you've already picked

Columns worth reading, left to right: Traffic, Traffic value, Top keyword, Position, Keyword difficulty, SERP title, and URL. Everything else is noise at this stage.

Before building a page for a keyword, study the page-one results Google already chose. For each keyword on your list, write down:

1. 1What page types rank: tool, landing page, blog post, listicle, Reddit thread, YouTube, app store, directory?
2. 2Do the titles use exact-match keywords?
3. 3Do results use modifiers like free, best, 2026, review, comparison?
4. 4Are low-DR pages ranking? If yes, the keyword may be beatable
5. 5Are Reddit, Quora or LinkedIn posts ranking? If yes, parasite/social SEO is part of the play
6. 6Are videos or images showing? If yes, consider YouTube or image assets
7. 7What sections repeat across the ranking pages?
8. 8What is missing that you can do better?

The SERP is the brief. Don't decide the content type first, and don't reuse the same playbook across keywords: one keyword wants a free tool, its neighbor wants guidance. Let Google show you the format, then make the best version of it.

Once the SERP has told you what to build, the on-page work is mechanical. Every page you publish should clear this checklist:

Steal the pattern, then follow the rule. For "profile picture maker", nearly every top result uses "free", the exact keyword in the title, an upload-first tool UX, a how-it-works section, examples, and an FAQ: that's the template to beat. And if every top result puts the exact keyword in the title, don't get clever. Use the keyword.

Search Console is Google telling you, for free, exactly how it sees your site. Set it up the same day as the keywords, then check in monthly:

1. 1Add your domain as a property at search.google.com/search-console and verify it (DNS record for the whole domain, or the HTML tag for a quick start)
2. 2Submit your sitemap.xml and confirm robots.txt isn't blocking anything public
3. 3Monthly: find pages with high impressions but low clicks, and rewrite their title and meta description
4. 4Monthly: find queries ranking on page 2 or 3, and strengthen those pages with content and internal links
5. 5Monthly: find queries landing on the wrong page, and build a focused page for them

Those three monthly checks are the entire junior-SEO job description. Search Console data lags a couple of days and new sites show almost nothing for weeks: that's normal, keep publishing.

Ahrefs' Site Audit finds the technical problems; an AI code editor fixes them. Set it up once: add your project to Ahrefs, connect Search Console if you're on the free Webmaster Tools tier, enable JavaScript rendering unless the site is fully server-rendered, and let it crawl weekly or monthly. Then every crawl becomes the same fix loop:

1. 1Open an issue in the audit report
2. 2Export the affected URLs as CSV
3. 3Screenshot the Ahrefs explanation of the issue
4. 4Drop both into Cursor or Claude Code
5. 5Ask it to identify the cause in your codebase and patch it
6. 6Re-run the crawl (or wait for the next scheduled one) to confirm the fix

Not worth your time: cosmetic warnings, low-value private URLs, and chasing a perfect 100 health score. The audit is a to-do list, not a report card.

Your competitors' unhappy customers search for alternatives every day. A fair, specific comparison page against each serious competitor is how they find you, and it converts at buyer intent, not reader intent.

"Best [category] for [use case]" keywords are some of the most valuable on the internet: run the volume x CPC math on one and it often clears tens of thousands of dollars a month in paid-search value. The page type that wins them is the listicle, and it works when three things are true:

Make the list helpful first. Rank the tools honestly, give each one a real verdict, and add your product only where it genuinely fits the search intent. A listicle that's obviously an ad for slot #1 loses the trust that made the format rank in the first place.

Stop thinking in pages and start thinking in systems: search demand follows patterns, and one template can capture every variation of a pattern. The four page types that do the work, in rough order of intent:

1. 1Best pages: "best [category] for [use case]". Buyers at the decision stage; make a clear #1 pick instead of listing everything.
2. 2"[X] vs [Y]" and "[X] alternatives" comparisons. Include a "who this is for" verdict: that's the part AI answers quote.
3. 3Use-case pages: "[product] for [audience]". Same structure, swap the persona; relevance is what converts.
4. 4Integration pages: "[your tool] + [other tool]". Narrow queries, near-zero competition.

How We Generate Millions Of Organic Clicks With Programmatic SEODeno Hawari's full playbook: the four page patterns and how to scale them.

This is the system behind teams reporting 692k clicks in 4 months and $1M+ in search revenue: structure written once, content produced at scale.

Backlinks are still the tiebreaker when two pages are equally good, and your competitors have already done the prospecting for you. Every site that links to them has proven it links to products like yours:

1. 1Open each competitor's backlink profile in Ahrefs
2. 2Keep the sites that link to pages like the ones you have (or plan to build)
3. 3Remove spam, directories with no editorial standards, and irrelevant sites

The output isn't a thousand-row export. It's a short list of sites, each with a one-line reason they'd plausibly link to you: they covered your competitor's comparison page, they maintain a tools roundup you belong in, they wrote about the problem you solve. No reason, no outreach.

Backlink profiles show who linked in the past. The SERP shows who is publishing right now. Search the keywords you're targeting and read page one as a prospect list instead of a competitor list. Four page types are worth an email:

Email the person who actually wrote the piece, not a generic inbox, and ask to be included in the list. Lead with why your page helps their reader: the option their roundup is missing, the competitor entry that's gone stale, the screenshot that answers the question their article raises. Asking for a link with no context is a delete.

When anyone can generate a competent article in one prompt, competent articles stop being an asset. The pages that hold their position have something behind them that can't be generated:

The test is one sentence: if a competitor can produce the same page with one prompt, it isn't a moat, it's filler with your logo on it.

AI-drafted pages all read the same because they're built from the same training data. The fix costs two minutes: before generating, record a voice note explaining the topic in your own words and feed it in with your screenshots. Include:

Then prompt: "Use this voice note and screenshots to rewrite the blog post. Keep my opinions and examples. Make it useful, not generic. Add sections where the screenshots prove the point. Avoid em dashes and generic AI phrasing. Add internal links to the relevant product pages." AI is useful for structure, but the moat is your proof of work: screenshots, opinions, examples and real product use make the page hard to copy.

A veteran SEO who managed local search for 14 years now runs the whole function through an 8-prompt Claude system: competitor audits, gap analysis, keyword-rich content calendars, all output as prioritized spreadsheets sorted by impact. The lesson transfers straight to SaaS: give the AI your real data (your site, your competitors, your keywords) once, then run repeatable audit prompts instead of paying an agency to do the gathering. The thinking stays yours; the 30 hours of manual work a month doesn't.

My chief of SEO, Claude CoworkThe full 8-prompt audit system, with the exact prompts and the reasoning behind each.7 Claude SEO promptsThe thread version: competitor audits and gap analysis prompts to steal.