How a Kerala SME Website Should Be Structured for Google in 2026

Structure is the ranking work most sites skip

A Kerala SME site usually fails Google for a boring reason. The services live on one dump page. The town the firm actually serves is a line in the footer. Malayalam, if it exists, is a PDF or a Facebook post. The homepage tries to be about, services, blog, and a slider. Then someone buys a “SEO package” that writes ten blogs about festivals and wonders why “industrial fabrication Kochi” still goes to a directory.

This is not a Core Web Vitals essay. Speed matters on public pages, and we already wrote that down. Structure is what you decide before you compress another hero image. Information architecture: which URLs exist, what each one is for, how they link, and which language they speak. Get that wrong and a fast site still has nothing a searcher can land on.

We build and rescue sites for firms that sell a real service from a real place — Kochi, Kozhikode, Thrissur, Kollam, an industrial estate, a shop that also takes WhatsApp orders. The advice below is for that shape. It is not for a nationwide marketplace and not for a logged-in product. If staff will live in the thing all day, you may be briefing a web application, not a website. Do not force one CMS to be both without two definitions of done.

One homepage job

The homepage should answer three questions in the first screen on a phone: what you do, for whom, from where. Not a mission paragraph. Not four rotating banners. A person on a mid-range Android in traffic should see the service and the place without waiting for a slider that loads last. The rest of the homepage can preview two or three services, a short proof, and a way to call or WhatsApp. That is enough.

What the homepage should not be: a blog index, a catalogue of every town you might visit, a second about page, or a gallery that never names the job. Those URLs can exist. They should not compete with the first screen. Google does not need your homepage to rank for every query. It needs a clean path from a query to a page that matches the query.

Write the homepage H2s as the site map, not as poetry. If you cannot point to a service URL from the homepage without a hamburger full of clutter, the IA is already broken. Fix the nav before you commission another banner.

Service pages that can actually rank

One URL per offer you would take money for this month. “Services” as a single scrolling brochure is a print leftover. A searcher who typed a job wants that job: web development, a Tally-facing internal tool, a shop-floor screen — named, with who it is for, what you need from them, and a next step. Related services link to each other. They do not hide behind one menu label.

Each service page needs a unique job. If two pages could swap titles and still read true, you have one page and a duplicate. Kerala SMEs often clone “our services” into thin variations to look bigger. Google is not impressed. Staff who have to update prices in three places will let two of them rot. Rot is a ranking problem.

• Title and H2s that match the work. The phrase people use, not the agency phrase. “Business website” beats “digital excellence”.
• A who-it-is-for paragraph. Manufacturers, dealers, clinics, shops — pick the ones you can name without inventing case studies.
• What you need from the client. Access, a first-slice sentence, who will sit with the build. Honesty converts better than a feature grid.
• Links to proof and the next step. Contact, WhatsApp, a relevant article. Not a popup.

If you sell in more than one language, the service page still has one primary language. A mixed paragraph that switches mid-sentence helps nobody. See Malayalam below. Do not bolt a machine-translated copy under the English and call it localisation.

Location without fake city pages

You have a real address, a real service area, or both. Put the address on contact, on the homepage, and in the same NAP form you use on Google Business Profile — name, address, phone, not a creative variant. If you serve Kochi and also travel to Thrissur, say so on the relevant service pages in a sentence, not as fourteen thin URLs that only change the city name.

Thin city pages are how 2018 SEO agencies padded retainers. In 2026 they still get built. They rarely rank, they confuse the crawler, and they are a maintenance tax. A useful location URL is a page with a reason to exist: a branch, a shop, hours, a map that is not copied onto every URL, a paragraph about what that office actually does. A page whose only unique token is “Calicut” in the H2 is not a location strategy.

Google Business Profile is the location surface most Kerala SMEs under-use. Categories, hours, photos of the real work, WhatsApp as a message action if you truly answer it, posts that are not festival spam. The website should agree with the profile. If the site says Edappally and the profile says an old rented desk, you have taught Google to distrust both.

Do not hide the place because you want to look national. A Kozhikode fabricator who writes like a Bengaluru SaaS company will lose the query that included the district. You can still take work from Chennai. You do not need a Chennai URL to do it. Say “we travel” where it is true. Do not invent offices.

Malayalam and English without a duplicate site

Many Kerala buyers search in English for B2B work and in Malayalam for local services. Some do both in one query. Your structure should pick a primary language per URL and make the other language a real alternative, not a widget that rewrites the same slug into broken copy.

If you have the staff to maintain two versions, use a clear pattern: /services/web-development and a Malayalam equivalent with its own slug, linked as alternates, each a full page. If you do not have the staff, pick English for the commercial pages and put Malayalam where it earns its keep — contact, a short homepage paragraph, a service you actually sell in Malayalam conversations. An unmaintained second language is worse than one honest language.

WhatsApp is often the Malayalam channel already. The site does not have to reproduce every chat. It has to not fight the chat: phone numbers that match, a name that matches the visiting card, a service list the person on the phone can find. When the site says “solutions” and the uncle on the call says “website for the shop”, you have an IA problem, not a translation problem.

Avoid auto-translate as the published page. A crawler will index the mess. A person will bounce. If you need Malayalam this quarter and cannot write it well, write less, write it by a person, and put it on the pages that get the calls.

A site map you can print

That is enough for most SMEs. Add industry pages only when you have a different offer or a different proof, not when you want more URLs. Add a careers page when you are hiring. Add a login only if there is software behind it. Extra top-level items are how the nav becomes a junk drawer and the crawler wastes crawl on thin URLs.

Print this table. If a proposed page is not on it, it needs a job or it does not ship. “We might need it later” is how you get a 2019 WordPress menu with twenty items and three that 404.

What not to put in the main nav

1. Every town you have ever sent a person to.
2. Every blog category. Categories are for the blog index, not the header.
3. PDF rate cards as nav items. Host them if you must; do not make them the IA.
4. English and Malayalam as two complete duplicate trees in one hamburger.
5. Admin, staff login, or “client portal” unless that portal is a product you sell on purpose.

The header on a phone can hold Home, Services (as a short list), Work, Knowledge or Blog, Contact. WhatsApp and Call can be actions, not more IA. Footer can repeat the address, the GSTIN if you publish it, and the legal pages. Legal pages are not ranking plays. Keep them short and findable.

Internal links that match how people search

Service pages should link to the two or three articles that answer objections — cost, website versus application, how you work — and back. Articles should link to the service, not only to more articles. The homepage should not be the only hub. A crawler that can only reach a service page from a slider is a crawler you have handicapped.

Use the words the searcher uses in the link, when they are true. “Web development” is a better anchor than “click here”. Do not stuff the district into every anchor. One clear mention of place on the page is enough. Repeated “best Kochi Kochi Kochi” is the 2014 leftover, and it reads like it.

Orphan URLs — published, in the sitemap, linked from nowhere — are how good articles die. If you publish a Monday series, the series should point at the service and the knowledge hub, and those should point back at the pieces that still earn the click. That is structure. It is also how a small site looks finished.

Sitemaps, URLs, and the mess you inherit

Clean URLs: lowercase, hyphens, no date unless the date is the point, no ?page_id= leftovers if you can 301 them. One host: pick lavistech.com-style apex or www and redirect the other. Trailing slash: pick one. These are dull. They are also how you stop splitting signals across four copies of the same page.

XML sitemap: only URLs you want indexed. No thank-you pages, no admin, no search-result URLs. HTML site map is optional; the printed table above is more useful internally. Robots.txt should not block the CSS and JS the public pages need, and should not allow a staging copy you forgot.

If you migrated from WordPress with default permalinks, Search Console will still show query-string URLs. Redirect them. Do not leave both the old and new URL 200. That is not “more pages”. That is two addresses for one thought.

This is not the vitals lecture

Once the URLs are right, make the money pages fast on a Kerala phone. Images sized, third parties off the first paint, layout that does not jump. That work is Core Web Vitals explained. Do it on the service and contact URLs that already have impressions. Do not start a vitals project to avoid writing the service pages.

A green Lighthouse score on a homepage that does not name the town or the job is a trophy. We would rather ship a plain, crawlable service URL that a person in Thrissur can finish on the bus. Web development here means that order: IA, then pages, then the boring performance work. Theme shopping is not a strategy.

A one-afternoon IA workshop

Write every URL you have today on a board. Mark index / noindex / redirect / delete. Write the table from this article next to it. Merge the dump “Services” page into real service URLs. Kill the fake cities. Pick a primary language. Agree the header. Assign an owner who will update prices and addresses. If nobody will own a URL, do not publish it.

Then look at Search Console queries. The phrases you already get impressions for should have a home. If “tally integration” or “website for workshop” shows up and you only have a homepage paragraph, that is next week’s page — not another blog about Onam. Structure follows demand you already have, plus the two or three jobs you can actually deliver.

Kerala SMEs do not lose Google because a competitor bought better adjectives. They lose because the competitor has a URL that matches the query and a phone number that matches the signboard. You can do that this month without a new brand film. You cannot do it if the site is still one page pretending to be a company.

If the thing you need is software people log into, stop reading this as a website brief and go back to the website-versus-application test. If what you need is a site Google can understand, print the table, cut the nav, and write the service pages like a person who answers the phone in Malayalam and English and is tired of repeating the same three facts.