LankaRank Research
Study, October 2026
Sri Lankan business websites in 2026: what we found when we checked 2,061
We ran the websites listed in four Sri Lankan business directories through the same checks our free tool uses, then tested the checker against a real browser. Every figure below has its count, a 95% range, and the limits of what we measured.
On this page
Key findings
- 39%of listed business websites didn't work810 of 2,061; 1 in 5 had expired
- 76%had no WhatsApp button42% had no one-tap way to call or message
- 53%didn't name their town where Google looksin the page title or main heading
- 62%gave Google no business detailsname, address and phone in structured data
- 79%took over 4 seconds to load on a phonethe typical site took 7.9 seconds
- 97%had little or no Sinhala or Tamilon their homepage
- 41%were hand-built by a developerso owners can't make changes themselves
- 51%of Google Maps profiles weren't claimedbusinesses with no website, 63 of 124
"Working sites" means the 1,246 sites that loaded normally when we re-checked them. Figures describe websites listed in these directories, not every business in Sri Lanka. See limits.
How many listed Sri Lankan business websites don't work?
About 2 in 5. 810 of the 2,061 websites listed in four directories (39%, 95% range 37–41%) didn't work, and 403 of them (20%) no longer existed at all because the domain had expired. Directories were still sending customers to these addresses.
The rate depends on the directory. Rainbow Pages and lankayp.com, which list more small businesses, are close to half. The exporters' directory, which lists more medium and large firms, is lower. Leaving out sites listed only as exporters, 46% didn't work.
| Directory | Sample | Broken | Sites | Expired |
|---|---|---|---|---|
| SLT Rainbow Pages | random sample | 47% | 318 of 678 | 26% |
| Sri Lanka YP (lankayp.com) | random sample | 48% | 224 of 466 | 26% |
| SLTDA accommodation register | every listing | 35% | 76 of 219 | 17% |
| EDB Exporters Directory | every listing | 27% | 216 of 789 | 9% |
| All four (each site counted once) | 39% | 810 of 2,061 | 20% |
| What the address showed | Sites | Share |
|---|---|---|
| Domain expired: the address no longer exists | 403 | 50% |
| No answer from the server | 109 | 13% |
| Page not found (404) | 92 | 11% |
| Empty, "coming soon" or a web server's default page | 90 | 11% |
| Parked or up for sale | 61 | 8% |
| Server error | 23 | 3% |
| Taken over: gambling, scam or unrelated foreign pages | 17 | 2% |
| Security certificate error | 14 | 2% |
| Sends visitors to an unrelated site | 1 | under 1% |
Of the expired addresses, 17 had been bought by someone else and now showed slot-gambling, scam or unrelated foreign pages under the business's old name. Free Google business.site pages fared worst: 53 of 65 were gone, because Google shut the service down in 2024.
Can customers get in touch from a phone?
Often not in one tap. 76% of working sites had no WhatsApp button and 49% had no tap-to-call number. 42% had neither, so a customer on a phone had to copy the number by hand. We looked at the homepage and the Contact page it links to, as a customer would.
| Finding | Share | Sites | 95% range We checked a sample, not every one. If we repeated the study with a different sample, the share would land in this range 95 times out of 100. |
|---|---|---|---|
| No WhatsApp button | 76% | 921 of 1,211 | 74–78% |
| No tap-to-call phone number | 49% | 592 of 1,215 | 46–52% |
| Neither of the two: no one-tap way to get in touch | 42% | 509 of 1,211 | 39–45% |
| No Google Maps map or link | 41% | 495 of 1,201 | 38–44% |
| No phone number shown at all | 15% | 184 of 1,216 | 13–17% |
| No link to a Facebook page | 32% | 393 of 1,214 | 30–35% |
Shares are of the sites each check applied to, so the totals differ slightly. 35 sites built entirely in JavaScript are left out of the WhatsApp figure because their buttons can't be read without running the page.
Are Sri Lankan business websites set up for Google?
The basics are often missing. 62% didn't give Google their business details in structured data A short piece of code, invisible to visitors, that tells Google your business name, address, phone and opening hours in a format it reads reliably. Also called schema., and 53% didn't name their town in their page title or main heading. Local searches often name a place: in Google's own figures for Sri Lanka, "cake shop colombo" is searched about 1,300 times a month and "dental clinic kandy" about 390.
None of this guarantees a ranking. But Google says businesses with complete and accurate information are more likely to show up in local results.
| Finding | Share | Sites | 95% range We checked a sample, not every one. If we repeated the study with a different sample, the share would land in this range 95 times out of 100. |
|---|---|---|---|
| Business details (name, address, phone) not given to Google in structured data | 62% | 778 of 1,246 | 60–65% |
| Town not named in the page title or main heading | 53% | 655 of 1,237 | 50–56% |
| Page title too short or too long to show in full on Google | 58% | 721 of 1,233 | 56–61% |
| Not exactly one main heading | 58% | 711 of 1,220 | 55–61% |
| No description for Google's results page | 35% | 438 of 1,246 | 33–38% |
| No working sitemap | 29% | 360 of 1,246 | 26–31% |
| Images without descriptions | 82% | 984 of 1,197 | 80–84% |
| Homepage hidden from Google (noindex) | 2% | 19 of 1,246 | 1–2% |
How fast are Sri Lankan business websites on a phone?
Mostly slow. On Google's PageSpeed Insights phone test, 79% of a random 194 working sites took more than 4 seconds to show their main content (153 of 194, 95% range 73–84%). The typical site took 7.9 seconds. Google's target is 2.5 seconds.
61
median PageSpeed score out of 100 Google's own score out of 100 for how fast and smooth a page is on a phone. 90 or more is good; under 50 is poor.
31%
scored "poor" (under 50)
10%
scored "good" (90 or more)
Only 31% had enough real Chrome visitors for Google to publish real-world speed data, so these are Google's lab tests: one run each, which can vary by 5 to 10 points from run to run. Google's research found that as a phone page's load time goes from 1 second to 3, the chance a visitor leaves goes up by 32%.
Are they secure and phone-ready?
Mostly, with one common gap. Almost every working site had a secure https:// address, but 18% didn't send visitors who type http:// on to it, so those visitors see Chrome's "Not secure" warning. Since 8 in 10 devices Sri Lankans use to go online are smartphones, the 6% with no phone layout at all are hard to use for most visitors.
| Finding | Share | Sites | 95% range We checked a sample, not every one. If we repeated the study with a different sample, the share would land in this range 95 times out of 100. |
|---|---|---|---|
| http:// visitors not sent on to the secure https:// address | 18% | 221 of 1,197 | 16–21% |
| No working https:// at all | 3% | 40 of 1,246 | 2–4% |
| Not set up for phone screens (no mobile viewport) | 6% | 72 of 1,246 | 5–7% |
Do Sri Lankan business websites speak Sinhala or Tamil?
Rarely. 97% of working sites (1,171 of 1,211) had little or no Sinhala or Tamil text on their homepage.
That isn't automatically a problem, because of how Sri Lankans search. We looked up common searches in Google's keyword data for Sri Lanka. 0 of 10 searches typed in Sinhala script showed any measurable volume. In Singlish (Sinhala typed in English letters) 4 of 8 did, in Tamil 4 of 4, and in English 5 of 6. Google may still receive Sinhala-script searches; its keyword tool doesn't report them. So what matters for a business is whether its pages use the words customers type, in English, Singlish or Tamil.
| Search | Searches a month |
|---|---|
| house for rent colombo | 14,800 |
| sinhala sindu | 8,100 |
| cake shop colombo | 1,300 |
| dental clinic kandy | 390 |
| guest house ella | 210 |
| gewal kuliyata | 170 |
What are Sri Lankan business websites built with?
Half ran WordPress, and 4 in 10 were hand-built. 49% of working sites ran WordPress and 41% were hand-built by a developer with no editor, so their owners need the developer for every change. Site builders such as Wix and Shopify ran only 7%.
| Finding | Share | Sites | 95% range We checked a sample, not every one. If we repeated the study with a different sample, the share would land in this range 95 times out of 100. |
|---|---|---|---|
| WordPress | 49% | 610 of 1,251 | 46–52% |
| Hand-built by a developer, no editor | 41% | 513 of 1,251 | 38–44% |
| Site builder (Wix, Shopify, Webflow and others) | 7% | 93 of 1,251 | 6–9% |
| Other content systems (Joomla, Drupal and others) | 3% | 35 of 1,251 | 2–4% |
How many Sri Lankan businesses have a website?
Few of those listed in directories. Only 5% of Rainbow Pages listings (751 of 16,353) and 12% of lankayp.com listings (507 of 4,083) gave a website of their own, and 39% of the listed websites we checked didn't work. Most businesses rely on their Google Maps listing instead, and many of those listings are incomplete.
In a sample of 2,635 businesses from the first page of 150 Google searches of the form "type of business in town", across all nine provinces, the 2,103 with no website of their own often lacked the basics. These are the more prominent businesses, so the real picture is likely worse.
| Finding | Share | Profiles | 95% range We checked a sample, not every one. If we repeated the study with a different sample, the share would land in this range 95 times out of 100. |
|---|---|---|---|
| No phone number | 19% | 397 of 2,103 | 17–21% |
| No opening hours | 24% | 503 of 2,103 | 22–26% |
| Fewer than 10 photos | 61% | 1,286 of 2,103 | 59–63% |
| Fewer than 10 reviews | 64% | 1,349 of 2,103 | 62–66% |
Google's public data doesn't say whether a profile has been claimed by its owner, so we looked up 330 Rainbow Pages businesses with no website one by one. 124 had a profile we could confidently match. Of those, about half (63, 95% range 42–59%) hadn't been claimed, so the owner can't reply to reviews, see how many people called, or stop others from changing the details. 73% had no description and 87% used a single category. You can check your own Google profile the same way.
How do Sri Lankan business websites score overall?
The typical working site scored 74 out of 100 on our health score, leaving speed aside. A quarter scored 64 or less and a quarter 83 or more. Checks count by how much they matter to customers, so a missing WhatsApp button costs more than a missing image description.
Band with the median score (74)
How we did it
- Listings. We collected business listings from SLT Rainbow Pages and lankayp.com (random samples of 16,503 and 4,083 listings), and every listing in the EDB Exporters Directory and the SLTDA accommodation register. We read each directory's robots.txt and requested one page at a time.
- Which sites. We kept listings with a website of their own, and left out government, universities, listed companies, banks, large chains and foreign parents. A site listed in several directories was checked once. That left 2,061 distinct sites we could test.
- Checks. Each site went through the same checker as our free tool, from a connection in Sri Lanka, on 13–15 October 2026. Sites we couldn't reach were checked again later the same day. Pages taken over by spam, parked or showing a server's default page were counted as not working.
- Re-run. After improving the checker, we re-ran it over every working site, reading the homepage and the Contact page it links to. The figures for working sites come from this re-run.
- Testing the checker. We opened 50 random working sites in real Chrome, as a phone would, and compared what Chrome saw with the checker's verdicts. 906 of 918 results matched. Of the 12 that didn't, the checker was right in 9; the 3 it got wrong were things only added by JavaScript after the page loads.
- Speed. A random 194 working sites went through Google's PageSpeed Insights phone test.
- Google profiles. A free Google Places search (2,635 businesses), and a paid lookup of 330 Rainbow Pages businesses for claimed status, each match checked by hand.
- Statistics. 95% ranges are Wilson score intervals: if we repeated the sampling, the true share would fall in that range 95 times in 100.
Limits
- Directories, not all businesses. These are websites listed in four directories. Rainbow Pages leans towards older firms with landlines; the exporters' directory towards larger firms. Per-directory figures are given where they differ.
- Homepage and Contact page only. A business may have WhatsApp or a map on another page.
- One snapshot. Sites go up and down; a site that was unreachable on the day may work now.
- Lab speed tests. One run each from Google's servers, not real visitors' phones.
- Search volumes are Google Ads estimates, rounded by Google, and the language test covered 28 searches, enough to show a pattern, not to measure it precisely.
- Google Maps samples favour prominent businesses, and the claimed-status sample is small (124 profiles).
Quoting these figures
You're welcome to quote any figure on this page. Please link back to this page so readers can see the method and limits. Suggested citation:
LankaRank, "Sri Lankan business websites in 2026: what we found when we checked 2,061", 15 October 2026. https://lankarank.lk/research/sri-lanka-business-websites
Other studies quoted: Google Ads keyword volumes via DataForSEO (October 2026); Google/SOASTA Research (2017) on load time; Department of Census and Statistics, Computer Literacy Statistics 2025, on smartphone use; Google Business Profile Help on local ranking.