What Is SEO for a Website? A Site Map With a Note on Every Node
What is SEO for a website? It is the work of assigning a specific job to every part of the site — home, category, product, article, and the plumbing underneath — so each page earns its place in search. Not one checklist run once, but a map with a note on each node: this page targets this query, this template needs these fields, this folder must stay crawlable. Get the notes right and the whole site compounds.
Most sites do not fail because nobody did SEO. They fail because the same three tasks were done on the home page and nowhere else.
SEO for a website means matching each URL on the site to one search intent, then removing every obstacle between that URL and the searcher: crawlable links, a clear title and description, content that answers the query, and a template that loads quickly on a phone. Every task on this page belongs to one of those four jobs.
Start with the map, not the task list
A website is not a thousand unique things. It is a handful of node types repeated at scale: a home page, a set of hubs, product or service pages, articles, utility pages, and the technical layer underneath all of them. Whether you run forty product pages or four thousand, the note attached to each node type is the same — which is why you plan against the map, not the page count.
Draw the map first and the work shrinks. You stop asking “what should we do for SEO this month” and start asking “which node is still unassigned”. That second question has an answer, and the answer has an owner.
Every node type takes one note. The note says what the page must do — not how clever it should be.
Node, note, and the cost of skipping it
| Node | The note we write on it | If the note is missing |
|---|---|---|
| Home | State what the company sells in one line, then route to every hub | Visitors land and cannot tell what the site is for |
| Category hub | One intent per hub, 80–120 words above the grid, filters not indexed | Hubs compete with each other for the same phrase |
| Product or service | Unique copy, spec table, schema, links to the guide and to siblings | Pages read as duplicates and get filtered out |
| Article | Answer in the first 100 words, one link to the page it supports | Sessions arrive and convert nothing |
| Utility pages | Keep crawlable, noindex where thin, link from the footer | Trust signals vanish at the moment of decision |
| Technical layer | Sitemap, robots, canonicals, HTTPS, speed, no broken internal links | Nothing else on this table is ever seen |
Utility pages are the ones teams delete in a redesign and then wonder why quote requests fall. They are not there to rank. They are there because a buyer who cannot find a phone number or a policy page does not fill in the form.
What a usable node note contains
A note that says “improve the category page” is not a note. It is a wish. Four lines turn it into work that somebody can finish on a Tuesday afternoon.
One line of intent
Write the query in the searcher’s words, not the company’s. “Industrial air filter supplier” and “replacement filter for a spray booth” may point at the same page, but only one of them is what the buyer types at the start. Pick one intent per node, write it at the top of the note, and let the neighbouring pages take the variants.
One line of template instruction
Something a developer can act on without interpretation: add a spec table, move the H1 above the grid, cut the intro to ninety words, add product schema. If two people could read the line and build two different things, rewrite the line.
One line of links
Every node needs links in and links out. The article links to the product. The product links to the guide and to two siblings. The hub links to its children and back to the parent. Links are what turn a list of pages into a map.
One owner and one date
A note with no name on it is a suggestion. A note with a name and a date is a commitment, and it is the only way to tell, three months later, whether the map is being followed or merely admired.
If the vocabulary is still unfamiliar — title tag, canonical, schema — read Whats Is SEO? first. Every note above assumes you know what those three words are doing.
Priority: three tiers, and the order is not negotiable
Priority here is not about effort. It is about dependency. Tier one unblocks, tier two earns, tier three compounds. Work them out of order and you spend the quarter polishing pages that search engines cannot reach.
Unblock
Crawlability, indexation, canonicals, sitemap, HTTPS, page speed, broken internal links, duplicate templates. While a URL cannot be found or crawled, no rewrite above it has any effect at all.
Rewrite
Titles, H1s, the first hundred words, internal links, schema, spec tables, image alt text. This is where a page stops describing a company and starts matching an intent.
Compound
New pages built on the proven pattern, links from other sites, seasonal refreshes, and retiring pages that no longer earn their place. Slow, cumulative, and the reason tiers one and two were worth doing.
On a mid-sized site, tier one is a finite job measured in weeks. Tier two is a template job: fix one pattern, ship it across every page that uses it. Tier three never ends, which is exactly why it must not start first.
Four ways a site map goes flat
Every page targets the same phrase
When the home page, three categories and four articles all chase the head term, they dilute each other and the strongest page rarely wins. One phrase per node, and the nodes below it take the long-tail versions of the same question.
The note has no owner
Notes get discussed in a meeting and die in a shared document. Attach a name, a deadline and a place where the work is tracked, or the map becomes decoration with better typography.
Filters and pagination get indexed
A store with colour, size and sort combinations can generate more URLs than it has products. Canonicalise, block the parameters, and keep the crawl budget for pages that sell something. A site map that ignores parameters is a map with a swamp in the middle.
Content is added, never revisited
An article from two years ago that still ranks but describes a discontinued product is worse than no article: it walks buyers into a dead end. Put a review date on every node and honour it.
How we work a site map
We list node types rather than pages, because that is the level at which decisions repeat. Then we write the notes, clear tier one, and ship tier two as templates. The rewriting, schema and internal linking work sits inside On-Page SEO Services; the crawl and index fixes come before any of it.
Every template, every section, every parameter. One page of output, not a two-hundred-row spreadsheet nobody opens.
Four lines per node: intent, template instruction, links, owner. Reviewed with whoever has to build it.
Crawl, index, canonical, speed. Measured before and after, with the fix list attached to the audit.
One pattern change, applied everywhere it appears. Then tier three on a monthly cadence.
Send us the site. We will send back the map.
A short note about what you sell, how many templates you run, and what has already been tried is enough to start. You will get a node list with a note on each one, and a priority order you can hand to a developer.
What changes when every node has a job
The first change is not traffic. It is that meetings get shorter. When every page has a stated job, a new product or article has a place to go before anyone writes it, and the argument about whether the blog “counts” disappears: it either links to a page that sells or it does not.
The second change is that reporting becomes readable. You stop measuring SEO as a single number and start measuring nodes — which hubs gained impressions, which product pages converted, which articles sent sessions to a page that sells. In our experience that is also when budgets get easier to defend, because the map shows exactly what the money bought.
Questions we get about SEO for a website
Is SEO for a website different from SEO for an online shop?
The nodes differ, the logic does not. A shop adds filter and pagination problems; a service site adds location and proof problems. Both need one intent per URL, working internal links, and a technical layer that lets crawlers through.
How many pages do we need before this is worth doing?
Fewer than most people think. Ten pages can carry a full map if they are distinct and linked. What kills a small site is not the page count but five pages saying the same thing in slightly different words.
Do we need new content or better pages?
Usually better pages first. Rewriting an existing URL that already has links and history is faster than publishing a new one that starts from nothing. Add content after the existing nodes are assigned, not before.
Who writes the notes?
Whoever owns the site commercially should agree the intent line; whoever builds it should agree the template line. We write the first draft and run it past both, because a note nobody signed off on is a note nobody follows.
How long before the map shows up in results?
Tier one fixes can change what is indexed within weeks. Rewrites typically need a quarter to settle, and new pages need longer. Anyone quoting a fixed date for a specific position is guessing.
Can we run this ourselves?
Yes, if one person owns the map and has two hours a week. The hard part is not the checklist; it is refusing to start tier three while tier one URLs are still uncrawlable.