- SEO is three jobs, not one: making the site accessible, answering the question better than the alternatives, and being credible enough to be trusted.
- Search intent decides page format. Get that wrong and no amount of optimisation rescues the page.
- Technical work is a floor, not a ceiling. It stops you losing; content and authority make you win.
- Measure business outcomes. Rankings explain what happened; they are not what you were paid to produce.
Most SEO advice is either too abstract to act on or a list of tactics with no order. This guide is the working version: what the discipline consists of, the sequence that avoids wasted effort, and the judgement calls that separate sites that grow from sites that plateau.
How search actually works, briefly
Three processes decide whether you appear. Crawling: a bot reaches your pages by following links and reading your sitemap. Indexing: the content is parsed, understood and stored - and not everything crawled gets indexed. Ranking: for a given query, the index is sorted by relevance, quality and context, including the searcher's location and device.
Almost every SEO problem is a failure in one of those three. Diagnose in that order. A page that is not indexed cannot be optimised into ranking, and a page ranking badly for the wrong query does not have a ranking problem, it has an intent problem.
Search intent: the decision that comes before the writing
| Intent | The searcher wants | The page that wins | Commercial value |
|---|---|---|---|
| Informational | To understand something | A clear explanation, well structured | Low, but it builds trust |
| Navigational | A specific site or brand | That brand’s own page | Only for your own brand |
| Commercial | To compare before buying | Comparisons, criteria, honest trade-offs | High - buyers are here |
| Transactional | To buy or enquire now | A service or product page that converts | Highest |
Getting this wrong is the most expensive mistake available: a beautifully written guide will not rank for a query where every result is a service page, however good it is.
Every query carries an expectation about the kind of answer. Informational queries want an explanation. Commercial ones want comparison. Transactional ones want a way to buy or book. Navigational ones want a specific place.
The practical test costs two minutes: search the term and look at what already ranks. If the first page is full of guides, a product page will not break in, and vice versa. Match the format first - guide, comparison, service page, tool - then compete on depth and clarity within that format.
Keyword research without the theatre
- Start from the business, not the tool. List what you sell, the problems it solves and the words your customers use in emails and calls.
- Expand. Use the search box suggestions, the People Also Ask block, and a keyword tool for volume and difficulty. Treat volume as an estimate, not a fact.
- Classify by intent. Each term gets one of the four labels above. This determines what you build.
- Cluster. Group terms that would be answered by the same page. One cluster, one page - splitting a cluster across pages makes both weaker.
- Prioritise. Rank clusters by commercial value first, achievability second, volume last.
On-page: the elements that carry weight
| Element | Weight | The mistake people make |
|---|---|---|
| Title tag | High | Stuffing it, or repeating the brand before the subject |
| H1 and heading order | Medium | Using headings as a type scale rather than structure |
| First hundred words | High | Warming up instead of answering |
| Internal links | High and underused | Linking from nowhere, or always with the same anchor |
| Meta description | Indirect | Treating it as a ranking factor; it is a click factor |
| Image alt text | Low but free | Leaving it empty, or stuffing keywords into it |
| URL | Low | Changing it later, which costs more than it ever gained |
- Title tag. The strongest single on-page signal and the thing people decide on in the results. Descriptive, specific, roughly 50-60 characters.
- Meta description. No direct ranking effect, real click-through effect. Say what the page gives and who it is for.
- Headings. One H1 stating the subject, H2s that map to the questions a reader arrives with. A structure someone can scan in ten seconds.
- URL. Short, readable, stable. Changing URLs later costs more than getting them right now.
- Internal links. The most underused tool in SEO. They distribute authority, and they tell search engines which of your pages matters for which subject.
- Images. Descriptive file names, real alt text, modern formats, correct dimensions.
- Structured data. Article, Product, FAQ, LocalBusiness where they genuinely apply. It does not lift rankings by itself; it changes how your result looks.
Content that holds its position
Length is not a strategy. Pages hold their position when they answer the question completely, are easy to scan, are specific where competitors are vague, and are kept current. A 900-word page that resolves the question beats a 3,000-word page that circles it.
Two habits matter more than any writing technique. The first is refreshing: revisit important pages every few months, update the facts, cut what aged badly, add what readers keep asking. The second is honesty about experience - concrete examples, real numbers, named constraints. It is also what quality raters are told to look for.
Technical SEO: the floor you stand on
Technical work rarely wins a ranking on its own. It removes the reasons you cannot have one, which is why it comes first:
No accidental noindex, no robots.txt block, no login wall, no orphan page.
Real text in the HTML, one canonical, sensible headings, structured data where it applies.
Most Madrid searches happen on mobile data, not on your office fibre.
A checked site that nobody re-checks is a site that breaks quietly.
Speed, mobile behaviour and stability are now table stakes. Aim for a largest contentful paint under 2.5 seconds, layout that does not jump while loading, and interaction that responds immediately. Serve over HTTPS, keep the sitemap accurate, and use canonical tags to resolve duplicates rather than hoping they resolve themselves.
Watch crawl coverage in Search Console: pages excluded, pages discovered but not crawled, redirect chains, soft 404s. These reports are the closest thing to a direct conversation with the search engine about your site.
Links and authority
Produces a profile that either does nothing or must be cleaned up later. The same money buys two pages that compound.
Third-party authority metrics are estimates. A link from a trade body with a low score can be worth ten with a high one.
Broken pages with inbound links are free recovery. Most sites have several and have never looked.
Links remain a primary trust signal, and the market for them is full of ways to waste money. What works is unremarkable: publish something worth citing, tell the people who would cite it, and build relationships in your sector. Digital PR, original data, useful tools and genuine partnerships produce links that survive algorithm updates.
What fails: bought link packages, private blog networks, mass guest posting on sites nobody reads, and reciprocal link schemes. They work briefly and then either stop counting or become a liability you have to clean up.
Measurement
Build the reporting around four questions. Which queries bring us impressions and clicks? Which pages produce enquiries or sales? Where do we appear but get ignored - high impressions, low click-through? And what changed since last month, in the site or in the results page?
Search Console answers the first and third, analytics the second, and a rank tracker plus your own change log answers the fourth. Report on outcomes; keep positions as the diagnostic layer underneath.
“Almost every SEO failure we are asked to diagnose is a sequencing problem. The individual activities were fine; they were simply done in an order that guaranteed a poor return.”
A sequence that avoids wasted work
- Month 1. Technical audit and fixes, analytics and Search Console properly configured, keyword research and clustering, an inventory of existing pages.
- Months 2-3. Fix and consolidate what already exists before writing anything new - most sites have pages competing with each other. Then publish against the highest-value clusters.
- Months 4-6. Content cadence, internal linking pass, first authority work, refresh cycle for the pages that matter.
- Ongoing. Monthly review, quarterly refresh, and one honest question each quarter: what did we stop doing because it did not work?
Frequently asked questions
How long does SEO take?
Technical fixes can show within weeks. New content in a competitive space typically needs three to six months to settle. Authority takes longer and compounds.
Is SEO dead because of AI answers?
Behaviour is shifting, not disappearing. AI answers cite sources, and the sources are the pages that would have ranked anyway: specific, credible and well structured. The work changes shape rather than value.
Should I write for people or for search engines?
For people, structured so machines can follow it. Those two stopped being in conflict some years ago.
What is the single most common mistake?
How many pages does a site need to compete?
Fewer than most people assume. Ten pages that each answer one query properly beat a hundred that half-answer the same five.
Do we need to publish constantly to hold a position?
No. You need to keep the pages that rank accurate and useful. Publishing cadence is a substitute people reach for when nobody is maintaining what already works.
Publishing more pages instead of fixing the ones that exist. Consolidation and refreshing usually produce faster gains than new content.
Find out which of these applies to your site
An audit that names the three things holding you back and the order to fix them in.
Request an auditSee also our SEO services and Local SEO for businesses in Madrid.