Website & Ecommerce Lesgo Media · 2026
- 6.6%median landing page conversion rate
- 21 timesdrop in qualifying odds at 30 min
- 3,000visitors per version to A/B test
The conversation usually goes the same way. Sales are flat, so the plan becomes more ads. Budget goes up, sessions go up, real enquiries barely move.
Traffic is the expensive input. Conversion is the one you have already paid for. If 400 people land on your site this month and 4 enquire, doubling ad spend costs you money every month after that. Getting those 4 to 8 costs one focused weekend, once.
This article covers how to measure your real rate, where to look for the drop-off, what usually breaks, and how to test honestly when your traffic is too small for statistics.
Work out your current conversion rate before you change anything
The formula is conversions divided by sessions, times 100. The argument is always about what counts.
For a Malaysian service business, a conversion is a form submitted, a WhatsApp button tapped, a phone number tapped, or a quote requested. It is not a page view or a brochure download you never follow up on. Count soft actions and your dashboard looks healthy while your calendar stays empty.
Measure over 90 days, not 7. One viral post, one public holiday, one week of Raya, and a weekly rate swings by half.
Then split the number by device and by channel. A site converting at 4% on desktop and 0.8% on mobile has a mobile problem, not a conversion problem. A site converting at 6% from Google and 0.5% from Facebook has a traffic quality problem on one channel. Our guide on what actually counts as a lead is worth reading before you settle your definition.
The short version
- Count only actions a human on your team has to respond to.
- Ninety days minimum, or you are reading noise.
- Always split by device and channel before drawing a conclusion.
What a realistic conversion rate looks like in your industry
Unbounce publishes the largest public dataset on this. Its Conversion Benchmark Report analysed 41,000 landing pages carrying 464 million visitors, and reported a median conversion rate of 6.6% across all industries, with industry medians ranging from 3.8% to 12.3%.
Read that carefully. Those are dedicated landing pages built for one offer and fed by paid traffic. Your homepage carries job applicants, suppliers, customers checking your address, and your own staff. Measured the same way it will always look worse, and that is arithmetic, not failure.
There is also a local distortion. Plenty of Malaysian sites push people to WhatsApp, and if that tap is not tracked as an event the site reports near zero conversions while the sales team is busy. Check that before you rebuild anything.
The honest benchmark is your own site last quarter. Beating 6.6% means nothing if the enquiries are unqualified, and 2% is fine if they close at RM40,000 each.
Finding where people drop off using GA4 and a session recording tool
Two tools, two questions. GA4 shows where people leave. A recording tool shows why.
- Mark real conversions as key events in GA4. Form submission, outbound click to wa.me, click on a tel: link. Until that is done every other number is decoration.
- Build a funnel exploration. Landing page, service page, contact page, key event. GA4 shows the percentage lost at each step.
- Open the Landing page report and add device as a secondary dimension. The mobile to desktop gap is usually the loudest signal in the account.
- Install a recording tool. Microsoft Clarity is free with no session cap, which makes it the sensible default for an SME. Hotjar has a free tier if you prefer it.
- Watch 20 sessions on your worst page. Not 200. Twenty is enough for patterns to repeat.
- Look for rage clicks and dead clicks. Somebody tapping a button that does nothing on mobile is a fix you can ship the same day.
Collect two weeks of recordings before you watch. Watching live while your own team browses the site pollutes the sample.
The seven things that usually break conversion on a Malaysian SME site
After enough audits the same list keeps appearing. Almost none of it needs a redesign.
| Problem | What the visitor does | Fix |
|---|---|---|
| Unclear offer above the fold | Reads two lines, cannot tell what you sell, leaves | One sentence: what you do, who for, where |
| Price hidden when they came for price | Hunts three pages, gives up, checks a competitor | Publish a range or a starting figure |
| Form asks too much too early | Sees nine fields, closes the tab | Name, phone, one line about the job |
| No WhatsApp option | Will not fill a form, will not email, leaves | Sticky WhatsApp button on mobile |
| Slow mobile pages | Never sees the page at all | Compress images, cut unused scripts |
| No trust signals | Believes nothing on the page | Named reviews, real photos, SSM number |
| Unclear next step | Reads everything, does nothing | One instruction, repeated three times |
The price row causes the most arguments and is the one worth winning. If somebody searched your service plus the word price, hiding it is not lead qualification, it is friction. A range with conditions attached filters out the wrong enquiries and earns trust from the right ones. Our practical checklist for a good business website covers the structural side.
The site converts fine, but nobody answered the enquiry
This failure never shows on a dashboard, because the dashboard says the conversion happened. The form was submitted. Nobody replied for two days.
The Lead Response Management study, run by Professor James Oldroyd on an InsideSales.com dataset of more than 15,000 leads and 100,000 call attempts, found the odds of qualifying a lead dropped 21 times when first contact came at 30 minutes instead of 5. Between 5 and 10 minutes the odds already fell fourfold.
In Malaysia the effect is sharper. The enquiry arrives on WhatsApp, where people expect a reply in minutes, and plenty of it lands after 9pm. A form emailing a shared inbox nobody opens on weekends is a leak you paid Meta or Google to create.
- Route every submission to a phone, not an inbox. A WhatsApp or Telegram alert to a named person, not a sales@ address four people ignore.
- Send an instant auto-reply. One line confirming receipt and when a human will respond holds the lead for an hour.
- Name one owner per day. Shared responsibility means nobody replies at 8pm on a Saturday.
- Log first response time. If you measure nothing else this quarter, measure this.
Fixing enquiry-to-customer is usually worth more than fixing click-to-enquiry
Take a site with 1,000 visitors a month converting at 2%, and a sales process closing 20% of enquiries. That is 20 enquiries and 4 customers.
Lifting the site from 2% to 3% is real work: new copy, new form, faster pages, maybe weeks of it. It gets you 30 enquiries and 6 customers.
Lifting the close rate from 20% to 30% is a follow-up discipline. Reply in ten minutes, call twice instead of once, send the quote the same day. It also gets 6 customers, and costs nothing but attention.
Most owners spend the quarter on the second row and never touch the third. Do the follow-up work first. It is faster, it is free, and it tells you whether your enquiries were ever worth anything.
Mobile speed is a conversion problem before it is a technical one
Google published data from roughly 3,700 mobile sites in 2016 stating that 53% of visits are likely to be abandoned if pages take longer than 3 seconds to load. Ten years on, Malaysian SME sites still fail this routinely, and the causes are consistent.
A hero video autoplaying on mobile. Product photos exported from the designer at 4MB each. Five tracking scripts nobody uses. A chat widget loading before the text. Each was added by somebody reasonable, and together they push a page past four seconds on 4G in a Klang Valley basement carpark, which is exactly where people check their phones.
Test on mobile data, not on office wifi, and fix the images first. Our guide to Core Web Vitals in plain language explains which numbers matter and which you can ignore.
Prioritising fixes by effort against impact
You will find fifteen problems and have time for four. Sort them honestly.
| Fix | Effort | Likely impact | Priority |
|---|---|---|---|
| Sticky WhatsApp button on mobile | 1 hour | High | This week |
| Cut form from 9 fields to 4 | 1 hour | High | This week |
| Route enquiries to a phone alert | 2 hours | High | This week |
| Publish a price range | Half a day plus an internal argument | High | This month |
| Compress every image | Half a day | Medium | This month |
| Add six named reviews with photos | 2 days | Medium | This month |
| Rewrite the homepage headline | 2 days | Unknown until tested | Next quarter |
| Full redesign | 6 to 10 weeks | Unknown | Only with evidence |
Everything in the one-hour rows ships this week, before anybody schedules a design meeting. A full rebuild is the most expensive way to test a hypothesis you have not written down, so it stays at the bottom until the cheap fixes run out. For pages built for paid traffic the rules are tighter, and we cover them in our guide to landing pages that convert in Malaysia.
How to test when your traffic is too low for A/B testing
Here is the part most conversion advice skips. With 400 visitors a month you cannot A/B test. At all.
The arithmetic is unforgiving. To detect a move from 2% to 3% at the usual 95% confidence and 80% power you need roughly 3,000 visitors per version, about 6,000 in total. At 400 a month that is more than a year for one test, by which time your offer, prices and competitors have all changed. A smaller lift, 2% to 2.4%, needs closer to 20,000 per version.
What to do instead
- Make the obviously better version and ship it. A 4 field form beats a 9 field form. You do not need a test to know that.
- Compare 8 to 12 weeks before against 8 to 12 weeks after, and accept it is directional, not proof.
- Change one meaningful thing at a time so the before and after is readable.
- Keep a dated change log. In six months it is the only record of what caused what.
- Replace statistics with evidence: 20 session recordings, 5 calls to recent customers.
Anyone selling A/B testing on 400 visitors a month is selling a report, not a result. Spend the money on fixes, and revisit testing past roughly 10,000 visitors a month. Building that traffic is a separate job, covered in our guide on generating leads with digital marketing.
Nak tahu angka sebenar untuk business anda?
Hantar detail ringkas. Kami semak akaun atau website anda, pastu tunjuk di mana duit bocor dan apa langkah paling berbaloi seterusnya.
Frequently asked questions about conversion rate optimisation in Malaysia
What counts as a conversion for a service business?
Any action that puts a real person in front of your sales team: a submitted enquiry form, a tapped WhatsApp button, a tapped phone number, a booked appointment. Newsletter signups and PDF downloads do not count unless somebody follows up on every one. Count soft actions and your dashboard improves while your revenue does not.
What is a good conversion rate for a Malaysian SME website?
Unbounce reports a median of 6.6% across 41,000 landing pages, with industry medians between 3.8% and 12.3%. Those are dedicated landing pages fed by paid traffic. A business homepage carrying mixed traffic normally sits lower, and that is not a fault. The rate worth beating is your own from last quarter.
How long should I measure before deciding a page is broken?
Ninety days. One week of data in a small business is noise, since a public holiday, a viral post or a school break can swing the number by half. Ninety days smooths that out and still reflects your current offer and prices.
Should I show prices on my website?
In most cases yes, as a range with the conditions attached. If a visitor searched your service plus the word price, hiding it does not qualify the lead, it sends them to a competitor who published theirs. A range filters out enquiries you cannot serve profitably.
How many fields should my enquiry form have?
Three or four. Name, phone number, and one line describing what they need. Everything else can be asked in the reply, when the person is already talking to you. Budget dropdowns and company size are questions for your CRM, not for a stranger deciding whether to contact you.
Do I need a paid tool to see where visitors drop off?
No. GA4 is free and shows where people leave. Microsoft Clarity is free with no session limit and shows why, including rage clicks and dead clicks. Between those two you have everything an SME needs. Paid tools become useful later, once you have enough traffic for real experiments.
Can I A/B test with 400 visitors a month?
No. Detecting a move from 2% to 3% at normal confidence needs roughly 3,000 visitors per version, about 6,000 in total, which is more than a year at that traffic level. Make the obviously better version, ship it, and compare 8 to 12 weeks before against after.
Is it better to spend on conversion work or on more ads?
Fix conversion first if you already get more than a few hundred visitors a month, because every improvement makes all future ad spend cheaper per lead. If almost nobody visits, no amount of conversion work helps and you need traffic first. The dividing line sits around 500 monthly visitors.
Conclusion: start with the visitors you already paid for
Measure your real rate over 90 days before touching anything, then compare it to your own last quarter rather than to the 6.6% median. Ship the one-hour fixes this week, put a phone alert on every enquiry, and treat first response time as the number that matters most. With a few hundred visitors a month you are making judgement calls, not running experiments, so make the obviously better version anyway. Versi Bahasa Melayu di sini: Tingkatkan Conversion Website. If you want somebody to look at your GA4 and tell you where the enquiries are leaking, talk to the Lesgo Media team for a free consultation.
Baca lagi
- High Converting Landing Pages in Malaysia: What Actually Works
- 5 Signs Your Business Website Needs a Redesign
- Beli Domain & Hosting Malaysia: Panduan Pilih untuk Business (2026)