- Decide the page structure before the visual design. Retrofitting pages into a finished design is where budgets go.
- One page per intent. A page that tries to serve three queries usually ranks for none.
- Give the designer a content outline, not lorem ipsum. Design around real text or the text gets cut to fit.
- Set a performance budget in the brief, not after launch.
A new site is the cheapest moment to get SEO right, and the moment it is most often skipped. This is the sequence that produces a site which ranks without a rescue project six months later.
Step 1: the page inventory comes before the wireframes
Start from the keyword research, grouped into clusters, and turn each cluster into exactly one page. That list - usually 15 to 40 pages for a service business in Madrid - is the site. Everything after that is a decision about how those pages look and connect, not about what exists.
Sites designed the other way round end up with beautiful homepages and no page that answers a specific commercial query, which is the single most common reason a new site loses traffic against the old one.
Step 2: information architecture
| Decision | Made by | Cost of getting it wrong later |
|---|---|---|
| Which pages exist at all | Search demand and the sales process | A rebuild, or pages nobody can find |
| URL structure | The inventory, before design | Redirects forever, and lost equity at each move |
| Navigation labels | The words customers use | Traffic that arrives and leaves |
| What is a page and what is a section | Whether it has its own search demand | Either thin pages or pages doing three jobs |
| Where the enquiry lives | The sales process | Traffic that converts nowhere |
Group the pages into a hierarchy no deeper than three levels: home, service or district hub, then the individual pages. Every page should be reachable in three clicks. Decide the URL pattern now and freeze it - /servicios/reformas-integrales/ rather than /page-id-42 - because changing it later costs redirects and authority.
Write the internal linking rules at this stage too: which pages link to which, and with what anchor text. This is easier as a diagram before the site exists than as an audit afterwards.
Step 3: brief the designer with content, not placeholders
Give each template a real outline: the H1, the sections in order, roughly how many words each needs, where the proof and the calls to action go. Designers asked to lay out lorem ipsum will produce layouts that cannot hold the content, and the content then gets trimmed to fit the design. That trimming is where rankings are lost.
Specify the components you need: FAQ blocks, comparison tables, testimonial units, a district module. Building them once in the design system is far cheaper than improvising them per page later.
Step 4: constraints that belong in the brief
Written into the brief, they cost nothing. Retrofitted after the design is signed off, they are the expensive part of most rebuilds:
Headings are structure, not a type scale. A designer who needs a big line can have a big H2.
Including the hero. If it matters enough to be the first thing read, it matters enough to be indexable.
A template that breaks past four hundred words will break on the pages that earn the most.
Layout shift and slow templates are design decisions, made by whoever exported the assets.
Not only the contact page. Most enquiries start on a service page.
- Performance budget. A page weight ceiling and a limit on third-party scripts, agreed before development.
- Headings. Used for hierarchy, styled independently, so nobody picks an H3 because it looks right.
- Text in HTML. No content baked into images or loaded only on interaction.
- Rendering. If the build is JavaScript-heavy, server-side rendering for content pages.
- CMS access. Whoever writes must be able to edit titles, meta descriptions, URLs and internal links without a developer.
Step 5: launch discipline
The launch checklist that prevents the drop everybody dreads:
One hop to the closest equivalent page. Never a bulk redirect to the homepage.
The single most common launch disaster, and the easiest to check.
A whole template moving to not-indexed is visible within days if anyone is looking.
You will need the before picture to know whether the launch cost you anything.
If a site already exists, map every old URL to a new one before development ends, keep the pages that have inbound links, and check the redirects on staging. Then verify indexing, structured data, analytics and Core Web Vitals in the first week, on real pages, on a phone.
Frequently asked questions
Can we design first and add SEO later?
You can, and it typically costs more than doing it in order, because the fixes touch templates rather than text.
How many pages does a new site need?
As many as you have clusters worth targeting and can maintain. Fifteen strong pages beat sixty thin ones, and the sixty are more work forever.
Who should own the URL structure?
Does a redesign always lose traffic at first?
No. A well-mapped launch is often flat or slightly up. The dip people accept as inevitable is usually redirects done on launch day rather than during the build.
Can we keep the old URLs even if they are ugly?
Usually yes, and often you should. Ugly URLs that rank are worth more than pretty ones that have to earn their position again.
Whoever owns SEO, with the developer implementing. It is a permanent decision made under time pressure if it is left to the build.
Planning a new site?
We produce the page inventory, the URL map and the content outlines before the designer starts.
Talk to Madrid SEORelated: SEO services.