What the visitor is deciding
Almost nobody browses a local business website. They arrived with a need and are deciding whether to contact you or the next result.
That decision is made on a small number of signals, and they are not the ones most redesigns focus on. Does this look like a business that still exists. Is there a phone number and will someone answer it. How much is this going to cost, roughly. Where are they and can I get there.
A site organised around those questions outperforms one organised around the company's internal structure, which is how most are built. The About page is rarely the reason anyone calls.
For businesses selling across the border there is one more: does this look like an operation a foreign buyer can trust. Clear pricing, real photographs of the actual premises, named people, and a site that does not look like it was assembled from a template in 2016.
Mobile first, on mobile data
The majority of local traffic arrives on a phone, often on a mobile connection rather than wifi, and often in a hurry.
That makes weight a business problem rather than an engineering preference. The most common cause of a slow local business site is uncompressed images uploaded straight from a camera or a phone, and it is also the easiest thing on this page to fix.
The phone number should be a tappable link in the header on every page. Making someone scroll to a contact page to find a number is friction placed exactly where you least want it.
Forms should be short. Every field you add is a reason to abandon, and for a first contact you rarely need more than a name, a way to reach them, and what they want.
Bilingual done properly
A translate widget bolted onto a Spanish site is not a bilingual site. Search engines cannot index what a script generates on the fly, so the English version effectively does not exist to them.
A real bilingual site has separate, indexable URLs per language, hreflang tags declaring the relationship, and copy written rather than machine-translated, because the English terms your buyers search are not literal translations of the Spanish ones.
It also means deciding what happens on first visit. Guessing the language from the browser and redirecting is convenient and occasionally wrong; the safer pattern is to serve a default and make switching obvious and one click away.
Not every Tijuana business needs this. Check your analytics for browser language and California traffic first. Building and maintaining two languages for an audience that is entirely local is work that could have gone somewhere else.
A site you can measure
A website with no measurement is an expense you cannot evaluate, and it is the norm rather than the exception.
At minimum: analytics installed and actually configured, conversion events on the actions that matter, and call tracking if the phone is how customers reach you, which for most local businesses it is.
That last one is where most local sites go dark. The site generates calls, the calls are not attributed, and six months later nobody can say whether the site paid for itself.
Build it in from the start. Retrofitting measurement onto a finished site is more expensive and usually gets postponed until it is never done.
