On-Page SEO in Nepal: The Complete 2026 Guide

How this guide was built

I wrote this for business owners, marketers & developers working on websites that sell in Nepal. It combines the page-level process I use in audits with official Google Search documentation reviewed on August 17, 2026. Wherever a recommendation is Google's, I link the source. Wherever it is my operating judgment, I say so.

There are two limitations worth stating before the checklist starts. First, I do not have Google Search Console query data for every industry in Nepal. Search volumes & keyword variants in the examples below are therefore examples, not claims about national demand. Validate them in your own Search Console, Keyword Planner, customer messages & sales calls.

Second, there is no universal on-page formula. A clinic in Lalitpur, a trekking company selling to Germany & a software exporter in Kathmandu have different audiences, languages, proof requirements & conversion paths. The method stays stable. The words do not.

What is on-page SEO?

On-page SEO is the work performed on an individual web page to help a visitor complete a task & help search systems understand what that page offers. It covers search intent, content, titles, descriptions, headings, URLs, internal links, images & page-level structured data.

That definition has two halves for a reason. A page that is easy for a crawler to classify but useless to a customer is not well optimized. Neither is a brilliant sales page that Google cannot confidently connect to a relevant search. Good on-page work closes both gaps.

The boundaries overlap. A canonical tag appears in a page's HTML but solves a technical duplication problem. Core Web Vitals affect the page experience, although performance work often belongs to a developer. A location page is on-site content that supports local search. Labels matter less than making sure somebody owns each problem.

One myth can go immediately: on-page SEO is not "30 percent" or "50 percent" of SEO. No credible source can assign a permanent share. If a page is blocked by noindex, technical SEO is the whole problem that day. If it ranks on page two with a title that promises the wrong thing, on-page work may be the fastest opportunity. Diagnose before allocating percentages.

What makes on-page SEO different in Nepal?

The fundamentals do not change at the border. Helpful content, crawlable links & clear page structure work in Nepal for the same reasons they work elsewhere. The inputs change.

The first difference is language. A customer might search in English, formal Nepali, mixed English-Nepali or Romanized Nepali. These are not interchangeable strings. Someone might type dentist in Kathmandu, काठमाडौंमा दाँतको डाक्टर, Kathmandu ma dentist or another natural variation. The useful question is not "which keyword has the most volume?" It is "which audience is this page for, & what wording will make them trust the answer?"

The second difference is geography. "Kathmandu" can describe the city, the district or the wider valley in casual business language. A customer in Bhaktapur may not consider a service in Baluwatar practical, even when both businesses casually say they serve Kathmandu. State actual service areas, travel conditions, delivery limits or appointment options instead of dropping city names into a footer.

The third is evidence. In the local audits I run, I often find addresses written differently on the website, Google Business Profile & directory listings. I also find services described in agency language while customers ask for them in plain English or Nepali. Neither is solved by adding the target phrase another five times. The fix is to make the business details & the offer unambiguous.

Three language forms, three research paths

These examples illustrate forms, not measured demand. Roman Nepali has no single standard spelling, so a tool may split what is functionally one intent across several tiny-looking phrases. Search Console query data, site search, WhatsApp messages, call notes & the language customers use at the counter are often more revealing than a global keyword database.

Start with search intent, not a keyword list

Search intent is the outcome a person is trying to reach. For on-page SEO, it is the decision that controls everything else: the page type, information order, proof, call to action & even whether you should create a page at all.

Take "SEO Nepal." One searcher could be learning what SEO means. Another could be comparing consultants. A student might want training. An agency outside Nepal could be looking for a delivery partner. One page cannot be the best answer to all four without becoming vague.

I check the live result page before choosing a format. If the results are mostly service pages, publishing a 4,000-word history lesson is unlikely to satisfy the same task. If the results are tutorials, a thin sales page will struggle for the opposite reason. The result page is not an instruction to copy competitors. It is evidence of what Google currently believes the query means.

A practical intent brief

Before writing, answer these seven lines in plain language:

1. Audience: exactly who is this for?
2. Situation: what has happened immediately before they search?
3. Task: what do they want to know, compare, buy or visit?
4. Language: which wording & script will feel natural to them?
5. Proof: what would make this answer credible?
6. Next step: what should they be able to do after reading?
7. Distinct value: what can this page say that a generic international article cannot?

If the team cannot agree on those lines, it is too early to choose headings. The disagreement will surface later as a page that mixes audiences & converts none of them.

How to research keywords for Nepal

Keyword research is demand research expressed through search language. It is not the collection of the largest possible spreadsheet.

For an existing website, I start with first-party evidence. Google Search Console shows the queries for which Google already displays the site. Google Business Profile performance can reveal how local customers find the listing. Analytics shows which landing pages contribute to enquiries or sales. Internal site search shows the vocabulary visitors use after they arrive. None of those sources is perfect, but they describe your actual audience rather than an estimated market.

Then I add qualitative evidence: sales-call notes, emails, Messenger & WhatsApp questions, proposal language, support tickets & reviews. If every customer asks whether delivery reaches Butwal but the ecommerce page only says "nationwide," that is an on-page information gap whether or not a keyword tool reports volume.

Finally, I use external tools to expand & compare: Google Trends for relative interest, Keyword Planner for advertising-oriented estimates, autocomplete & related searches for phrasing, and third-party suites for competitive discovery. Google's June 2026 guidance is useful here: third-party tools do not have access to Google's internal ranking data, & their predictions are their own. Treat volume, difficulty, DA & DR as directional measurements, not facts imported from Google.

Build a query universe in four passes

1. Seed the service or product. List what the business actually sells, using customer language rather than internal department names.
2. Add the task. Buy, book, compare, repair, learn, calculate, visit or contact.
3. Add relevant geography. Nepal, city, neighbourhood or service area only where location changes the answer.
4. Repeat in natural language forms. English, Devanagari, Roman Nepali & genuine mixed-language phrasing where the audience uses them.

Do not mechanically translate the English list. A technically correct Nepali phrase may sound like a government circular while customers use an English service noun inside a Nepali sentence. Ask a fluent speaker who understands the industry, then validate against real queries.

Prioritize by business fit, not volume alone

This is not a Google score. It is a prioritization tool. I would rather publish a page scoring 13 with modest estimated volume than a broad 5-point topic whose visitors cannot buy anything.

Map one intent to one primary page

Once the query groups are clear, assign each meaningful intent to a URL. This prevents two common problems: no page owns the topic, or five pages weakly compete for the same one.

A page can rank for many related searches. "On-page SEO Nepal," "on page optimization Nepal" & "Nepali on-page SEO checklist" can reasonably belong to one guide because the visitor expects substantially the same answer. "SEO consultant in Kathmandu" belongs on a service page because the visitor expects scope, credibility & a route to contact, not another definition.

Create a new page when at least one of these changes materially:

• The user's task changes from learning to hiring, buying or visiting.
• The product or service has a distinct offer, price logic or proof set.
• The location changes availability, delivery, regulation, travel or contact details.
• The language audience needs a complete translation, not a few inserted keywords.
• The current page would become confusing if both intents were combined.

When several existing URLs overlap, do not automatically delete the weaker ones. Check their traffic, backlinks, conversions & purpose first. Then choose among consolidation with a redirect, a clearer differentiation, or a canonical when duplication genuinely must remain. A canonical is a signal of preference, not a substitute for fixing a confused information architecture.

Write titles & descriptions for a decision

The title link is often the first line a searcher evaluates. Google can build it from the HTML <title>, visible main title, headings, Open Graph title, other prominent text & link text. That is why consistency matters. If the title tag says "SEO Services Nepal" while the H1 says "Digital Solutions for Tomorrow," the page is asking Google to reconcile two different descriptions.

Title tag formula

A reliable editorial pattern is:

Primary topic or outcome + useful distinction + brand when needed

There is no Google rule requiring 60 characters. Results truncate to available width, which varies by device & characters used. As a working habit, I keep the distinguishing words early, remove repetition & preview the result on mobile. If a useful title needs a few more characters, accuracy beats an arbitrary green bar.

Meta descriptions

A meta description is a candidate summary, not guaranteed result text. Google primarily builds snippets from page content & may use the meta description when it better describes the page for the query. Write one because a specific pitch is better than leaving the option empty, but make sure the opening content can also stand on its own.

A useful description tells the searcher what the page contains, who it is for & what is distinct. Avoid a comma-separated keyword list. Avoid claiming "No. 1" unless you can define & prove the ranking. Avoid putting a price in the snippet if the page does not clearly explain what that price includes.

Build a page people can scan & quote

A heading should tell a reader what the next section answers. It should not exist merely to hold a variation of the primary keyword.

I use one clear visible H1 because it removes ambiguity for readers & assistive technology. Below it, H2s represent the main questions or stages; H3s divide a stage when needed. Google explicitly says there is no magical heading count & heading order is not a ranking trick. Semantic order is still worth maintaining because people navigate with it.

Use definition-first sections

When a heading asks "What is local SEO?", answer it in the first sentence. Do not make the reader cross three paragraphs of scene-setting. Then explain the nuance, give a Nepal-specific example, state a limitation & offer the next action. That structure is useful for humans, featured passages & AI systems because the answer remains understandable outside the full article.

Write to complete the task

Google publishes no minimum or maximum word count for ranking. A 700-word service page can be complete. A 5,000-word guide can still be thin if it repeats the same advice. Length should follow the task.

For a commercial page, completeness often means:

• A plain-language statement of what the service is & who it is for.
• The problems it solves & what it does not solve.
• Scope, process, deliverables & realistic expectations.
• Relevant proof that can be verified.
• Location, availability, language & contact details.
• Answers to objections a real buyer raises before contacting you.
• A clear next step that works on mobile.

For an informational page, completeness means the reader can act without returning to search for the missing middle. Show the steps, explain decisions, name the risks & cite primary sources. Do not pad the introduction to hit a word count someone saw in a content tool.

Demonstrate experience without manufacturing a case study

Specificity is the difference between experience & performance theatre. You can explain what you inspected, which tradeoff you made, what commonly fails & how you verify the work without inventing a result. If you publish client numbers, include the timeframe, baseline, scope & relevant context, with permission. "Traffic increased 300 percent" without those details is decoration.

Google's helpful-content guidance describes experience, expertise, authoritativeness & trust as concepts its systems try to identify. It also says E-E-A-T is not one specific ranking factor. Treat the framework as an editorial trust test: who wrote this, how do they know, why was it created, what evidence supports it & when was it honestly reviewed?

How to handle English, Nepali & Roman Nepali

The safest multilingual strategy is audience-first, not keyword-first.

If a page is for English-speaking buyers, write it in natural English. It can mention a Nepali term when that genuinely helps, but inserting Devanagari phrases into every heading does not create a useful Nepali version. If Nepali-speaking readers need the same page, translate the whole meaningful experience: navigation, body copy, labels, calls to action, forms, confirmation messages & support route.

Mark a Nepali passage with a language attribute so browsers & assistive tools can process it appropriately:

<p lang="ne">नेपालमा अन-पेज SEO भनेको वेबसाइटको प्रत्येक पृष्ठलाई
प्रयोगकर्ता र खोज प्रणालीका लागि स्पष्ट बनाउने काम हो।</p>

नेपालमा अन-पेज SEO भनेको वेबसाइटको प्रत्येक पृष्ठलाई प्रयोगकर्ता र खोज प्रणालीका लागि स्पष्ट बनाउने काम हो।

For transliterated Nepali, a more specific BCP 47 tag can identify Latin script:

<p lang="ne-Latn">Nepal ma on-page SEO bhaneko harek page lai
user ra search system dubai ko lagi spashta banaune kaam ho.</p>

Nepal ma on-page SEO bhaneko harek page lai user ra search system dubai ko lagi spashta banaune kaam ho.

Do not expect the HTML lang attribute to make a page rank for that language by itself. Google says it detects page language algorithmically from visible content. The attribute is still correct HTML & useful to accessibility tools.

Use separate URLs for complete language versions

If you publish equivalent English & Nepali pages, give each a stable URL & connect them with reciprocal hreflang annotations. Each page must list itself & its alternate:

<link rel="alternate" hreflang="en-NP"
      href="https://example.com/en/seo-services">
<link rel="alternate" hreflang="ne-NP"
      href="https://example.com/ne/seo-services">
<link rel="alternate" hreflang="x-default"
      href="https://example.com/seo-services">

Do not point en-NP & ne-NP at two pages whose main content is still English. Do not canonicalize the Nepali page to the English page merely because the topic is equivalent. Each translated page normally self-canonicalizes, while hreflang explains the relationship.

Where Roman Nepali belongs

Roman Nepali often belongs in conversational content, FAQs, community questions or examples when real customers use it. It does not need its own page for every spelling. Search systems can understand many close variants, & repeating kasari, kasri & kasaree in one paragraph makes the page look broken.

Use evidence. If Search Console shows a Roman Nepali query earning impressions for a relevant page, answer that question naturally. If the business receives the same mixed-language question every week, include it. If a keyword tool alone produces an awkward phrase nobody on the team would say, leave it out.

URLs, canonicals & duplicate pages

A useful URL is stable, readable & descriptive enough to set an expectation. Keep it in lowercase, separate words with hyphens & avoid parameters when a permanent content URL will do. A Nepal page does not need every modifier in the slug.

Add a self-referencing canonical to the preferred page. Keep internal links & sitemap entries consistent with that URL. If tracking parameters, print views or faceted filters create duplicates, decide whether to redirect, canonicalize, restrict crawling or let the variants exist based on their purpose. Do not use robots.txt as a canonicalization tool; a blocked URL can still be known without Google seeing the canonical.

Also remember that a canonical is not a guaranteed instruction. Google combines it with redirects, sitemap inclusion, internal linking & content similarity. Conflicting signals create avoidable ambiguity.

Internal links should explain the site

Internal links do three jobs. They help people continue a task, help crawlers discover pages & show how topics relate. That is enough reason to treat them as part of the content rather than a footer afterthought.

Link from a sentence where the destination genuinely adds the next layer. If this guide gives the page-level process, my 60-minute SEO audit checklist explains how I inspect the wider site. If a reader is choosing support rather than doing the work, the SEO consultant service in Kathmandu is a sensible commercial next step. Both anchors describe the destination before the click.

Build a hub-and-spoke cluster

A useful topic cluster has a broad hub that defines the system & focused spokes that solve distinct subproblems. The hub links to each spoke. Each spoke links back to the hub & sideways only where relevant. It is not a ring of exact-match links added to manipulate authority.

For an SEO cluster in Nepal, the eventual structure could be:

• Hub: SEO in Nepal, covering the complete strategy & linking outward.
• Spoke: this guide to on-page SEO in Nepal.
• Spoke: off-page SEO in Nepal, focused on earning authority.
• Spoke: a technical audit guide, focused on crawling, indexing & performance.
• Commercial page: the service for people ready to hire.

There is no ideal number of internal links. A 600-word service page may need four; a major guide may need twenty. Count usefulness, not links. Review orphan pages, broken links, redirect chains & generic anchors such as "click here" during every content update.

Optimize images without writing alt-text spam

Images can explain, reassure & help a page appear in visual search. They can also become its heaviest resource.

Use an image that is genuinely relevant, place it near the text it supports, serve an appropriate size & include width and height so the layout does not jump. Compress it in a modern format when the browser support & build process allow. A hero photograph exported straight from a camera should not make a user on a slower connection download several megabytes to read a service page.

Alt text describes the image's purpose in context. For a photograph of a dentist examining a patient in a Lalitpur clinic, "Dentist examining a patient at the clinic in Lalitpur" may be useful. "Best dentist Nepal Kathmandu Lalitpur dental clinic cheap" is not. Decorative shapes should use an empty alt="" so a screen reader can skip them.

File names can be descriptive, but they are a minor clarity choice, not a place to list keywords. Keep a single image URL consistent when the same asset is reused; unnecessary duplicates make crawling & asset maintenance harder.

On-page SEO for local businesses in Nepal

A local page should answer "Can this business solve my problem here?" A city name in the title is only the start.

Include the official business name, practical location, phone, opening or appointment information, map or directions when useful, services available at that branch, accessibility details & a clear contact action. Keep the core facts aligned with the Google Business Profile. If an address is commonly described using a landmark, include it for humans while keeping a consistent formal version where structured address fields are used.

When a location deserves its own page

If you serve Kathmandu, Lalitpur & Bhaktapur from one office with the same process, one strong valley-wide service page may be more honest than three near-duplicates. If each city has a branch, team, inventory & contact route, separate pages become useful.

Reviews support trust, but no confirmed Google formula says a specific volume, recency or owner reply speed will produce a position. Ask for genuine reviews without incentives that compromise them, respond helpfully & use the feedback to improve the page. Do not copy private customer messages into public testimonials without permission.

How the page changes by industry

A reusable checklist still needs industry judgment. Here is where I would shift the emphasis.

Clinics & healthcare

Put practitioner credentials, service boundaries, appointment details, emergency limitations & review dates where visitors can see them. Avoid diagnosing an individual through generic copy. Medical pages affect safety, so sourcing & accountable authorship matter more than sales adjectives.

Education consultancies

Separate destination, intake, course & service intents instead of putting every country on one page. State which information changes, where official requirements come from & when the page was checked. Do not promise visas or outcomes outside the consultancy's control.

Trekking, hotels & tourism

International visitors need route facts, seasonality, inclusions, exclusions, physical requirements, permits, safety detail, cancellation terms & a clear booking path. A page targeting Nepali domestic travellers may need different language, prices, transport context & assumptions. That can justify a regional or language version when the content is genuinely localized.

Ecommerce

Category pages should help visitors filter & compare, not just display a grid. Product pages need unique factual descriptions, accurate availability, delivery coverage, returns, specifications, clear images & variant handling. Use Product structured data only for details visible on the page. Decide how out-of-stock, filtered & duplicate variant URLs behave before the catalogue grows.

B2B services & software exporters

Lead with the business problem, delivery model, technical capability, communication window, security expectations & evidence. "Based in Nepal" can be relevant, but it is not the value proposition by itself. A buyer in Australia or Dubai needs to understand outcomes, process & risk before they need a paragraph about Kathmandu.

Structured data, AI search & what not to promise

Structured data is a machine-readable description of content that already exists visibly on the page. For an article, that may include the headline, author, dates & representative image. A local business may use the most specific applicable LocalBusiness subtype. A product page may describe a real product, offer & availability.

It does not replace the content. Google's structured-data guidelines say markup must represent the main visible page & must not describe hidden or misleading information. Valid markup makes some pages eligible for supported features. Eligibility is not a guarantee.

Can on-page SEO help with AI citations?

It can make a page eligible, understandable & useful, but it cannot secure a citation. Google's own AI-features guidance says there are no additional technical requirements or special schema needed for AI Overviews or AI Mode. A page must be indexed & eligible to show with a snippet, then the usual SEO fundamentals apply.

For people & answer systems, I make important passages self-contained: a direct answer, the necessary qualification, a concrete example & a source where the fact is external. Original observations, clearly described methods, comparison tables & accountable authorship give another publisher a reason to cite the page. Generic summaries do not.

No schema type can force ChatGPT, Gemini, Perplexity or Google to mention a page. No llms.txt file guarantees visibility either. Build pages worth retrieving, allow the relevant crawlers according to your publishing policy & measure referral traffic where the platform exposes it.

Two 2026 developments are worth tracking rather than reacting to. Google's guidance on AI features, last updated on December 10, 2025, is unambiguous: there is no separate AI checklist, & the same helpful-content & technical fundamentals apply. Separately, Search Console began surfacing a generative AI performance view, but it reached only a subset of properties & reports impressions without the full click & position breakdown available for standard search. Bing opened a comparable AI performance report in public preview in February 2026.

The practical consequence for a Nepali site is measurement humility. If enquiries mention that someone "asked an AI" about you, treat that as qualitative evidence, because the reporting to prove it at scale does not exist yet. Do not rebuild a page around a metric no platform fully publishes.

Mobile experience & Core Web Vitals

A page is not useful if its main content arrives late, moves under the reader's finger or ignores the first tap. This is usually owned under technical SEO or development, but it belongs in a page review because layout, images, embeds & scripts are content decisions too.

The current Core Web Vitals measure loading, interaction & visual stability. The recommended "good" thresholds are:

• Largest Contentful Paint (LCP): 2.5 seconds or less.
• Interaction to Next Paint (INP): 200 milliseconds or less.
• Cumulative Layout Shift (CLS): 0.1 or less.

Evaluate these at the 75th percentile separately for mobile & desktop, using field data when available. Lighthouse is useful in the lab but cannot directly measure INP without real interaction. Search Console's Core Web Vitals report & PageSpeed Insights can reveal field groups; your own real-user monitoring gives the detail needed to diagnose a specific template.

For a Nepal-facing page, I test on a real mid-range phone as well as desktop, throttle the connection, complete the form with one hand & check whether a phone number, map or payment step works. A perfect lab score does not rescue an enquiry form whose district selector is impossible to use.

My page-level optimization workflow

This is the sequence I use because it prevents polishing the wrong page.

1. Establish the baseline

Record the URL, index status, canonical selected by Google, current queries, impressions, clicks, conversions & meaningful internal links. Annotate the date. Without a baseline, every later opinion becomes vulnerable to memory.

2. Define the audience & intent

Write the seven-line intent brief. Inspect the result page in the relevant location & language. Identify what the current page gets wrong: format, promise, depth, local context, proof or next step.

3. Choose the primary URL

Check whether another page already owns the intent. Decide whether to improve, differentiate, consolidate or create. Confirm that internal links, sitemap & canonical signals can support the choice.

4. Build a content inventory before rewriting

Mark what remains accurate, what needs evidence, what is duplicated, what customers still ask & what no longer reflects the offer. Preserve useful equity instead of replacing the page simply because a rewrite feels productive.

5. Rewrite from the answer outward

State the main answer early. Organize sections around the decisions a visitor makes. Add examples, constraints, proof & primary sources. Write the title & description after the page is clear, not before.

6. Add language & local relevance

Check the wording customers use. Add accurate service areas, branch details, delivery conditions or local examples where they affect the decision. Do not translate isolated keywords into an otherwise unrelated page.

7. Connect the page

Add links from relevant established pages, not just links out of the new page. Use anchors that describe the destination. Confirm that breadcrumbs, category pages & the XML sitemap place it in the site graph.

8. Validate the implementation

Inspect the rendered title, canonical, robots directive, headings, image alternatives, mobile layout, links & structured data. Test the call, form, purchase or booking action. A page is not finished when the copy document is approved.

9. Request indexing only when useful

For an important new or substantially changed page, use Search Console's URL Inspection after deployment. A request is not a guarantee or an accelerator you should repeat daily. Google still decides when to crawl & index.

10. Review evidence, not rankings alone

After enough time for recrawling, compare query groups, clicks, qualified sessions & conversions. Google says changes may take from hours to several months & advises waiting at least a few weeks to assess many improvements. Competitive pages need a longer observation window. Record what changed, what did not & the next hypothesis.

The complete on-page SEO checklist

Intent & page mapping

• The page has one named audience & one coherent primary intent.
• The format matches what the searcher needs: guide, service, category, product, tool or location.
• A different existing URL is not already serving the same intent better.
• English, Devanagari, Roman Nepali & mixed-language variants were considered where relevant.
• Keyword-tool estimates were checked against first-party or customer evidence where possible.
• The page maps to a real offer, expertise area or audience need.

Search appearance

• The title is unique, descriptive & concise.
• The distinguishing topic or outcome appears early in the title.
• The title, visible H1, Open Graph title & page content describe the same thing.
• The meta description summarizes this page specifically rather than listing keywords.
• The description does not make a claim the visible page cannot support.
• The representative image is relevant, crawlable & suitable for sharing.

Content & trust

• The opening answers the main question without unnecessary scene-setting.
• Headings describe the questions or stages a reader can scan.
• The page covers the information needed to complete the task, without padding to a word count.
• Examples are concrete & relevant to the intended Nepal audience.
• External factual claims link to primary or authoritative sources where practical.
• First-person claims distinguish observation from measured research.
• Results include a baseline, timeframe, scope & permission when client-specific.
• The author, background & contact route are visible.
• The published & modified dates are accurate rather than changed for cosmetic freshness.
• Grammar, spelling & language feel natural to a fluent reader.

HTML & media

• There is one clear visible H1.
• Heading levels follow a useful document outline.
• Lists are real HTML lists & comparisons use accessible tables where appropriate.
• Every informative image has contextual alt text.
• Decorative images use empty alt text.
• Images have width & height to reduce layout shift.
• Important information exists as text, not only inside an image or video.
• Nepali passages use an appropriate lang attribute.

URLs, links & localization

• The URL is stable, readable & not stuffed.
• The canonical points to the intended preferred URL.
• Internal links use the canonical URL consistently.
• Relevant pages link into this page as well as this page linking out.
• Anchor text describes the destination without repeated manipulation.
• No important internal link is broken, redirected through a chain or blocked.
• Full language alternates use separate URLs & reciprocal hreflang.
• Each language version self-canonicalizes unless there is a specific reason not to.

Local & commercial details

• The business name, location, phone & hours are accurate where relevant.
• The page explains actual service or delivery coverage rather than listing cities.
• Location pages contain distinct local value, proof & contact information.
• The offer, scope, exclusions & next step are clear.
• The primary call to action works on mobile.
• Claims, testimonials & reviews are genuine & permissioned.

Indexing, schema & performance

• The page returns the intended HTTP status & is not accidentally noindex.
• Google can render important content & access required resources.
• Structured data uses the most specific relevant type & matches visible content.
• Article author, date & image fields are accurate where Article markup is used.
• No schema is present solely to promise a ranking or unsupported rich result.
• The page is usable on a real phone & a constrained connection.
• Core Web Vitals are checked in field data when enough data exists.
• Forms, calls, bookings, maps & payments have been completed in a real test.

How to measure whether the work succeeded

Rankings are one diagnostic, not the business result. Measure the chain.

Compare like with like. Segment brand & non-brand queries, mobile & desktop, Nepal & foreign traffic, and informational & commercial pages. A title change can raise CTR while a ranking update changes position at the same time, so do not attribute every movement to your edit.

For low-volume pages, weekly noise can be larger than the signal. Use a longer window, annotate deployments & compare year over year when seasonality matters. A trekking page in an enquiry season should not be judged against an off-season month without context.

The on-page mistakes I would fix first

1. A page that is not indexable. Perfect copy cannot rank behind an accidental noindex, broken canonical or inaccessible render.
2. Intent mismatch. A blog post ranks weakly for a hiring query while the service page says almost nothing.
3. Several pages fighting for one task. Consolidate or differentiate before writing another.
4. Generic titles & H1s. "Welcome" & "Our Solutions" waste the strongest labels on the page.
5. English-only research for a mixed-language audience. This misses how customers actually ask, even when the final page remains English.
6. Location swapping. Fifty city pages with one changed noun are not local expertise.
7. Keyword density targets. Repetition is not completeness & can cross into stuffing.
8. Plugin-score optimization. A green score cannot understand the offer, evidence or searcher's decision.
9. Unsupported superlatives. "Best SEO company in Nepal" is a claim, not proof.
10. Schema theatre. Marking up hidden FAQs, invented ratings or unavailable products creates risk rather than relevance.
11. Publishing without links in. A page that only exists in the XML sitemap is structurally isolated.
12. Changing the date without changing the substance. Cosmetic freshness weakens trust when readers notice stale advice.

A realistic first 30 days

If the site has many issues, do not optimize every page halfway. Work in a sequence that can produce evidence.

Days 1 to 5: export landing-page & query data, crawl the site, identify indexation problems, map commercial intents & choose the five pages closest to revenue.

Days 6 to 12: interview whoever speaks to customers, build English & Nepali query groups, inspect competing results & write an intent brief for each priority page.

Days 13 to 20: fix page mapping, titles, headings, copy, evidence, calls to action, local detail & internal links. Consolidate obvious overlap carefully.

Days 21 to 25: validate canonicals, robots directives, structured data, images, mobile behaviour, forms & Core Web Vitals risks with the developer.

Days 26 to 30: deploy, inspect rendered pages, request indexing for the small set of materially changed URLs, annotate Analytics & create a review date. The month ends with a baseline & a measurement plan, not a promise of first position.

Sources & further reading

I used primary Google documentation wherever the recommendation describes Google Search. The local application & prioritization method are mine.

• Google Search Central: SEO Starter Guide, including content, links, images, title links & common misconceptions.
• Google Search Central: Creating helpful, reliable, people-first content, including E-E-A-T, authorship & the Who, How & Why framework.
• Google Search Central: Influencing title links, including the sources Google may use.
• Google Search Central: Controlling snippets, including meta-description guidance.
• Google Search Central: Canonical URL methods, including canonical signals & best practices.
• Google Search Central: Localized versions & hreflang, including reciprocal-link requirements.
• Google Search Central: Qualifying outbound links, including sponsored, UGC & nofollow annotations.
• Google Search Central: Article structured data, including author, date & image recommendations.
• Google Search Central: General structured data guidelines, including visible-content & no-guarantee requirements.
• Google Search Central: AI features & your website, including eligibility & the absence of special schema requirements, last updated December 10, 2025.
• Google Search Central: Third-party SEO tools & advice, published June 2026.
• Google Search documentation updates: FAQ rich-result deprecation, May & June 2026.
• web.dev: Core Web Vitals definitions, thresholds & measurement.

Related guides

Technical SEO

How I Audit a Website's SEO in 60 Minutes

13 minFeb 2026

Marketing

Digital Marketing in Nepal: The Complete Growth System

42 minAug 2026

SEO & GEO

GEO vs Traditional SEO: How to Rank in Google AI Overviews

11 minMay 2026