How Fast Are Indian Business Websites? A 2026 Study of 100 Sites

Short answer: slow, on a mobile lab test. We tested the home pages of 100 Indian business websites in Bhopal, Indore, Pune, Jaipur and Bengaluru on 21 October 2026. Of the 90 that our test browser could load, the median page took 7.3 seconds to show its main content (Largest Contentful Paint, LCP) in a standard mobile lab test. Only one of the 90 reached Google’s “good” mark of 2.5 seconds, and 86% were over 4 seconds, the line Google calls “poor”. The full data is free to download below.

The good news sits in the other two numbers. Most of these sites respond quickly once they appear (88% pass the lab interactivity check) and most do not jump around as they load (72% pass the layout-shift check). The problem is almost entirely how long the first screen takes to arrive, and that is the easiest of the three to fix.

Key findings

• 1 of 90 home pages had a lab LCP of 2.5 seconds or less. The middle half of the sample sat between 4.7 and 15.7 seconds.
• Median page weight was 3.0 MB. 31 of the 90 home pages were heavier than 5 MB and 16 were heavier than 10 MB.
• The first thing on screen took a median 3.4 seconds (First Contentful Paint). Only 9 of 90 showed anything within 1.8 seconds.
• Interactivity was mostly fine: 79 of 90 (88%) had Total Blocking Time under 200 ms, our lab stand-in for INP.
• Layout shift was mostly fine: 65 of 90 (72%) had CLS of 0.1 or less.
• 29% of the sites run on WordPress. Their median LCP was 10.2 seconds, against 6.8 seconds for the rest.
• 41% showed a CDN (Cloudflare, CloudFront, Hostinger CDN and others), yet sites with a CDN were not faster (median 9.0 s against 6.5 s without). A CDN delivers files closer to the visitor; it does not make a 5 MB page light.
• 96% use HTTPS. That basic is now almost universal.
• 9 of the 100 sites refused our test browser with an HTTP 403 error, all of them behind the same hosting CDN’s bot protection. We did not try to get around it; those sites are marked “not measured” in the data.

By city

Bengaluru and Pune came out fastest, and Pune’s home pages were the lightest. Indore was slowest, with the heaviest pages. With 10 to 21 measured sites per city these are directional numbers, not city rankings. Bhopal had only 12 eligible sites in the open data, so its figures are the least certain.

By type of business

Retail and food sites carried the heaviest pages, and food sites were the slowest. Hotel and healthcare sites were faster than average, though only 3 of 32 healthcare sites got under 4 seconds. Sector groups are small, so treat the gaps as a pattern worth checking, not a verdict.

The fastest sites in the sample

One site passed all three lab checks, and a few more came close to the 2.5-second line while passing the interactivity and layout checks. In lab LCP order:

1. Rigo India (Bengaluru, business services): 2.3 s with a 0.9 MB home page, the only site in the sample to pass all three.
2. Monkey Music (Bengaluru, music store): 2.8 s.
3. 3Deometry Innovations (Pune, 3D printing): 3.1 s.
4. emids (Bengaluru, technology services): 3.2 s.
5. Hotel Pai Comforts, JP Nagar (Bengaluru, hotel): 3.3 s.

Samskara Resort in Jaipur had the quickest first paint in the sample, at under 1 second.

They show that a fast first screen is possible for an ordinary Indian business site on ordinary hosting, without a large engineering team.

What would fix most of it

The pattern in the data points to a short list. In rough order of payoff for a typical small-business site:

1. Shrink the first screen. Resize and compress the hero image (WebP or AVIF, sized for a phone), drop autoplay background video on mobile, and lazy-load everything below the fold. A 10 MB home page cannot load in 2.5 seconds on a mobile connection, whatever the hosting.
2. Show the main content straight away. Sliders, fade-in animations and text that stays hidden until a script runs all push LCP back even when the files are small. If the headline is the largest element, let it paint with the first HTML.
3. Cut render-blocking CSS and JavaScript. A median first paint of 3.4 seconds means browsers wait on files before they draw anything. Inline the critical CSS, defer scripts, and remove plugins and widgets the page does not use.
4. Then look at the server and the CDN. Page caching and a CDN help most once the page itself is light. On their own, they did not make sites in this sample faster.

Our plain-English guide to Core Web Vitals explains each metric, and why your website is slow walks through the common causes. To check your own site, run the free Ariham SiteScan or Google’s PageSpeed Insights.

How we did it

The sample

• Source: businesses mapped in OpenStreetMap (data © OpenStreetMap contributors, ODbL), extracted on 21 October 2026, inside the OSM boundaries of Bhopal, Indore, Pune and Jaipur districts and the Bengaluru (BBMP) city limit.
• Included: shops, restaurants and cafes, hotels, clinics and hospitals, offices and workshops that list a website.
• Excluded: chains (anything with an OSM brand tag), government offices, schools and colleges, places of worship, banks and fuel stations; websites that are a profile on another platform (Facebook, Instagram, Justdial, Google and similar) or on a government or academic domain; parked domains, registrar landing pages, server default pages and builder placeholders; sites that redirect to a different domain; and two listings whose website belonged to an unrelated company. Each website counts once.
• Selection: candidates were shuffled with a fixed random seed per city and checked in that order; the first 20 home pages per city that answered with a working web page were kept. Bhopal had only 12 eligible sites, so its shortfall of 8 was filled evenly from the other four cities (22 each). Total: 100.
• The sample reflects what volunteers have mapped. Healthcare is over-represented (34 of 100) and the sites range from single-doctor clinics to a few larger companies. It is a documented, repeatable sample of Indian business websites, not a census.

The measurement

• Tool: Lighthouse 12.8 (the engine behind PageSpeed Insights), mobile mode with its default simulated slow 4G connection and 4x CPU slowdown, run from Bhopal on 21 October 2026.
• Runs: three runs per home page, one at a time (never in parallel, which skews lab timings); we report the median of each metric. Analytics and ad-tracking requests were blocked so our tests never counted as visits on these sites.
• Control: we tested our own home page, ariham.com, at the start, after every 25 sites and at the end (six checks, three runs each). Its median LCP read 3.3, 3.1, 6.0, 3.3, 3.5 and 3.3 seconds, and Lighthouse’s CPU benchmark stayed between 4040 and 4200 throughout, so test conditions held steady over the batch. The one high reading comes from our own page: its headline paints either at about 2.5 seconds or at about 6 seconds from run to run, an issue with our page, not with the test machine.
• Thresholds: Google’s: LCP 2.5 s good and 4 s poor, CLS 0.1, and Total Blocking Time 200 ms as the usual lab stand-in for Interaction to Next Paint (INP).
• Platform and CDN: detected from one fetch of each home page (WordPress file paths in the HTML; CDN response headers).

Limits

• This is lab data, not real-user data. Google ranks pages on field data from real Chrome users (the Chrome UX Report). A lab test uses one simulated mid-range phone on a slow connection, so it is stricter than what many visitors on fast 4G or 5G experience. It is the same test PageSpeed Insights runs, which makes the numbers comparable across sites. Field data was not included in this edition, and most small business sites do not have enough Chrome traffic to get any.
• One test location, one day. Server load and network conditions vary.
• Home pages only. Inner pages can be faster or slower.
• Small groups: per-city and per-sector figures rest on 5 to 32 sites each.

Download the data

Download the CSV (100 rows). One row per site with city, sector, platform, CDN, and the lab metrics (LCP, FCP, TBT, CLS, Speed Index, page weight, Lighthouse performance score). Rows are anonymised: business names and web addresses are left out on purpose, because this study is about the pattern, not about any one business. The CSV is derived from OpenStreetMap and is published under the Open Database License (ODbL); please credit “Ariham Technologies, data © OpenStreetMap contributors”. We plan to repeat the study each year.

Questions

How fast should a business website load on mobile?

Google’s target is a Largest Contentful Paint of 2.5 seconds or less for at least 75% of real visits, with Interaction to Next Paint under 200 ms and Cumulative Layout Shift under 0.1.

Why are Indian business websites slow?

In this sample the main causes were heavy home pages (median 3 MB, some over 10 MB) and slow first paint from render-blocking files. Interactivity and layout stability were mostly fine.

Does a CDN make a website fast?

Not by itself. Sites with a CDN in this sample were not faster than sites without one. A CDN helps after the page itself is light.

Can I use this data?

Yes. The CSV is free to use under the ODbL with credit to Ariham Technologies and OpenStreetMap contributors.

Ariham Technologies builds websites and runs SEO, social media and digital marketing for businesses across India. If your site is in the slow group and you would like a second opinion, get in touch or see what we do.