How to SEO: Seven Moves, and the Order They Must Run In
If you want to know how to SEO, start by accepting that sequence decides the result. Technical indexability comes before intent, intent comes before depth, depth comes before links, and none of it matters if you never re-measure. Run the same seven tasks in a different order and you can spend six months producing work that has to be redone from the foundation.
The rail at a glance
Each node removes a reason the next node would fail. That is why the rail is a rail and not a checklist.
What Is the Short Answer to How to SEO?
Fix indexability first. Then match each page to the intent people actually search with. Then deepen the content until the question is fully answered. Then wire internal links. Then earn outside references. Then improve click-through. Then re-measure. Every step removes a reason the next one would fail, so the order is not a preference.
Why Order Beats Effort
Most stalled SEO programmes are not lazy. They are busy in the wrong direction. A team writes twenty articles before the site has a clean crawl, then discovers half the URLs are blocked by a staging rule and the articles sit in limbo. Another team buys links to a category page whose content does not match the query, so the links lift a page that never converts. The work was real in both cases. The sequence was not.
Think of the rail as a chain of preconditions. Content depth cannot rescue a page that a crawler cannot reach. External links cannot rescue a page that answers the wrong question. Click-through improvements cannot rescue traffic that was never relevant. Each stage assumes the one before it holds. Skip a stage and you are not saving time, you are deferring the same cost with interest.
There is a second reason the order matters: cost. Steps one to four are mostly internal and get cheaper as your process matures. Steps five and six cost money and negotiation. Doing the free steps first means you spend on authority only where it will actually move something, which is a materially different budget conversation.
The Seven Steps, One by One
Each step below lists what to do, what it produces, and what breaks if you run it late. Treat the output line as the exit condition. If you cannot name the artefact, the step is not finished.
Indexability week 1
Crawl the site as a search engine would, not as a browser. Confirm every page you care about returns a 200, appears in the XML sitemap, has a self-referencing canonical, and is not blocked by robots rules, noindex headers or a JavaScript render that fails. Check that pagination, filters and search-result URLs are handled deliberately rather than accidentally indexed.
Then look at speed and stability: server response, layout shift, and whether the page renders its main content without a script error. This step is unglamorous and it is the one that silently caps everything downstream.
If you skip it: content and links get spent on pages nobody can reach.
Intent match week 2 to 4
Group the queries you care about by what the searcher is trying to do: learn, compare, choose, buy, or fix. Then assign exactly one primary intent to one URL. Two pages chasing one query cannibalise each other; one page chasing four intentions satisfies none.
Read the current results for each query before writing anything. The format of what already ranks is the strongest signal about the format expected. If the first page is a comparison table and you publish an essay, you are answering a different question than the one asked.
If you skip it: you write accurate pages that rank for nothing, because they answer a question nobody asked in that shape.
Content depth week 3 to 8
Depth is not length. Depth is the number of follow-up questions a reader would ask that your page already answers. For a product page that means specifications, tolerances, compatibility, lead times in ranges, and what happens when something goes wrong. For an explainer it means the definition, the exception, the worked example and the limit.
Write the page so a competent person could act on it without emailing you. That is also the page that earns links in step five without asking, which is why depth comes before outreach.
If you skip it: you win impressions, lose clicks, and train the search engine to stop showing you.
Internal links week 4 to 9
Internal links are how a crawler learns which page is the authority for a topic, and how a reader moves from a question to a purchase. Build them deliberately: from high-traffic explainers to commercial pages, from category pages to the specific product, and from every new article back to its parent hub.
Use descriptive anchor text that names the destination. A link labelled "click here" transfers nothing but a click.
If you skip it: new pages stay orphans and the site's own authority never reaches them.
Authority month 2 onward
Now that you have something worth citing, go and earn references. Not bulk directory submissions. Specific, relevant pages that would genuinely link to the thing you made: a data set, a calculator, a comparison nobody else published, a specification table that solves a real problem.
Outreach works when the ask is small and the asset is obvious. If you cannot explain in one sentence why the recipient's reader benefits, the link will not happen.
If you skip it: a new domain competes on content alone, which works slowly and unevenly.
Click-through month 2 onward
Once you hold a position, the remaining lever is whether people choose you over the nine neighbours. That means the title, the snippet and the proof. Titles should name the outcome, not the keyword twice. Snippets should promise something the page delivers in the first screen.
Improving click-through on a page that already ranks is usually the cheapest available gain, and it is the step almost everyone leaves last because it feels cosmetic.
If you skip it: you pay for positions and collect a fraction of the visits they should return.
Re-measure every quarter
Compare non-branded impressions, clicks and qualified visits, not total traffic. Mark the date of every change so you can tell correlation from coincidence. Kill what did not work, and write down why, because the same idea will be proposed again in eight months.
Then loop back to step one. Sites drift: a plugin update, a new template, a filter rule. The rail is a cycle, not a finish line.
If you skip it: the programme becomes folklore, and nobody can say what works.
What Each Step Should Produce
If a step has no artefact, it is a meeting rather than a step. This table is the version we hold ourselves to on client work.
| Step | Artefact | Done when |
|---|---|---|
| 1 Indexability | Crawl diff and redirect map | Priority URLs all return 200 with a self canonical |
| 2 Intent | Intent map, one row per URL | No query is claimed by two pages |
| 3 Depth | Published page plus gap list | A reader can act without emailing you |
| 4 Internal links | Link plan and orphan list | Every priority page has descriptive inbound links |
| 5 Authority | Scored target list and log | Asset exists and outreach has a record |
| 6 Click-through | Rewritten titles and descriptions | Top impression pages are rewritten and dated |
| 7 Re-measure | Quarterly note | Changes dated, outcomes named, next test chosen |
Notice that only two of the seven depend on buying anything. The rest depend on decisions, which is why a small team that sequences well outperforms a large team that does not.
Three Ordering Mistakes That Cost Months
These are the sequences we are asked to rescue most often. Each one looks reasonable at the start, which is why it gets approved.
Authority spent on a page that answers the wrong query. The links work, the rankings arrive, the page does not convert, and everyone concludes SEO does not pay.
A large batch published onto a site with an indexing problem. The articles are fine and invisible. Fixing the crawl six weeks later wastes the launch window entirely.
The whole site rewritten in one push with no baseline. When traffic moves, nobody can attribute the change, so the next decision is a guess again.
A fourth mistake is subtler: doing steps one to four perfectly and never starting five. That produces a technically clean site that ranks for long-tail questions and never competes for the head terms where the money is. The rail has to reach the end.
How Long the Rail Takes
Steps one and two are usually weeks, not months, for a site under a few thousand URLs. Depth is the slow one, because it depends on subject-matter time rather than effort. Authority is the slowest to show, and click-through is the fastest to prove.
What matters is not the calendar but the overlap. Step three can start while step two finishes; step six can begin as soon as a page holds a position. The only hard rule is that step one is not skipped and step five is not first.
Who Does What
- A developer owns crawlability, canonical logic and redirects.
- An editor owns intent, depth and internal links.
- Subject-matter experts own accuracy, which no writer can invent.
- A marketing lead owns authority targets and the measurement note.
If one person is expected to hold all four, one of them will quietly stop happening. Our Whats Is SEO? page covers the editorial half of that split in more detail.
A Worked Example
An industrial supplier had three hundred pages and almost no non-branded traffic. The instinct was to commission sixty articles. Instead we ran the rail. The crawl showed that a filter parameter had generated thousands of near-duplicate URLs, which had consumed most of the crawl budget. Fixing that took nine days. The intent map then showed that only forty of the three hundred pages matched a real query, and that eleven queries had no page at all.
Those eleven pages were written properly, each answering the follow-up questions the sales team fielded on calls. Internal links moved from the four existing explainers into the product range. Only after that did we approach trade directories and two industry publications with a specification table worth citing. The site did not triple its traffic. It did start producing enquiries from queries that had previously gone straight to a competitor, which is the outcome that was asked for.
Same budget as sixty articles. Different order. That is the whole argument.
Where are you on the rail?
Send us the URL and one line about what is not working. We will tell you which step is missing, in plain language, before you spend anything.
If the terminology is still unfamiliar, start with On-Page SEO Services and come back to the rail with the basics in place.
Questions About Getting the Order Right
Can I run the steps in parallel?
Yes, with one constraint: step one must be finished or at least diagnosed before you publish at volume. Steps three and four usually run together, and step six can start as soon as a page holds a position.
How do I know step one is done?
When every priority URL returns a successful status, has a self-referencing canonical, appears once in the sitemap, and renders its main content with scripts blocked. If you cannot demonstrate that with a crawl export, it is not done.
Do we need new content, or can we fix what we have?
Usually fix first. Most sites have pages that rank for nothing because they answer half a question. Deepening an existing page that already has history is faster than publishing a new one that starts at zero.
What if we only have budget for two steps?
Take one and two. Indexability plus intent match is the highest-return pair, and it costs decisions rather than media. Depth without those two is a well-written page in a queue.
How many pages should we add per month?
Fewer than you think, if quality is real. Six to ten substantial pages a month is a working pace for most teams, provided each one is linked and each one answers a query from the intent map.
When does the rail restart?
Every quarter, at step one. Crawl the site again, because templates change, and check that the pages you built last quarter are still reachable. Compounding only continues if the base holds.