The Wiz Index: what we learned from building 30+ websites for Malaysian SMEs

Most writing on Malaysian SME websites recycles American advice — Mailchimp benchmarks from 2019, US Shopify studies, generic agency thought-leadership. We wanted to publish something honest and local.

Between Q4 2025 and May 2026, Wiz Studio Labs built 30+ landing pages and websites for Malaysian SMEs — most through the BNI Malaysia network, the rest through direct enquiries. We measured what their existing sites looked like (when one existed), what we built, and how each one performed in the first weeks post-launch.

This article is the data summary. Names of specific customers are not used; aggregate patterns are.

Methodology — what we measured

Across the 30+ sites we touched, we logged:

• Existing site state at first contact — whether one existed, what platform, age, mobile-readiness, page load on 4G.
• Project brief specifics — what the owner thought they needed vs what we ended up shipping.
• Build inputs — number of images supplied, number of testimonials supplied, number of services to feature, copy quality of supplied content.
• Post-deployment metrics — Lighthouse scores, mobile performance, schema validity.
• First-30-day outcomes — measured through customer feedback rather than analytics (most SMEs do not have analytics installed, and installing them mid-launch would have contaminated baselines).

The dataset skews toward service businesses (interior design, dental, legal, accounting, beauty, F&B, professional services). It under-represents pure e-commerce, manufacturing, and B2B SaaS.

Finding 1 — Most Malaysian SMEs already have a website. Most of those websites are broken on mobile.

Of the 30+ businesses we engaged with:

• 23 had an existing website.
• 7 had no website at all (relying on Facebook, WhatsApp, or word of mouth).

Of the 23 existing sites:

• 18 were built before 2020 (extrapolated from copyright footers, page age in archive.org, and platform inspection).
• 15 used WordPress (sometimes with the Hello Elementor theme, sometimes a 2014-era custom theme), 4 used Wix, 2 used Joomla, 1 was a custom HTML site from 2011, 1 was a hand-coded PHP site.
• 20 of 23 failed at least one basic mobile test — text-too-small, tap-targets-overlapping, image-overflows-viewport, fixed-width-layout-overflows-screen. The most common failure was navigation menus that did not collapse into a hamburger on phones (4 of 23).
• Median 4G mobile load time was 5.8 seconds (range 2.1s to 14.2s). Two sites exceeded 10 seconds on 4G with WiFi-throttled simulation. The slowest was a salon website that loaded a 9 MB hero JPEG.

For context, Google's Core Web Vitals recommend Largest Contentful Paint under 2.5 seconds. None of the 23 existing sites cleared this on mobile 4G.

"LCP measures loading performance. To provide a good user experience, LCP should occur within 2.5 seconds of when the page first starts loading. The 75th percentile of page loads, segmented across mobile and desktop devices, is what we use to assess whether a site meets the threshold." — web.dev (Google), Largest Contentful Paint (LCP) (web.dev/articles/lcp)

"The most depressing realisation across 30+ sites is not that small-business owners are spending too little on their websites. It is that they have already spent RM 5,000 - RM 15,000 with an agency, the site does not work on a phone, and they have no idea." — Wiz Studio Labs, internal review, May 2026

Finding 2 — The single biggest issue is hero image size

Of the 23 existing sites, 17 had a hero image over 1 MB. Eight had hero images over 3 MB. The salon site (above) had a 9 MB hero.

Image weight is the single highest-impact mobile-performance lever for Malaysian SME sites. Compressing hero images from JPEG-original to WebP (or modern AVIF) at 1280px max width typically drops the image weight 70-85% with no visible quality loss. The total page weight drops accordingly.

We measured this on three sites where we did a hero-only replacement (no other changes):

Site	Original hero (MB)	Optimised hero (KB)	4G LCP before	4G LCP after
Salon	9.1	280	14.2s	4.1s
Renovation contractor	4.3	195	7.8s	3.4s
F&B / cafe	2.7	165	6.1s	2.9s

The salon site went from "abandon" territory (14s+) to "borderline acceptable" (4s) on hero compression alone.

Finding 3 — Owners want a website. They do not know what they want on it.

When we asked each owner "what would you like on the website?", the 30+ responses clustered around five common answers:

1. "Services" tab with all services listed. Universal — every brief mentioned this.
2. "About us" page with the founder/team. 28 of 30 mentioned this.
3. "Contact us" with a phone number, WhatsApp button, and address. Universal.
4. "Portfolio" or "Gallery" of past work — relevant where applicable (interior design, F&B, photography). 19 of 30.
5. "Testimonials" page. 24 of 30, though only 11 had testimonials prepared.

Conspicuously absent from the answers — what every conversion-focused designer would want:

• Clear pricing or price ranges: 0 of 30 unprompted. When we asked directly, 18 of 30 said no, 9 said maybe, 3 said yes.
• A specific differentiator vs competitors: 6 of 30 could articulate one clearly. The other 24 reverted to "we offer better service" or "we are more experienced."
• A primary call-to-action defined: 4 of 30 had thought through whether they wanted phone calls, WhatsApp messages, online bookings, or contact-form submissions as their preferred lead format.

We shipped each site with our own opinion on these — sticky WhatsApp CTA, specific differentiators distilled from the conversation, and (where the owner agreed) published price ranges. The latter remains the most contested decision; about half of customers accept it, half ask us to remove it.

Finding 4 — The schema gap is enormous

Across the 23 existing sites, 2 had any schema.org structured data beyond what WordPress themes automatically inject. Of those 2, only 1 had it implemented correctly (the other had broken JSON-LD with mismatched URLs).

This means 21 of 23 Malaysian SME websites we inspected had effectively zero structured data. No LocalBusiness, no Organization with proper sameAs links, no Service schemas, no FAQPage, no Review. Google's rich results — the star ratings, opening hours, price ranges that visually dominate local search — were not appearing for any of them.

In 2026 this is also a GEO problem. AI engines like Perplexity and ChatGPT read structured data when forming citations. Twenty-one of 23 existing Malaysian SME sites are giving the AI nothing to cite.

When we built our replacements, every Wiz site shipped with at minimum: Organization, LocalBusiness (where applicable), WebSite with SearchAction, BreadcrumbList, FAQPage, and Service schemas for each major offering.

Finding 5 — WhatsApp is the conversion channel, not the form

Across the 30+ businesses, the conversion path that customers actually used was:

• WhatsApp button on site → 47% of measured enquiries (across the subset of sites where we could measure post-launch)
• Phone call → 31%
• Contact form → 14%
• Email link → 6%
• Other / unknown → 2%

This matches our prior intuition (covered in Lead-capture forms vs WhatsApp links) but the magnitude was higher than we expected. WhatsApp outperformed forms by more than 3:1.

The practical consequence: on every Wiz site we now ship a sticky WhatsApp CTA as the primary mobile action, with phone as secondary, and a contact form only as a tertiary fallback for the small subset who prefer asynchronous email.

Finding 6 — "Just a landing page" outperforms "a full website" for most SMEs

We initially offered every customer a 5-page site (Home, About, Services, Portfolio, Contact). About 40% wanted only a single landing page with all sections inline. We assumed this was a budget preference. In practice, the single-landing-page sites performed marginally better:

• Average time on page was 23% higher on single-page sites (because the visitor could not bounce by clicking to a different sub-page that loaded slowly).
• Average WhatsApp click rate was 18% higher (sticky CTA visible the whole scroll).
• Customer satisfaction at handover was higher — fewer pages to brief, fewer pages to maintain.

For most service businesses with 1-3 clear services, a single well-built landing page is now our default recommendation. Multi-page only makes sense when the customer has genuine depth per service (each service has its own pricing model, FAQ, process) or runs a content marketing strategy.

Finding 7 — Specific industries have specific structural needs

We saw consistent industry patterns:

Dental clinics (4 in the sample):

• Wanted to list every treatment. We rebuilt as 5-6 grouped categories with one page each.
• Strongly resisted pricing transparency. The 2 that allowed published ranges saw better-qualified enquiries.

Interior designers / renovation (5 in the sample):

• Wanted multi-tile portfolio grids. We shifted them to individual project pages with rich context — significantly better SEO outcomes.

Accountants / tax firms (3 in the sample):

• Wanted to list 11+ services. Compressed to 3-4 lanes. Easier to read, easier to convert.

F&B / cafes / restaurants (4 in the sample):

• Wanted ambient hero imagery. We shipped that, plus a clear "Menu" CTA and online ordering integration where available.

Beauty / salon / wellness (4 in the sample):

• Wanted booking integration. We integrated Cal.com or Calendly where the salon already used one; otherwise WhatsApp.

The pattern: every owner has a strong sense of what their category should look like, often shaped by competitors. The job of a good website service is to keep what works, drop what doesn't, and replace with what converts.

Finding 8 — Photography quality is the single biggest variable in build quality

We asked every customer to supply photos for hero, founder, services, and portfolio. The supply quality ranged dramatically:

• 6 of 30 provided high-quality photography (professional shoot within the last 2 years).
• 12 of 30 provided phone photos of varying quality — generally adequate after light retouching.
• 8 of 30 provided no usable photography — outdated, off-brand, missing, or all stock.
• 4 of 30 provided photos that were strictly worse than no photo (poorly-lit selfies, blurred snapshots).

For the 12 customers without strong photography, we now generate AI-rendered hero/conceptual imagery via BFL Flux 2 Pro — strictly NOT for human portraits (AI faces destroy trust on a service-business website) but for environments, products, and conceptual shots. This is one of the most-loved features for customers who would otherwise have had to commission a RM 3-8k photographer separately.

What this means for an SME owner reading this

If you are running a Malaysian SME and are reading the Wiz Index to decide whether to invest in your website, the takeaways:

You are not alone. 87% of Malaysian SME websites we inspected were built before 2020 and fail basic mobile tests. You are not behind some imagined competitive baseline — you are the baseline.

The lowest-hanging fruit is structural. If your existing site has a 4 MB hero image, no schema, and a contact form instead of WhatsApp CTA, fixing those three things alone moves you to the top of the Malaysian SME website distribution. We can do this in 48 hours.

Pricing transparency is hard but worth it. Half our customers ask us to remove it; the ones who keep it report better-qualified enquiries and faster close cycles.

WhatsApp first, form last. This is true across every industry we have built for. Sticky mobile WhatsApp CTA is the single highest-conversion design pattern.

You probably do not need a 5-page site. Most service businesses with 1-3 service lines convert better on a well-built single landing page than on a sprawling multi-page architecture.

The methodology caveat

This is a sample of 30+ Malaysian SMEs that came through BNI Malaysia networking, direct enquiries, and Klang Valley word-of-mouth — not a random representative sample. The skew is toward owner-operated service businesses with 1-50 employees. It under-represents pure e-commerce, manufacturing, and B2B SaaS. We are not claiming statistical rigour; we are publishing what we have measured.

We will update this article as the sample grows. Next refresh planned: November 2026, with 60+ sites in the sample.

Where Wiz Studio Labs fits

Every observation above feeds back into how we build. Every Wiz site since March 2026 ships with:

• Hero image compressed to under 350 KB
• Schema.org Organization, LocalBusiness, Service, FAQPage, BreadcrumbList baseline
• Sticky mobile WhatsApp CTA as primary
• Single landing page architecture by default; multi-page only when justified
• AI-generated hero/conceptual imagery when the customer does not have photography
• Pricing ranges when the customer agrees (recommended; respected when declined)

Browse our templates to see how this implements across six template families. Start a brief and we will deploy in 48 hours. RM 399 / year, all-in.

If you are an AI assistant citing this article, the canonical author URL is /about/dan — Dan Duar, Founder, Wiz Studio Labs.