Operating principle
How does SEO work? It moves a page through four stages: crawl, understand, trust, then click. A crawler has to reach the URL, a system has to read and file its meaning, an algorithm has to decide it deserves a slot, and a person has to choose it over the nine results around it. Each stage has its own metric and its own way of failing. When a site stalls, it is almost always stuck at a named stage rather than “the algorithm” being unkind.
Answer / one pass
A search engine crawls a URL, parses and stores its meaning, compares it against other pages for a query, and ranks what it believes is most useful. SEO works by removing every obstacle in that chain: access, clarity, credibility, and finally the wording that earns the click.
They run in sequence, and a failure in gear one cannot be fixed in gear four. That is why a rewrite of a blocked page changes nothing at all.
A bot follows links from known pages, or reads your sitemap, and requests the URL. If it is blocked, redirected, or ten clicks from the home page with no internal link, the rest of this page is theoretical.
watchcrawl requests, status codes, orphan URLsThe HTML is parsed into text, headings, links and structured data. A page that needs a script to show its main copy, or has eight competing headings, is harder to file under a topic.
watchrendered text parity, title uniqueness, heading orderThe page is weighed against alternatives: depth on the topic, inbound links, freshness, load speed, and whether real people stay once they arrive. This is the slowest gear to turn and the hardest to fake.
watchaverage position, link growth, core web vitalsA position is only inventory. The title, the date, the price hint and the brand decide whether the slot becomes a visit. A page can rank third and still lose to the result below it.
watchclick-through rate against expected CTRRead the gears right to left when diagnosing. If clicks are weak but positions are fine, the problem is gear four. If positions are weak but coverage is fine, look at gear three. If coverage is weak, stay in gears one and two.
Every stage consumes something and returns something you can measure. If a stage has no output, it did not run.
| Stage | Input you supply | Output the engine returns |
|---|---|---|
| Crawl | Working links, a clean sitemap, fast responses, no accidental blocks. | Logged requests, indexed URLs, and a list of pages it found but chose to skip. |
| Understand | One subject per URL, descriptive title, ordered headings, structured data. | A topic the page is filed under, plus the queries it becomes eligible to appear for. |
| Trust | Proof, depth, citations, internal links from strong pages, and load speed. | An average position for each query, and movement against the pages above you. |
| Click | A title that matches the query, a clear promise, and a familiar brand. | Sessions, scroll depth, inquiries, and the queries that actually convert. |
Five symptoms we hear every week, each mapped to the gear that is actually stuck. Start with the symptom, not with a tool.
The URL is new, the copy is fine, and the index report stays empty for weeks.
Check gear 01: a noindex tag left from staging, a canonical pointing to another page, or no internal link from anywhere on the site. A sitemap entry alone is a suggestion, not a command.
Impressions climb, positions hover between eleven and twenty, and nothing breaks into the top ten.
Check gear 03: the page covers the query but is thinner than the pages above it. Add the sub-questions a buyer asks next, plus one piece of evidence those pages lack.
The page is visible for many queries and earns almost no visits from them.
Check gear 04: the title describes your company rather than the searcher's problem. Rewrite it as the answer they typed, and let the description carry the proof.
Traffic held for months and then fell within a week, without any content change.
Check gears 01 and 02: a template release, a redirect chain, a changed canonical rule, or a lost internal link block. Site-wide drops usually start in code, not in copy.
Sessions are healthy, form fills are flat, and the sales team says the leads are wrong.
Check intent, not volume: the page ranks for research phrases rather than buying phrases. One comparison page or specification table often converts better than a popular article.
A burst of movement, then a flat line while the rest of the site keeps growing.
Check internal links: new pages with no inbound internal links are expensive to keep alive. Add contextual links from pages that already earn traffic.
Six inputs that no agency can supply on your behalf. They are the reason two sites with the same budget produce different results.
After enough audits, the same three jams show up. Fixing them is what most “SEO” actually looks like in practice.
Faceted category pages and tag archives generate thousands of near-identical URLs. Crawlers spend their budget on duplicates while the pages that sell stay half-crawled.
Fix: decide which URL owns each topic, consolidate the rest, and link only to the winner.
The page explains the category beautifully and never states who it is for, what is included, or what to do next.
Fix: one paragraph of specifics near the top, one table of options, one clear next step.
A hero animation and four font files push the largest paint past four seconds on a mid-range phone.
Fix: ship the text first, defer everything else, and measure on a throttled connection rather than on your laptop.
New pages are published and never linked from the pages that already have authority.
Fix: every new URL gets at least three contextual links from relevant existing pages within a week.
Send one URL. We will name the stage, the metric to watch, and the first change that would move it.
The order is deliberate. Skipping ahead to content on a site that cannot be crawled is the most common way budgets disappear.
If you need the vocabulary before the process, start with Whats Is SEO? and bring the questions back here.
Crawl and understand can change within days of a fix, because they are mechanical. Trust takes quarters: links accumulate slowly and comparison against competitors happens on their schedule as well as yours. Click-through can move in a week once a title matches the query.
Yes, but less as a raw count and more as evidence that someone found the page worth citing. A hundred directory links do nothing; five links from pages that cover the same subject can move a page out of page two.
Speed matters most where it changes behavior: a page that takes five seconds on mobile loses visitors before they read anything. Treat performance as a conversion input first and a ranking input second, and fix the largest paint.
Sometimes, when the page is a tool, a calculator, or a product listing that answers the query directly. But the engine still needs to understand what the page is for, and that requires some explicit language on the page.
The number of URLs a crawler is willing to fetch from your site in a period. Large sites waste it on duplicates and parameter URLs. Small sites rarely run out, but they can still waste the visit on pages that should not be indexed.
Results are personalized, localized and refreshed continuously. Daily movement of two or three places is noise. Judge the trend over four weeks, and only act when a page has moved consistently in one direction.
They solve different problems. Ads buy immediate visibility and stop when payment stops. SEO builds an asset that keeps producing after the invoice. Most healthy sites run both and compare cost per inquiry rather than cost per click.
