Two different jobs.

Technical — so Google can read the site

• Every page loads its full content from the server, without depending on JavaScript. If the content only appears after code runs, many crawlers never see it.
• Sitemap and crawl file declared and kept up to date.
• Load speed: static sites, images in modern formats with declared dimensions, preloaded fonts.
• Clean, permanent addresses. One page, one address. 301 redirects when something changes.
• Security certificate and a single version of the domain: with www or without it, never both.

Content — so Google understands what you sell

• A distinct title and description for every page, written for the person about to click, not for the search engine.
• One main heading per page, and secondary headings that describe what follows.
• Real text about every product and service. Not three adjectives: what it is, who it is for, what it includes, what it costs and how it is different.
• Descriptive image filenames and alt text.

Local — so Google knows where you are

• A complete Google Business Profile: correct category, hours, photos, services and attributes.
• Name, address and phone number identical — character by character — on the site, on Google and in every directory where the business appears. A single inconsistency splits the signal.
• Local business structured data with coordinates, hours and price range.
• Reviews: a system for requesting them consistently and answering every one.

Authority — so Google trusts you

• Mentions and links from real local sites: chambers of commerce, industry directories, regional press, suppliers and partners.
• Content other people want to cite because it says something nobody else says.

Open crawling

Assistants use their own crawlers: OpenAI's GPTBot and OAI-SearchBot, Anthropic's ClaudeBot, PerplexityBot, Google-Extended and Bingbot for Copilot. If the crawl file blocks them, the business is out by its own decision.

Sarape Web Studio checks that they are allowed and explains what each one implies, because it is your decision: allowing the crawl also means your content can be used to train models.

Complete structured data

It is the difference between a model interpreting your page and a model reading it. Sarape Web Studio implements it in JSON-LD, with the schemas that match your type of business.

Every schema is validated before publishing. The aggregate rating is only marked when there are real, visible reviews on the page itself: inventing it or importing it from elsewhere is grounds for a penalty.

Content shaped as an answer

Models extract fragments, not pages. The heading is the question, phrased the way a person would ask it. The first paragraph is the complete answer in 40 to 60 words, direct and without preamble. The detail comes after.

And the rule most often forgotten: every section has to make sense on its own, torn out of the rest. If a paragraph says "we prepare it fresh every day", the model that extracts it does not know who "we" are or what "it" is. That is why Sarape Web Studio repeats the name of the business and the product inside every block. It reads oddly for someone reading straight through; it reads perfectly for a model reading one piece.

One name everywhere

Models cross-reference sources. If the site says "Restaurante Don Artemio", Google says "Don Artemio Restaurant" and a directory says "Don Artemio Cocina Mexicana", the model sees three weak businesses instead of one strong one.

Sarape Web Studio defines a canonical form of the name, the address and the phone number, corrects it everywhere and links the sources to each other.

Formats that are easy to cite

Models cite more readily what already comes organized: pricing tables, numbered lists of steps, short definitions, side-by-side comparisons, questions and answers.

Wherever the content allows it, Sarape Web Studio presents it in those forms, in real HTML — table, list, heading — never as an image and never as text pretending to be a table.

Signals that the content is alive and yours

Publication and last-updated dates, visible and marked in the structured data. Plus content only that business can write: its own numbers, years in operation, processes, ingredient origins, concrete cases.

A model prefers to cite the source that says something specific over the source that repeats what is generic.