The browser worker at work: 30 companies, real money.

Alibaba.com

B2B e-commerce marketplace

• 14% more monthly active users on iOS; 30% more on Android
• 4× higher interaction rate from Add to Home Screen
• Push notifications at the same open rate as the native app

AliExpress

Consumer e-commerce · Alibaba Group

• 82% increase in iOS conversion rate
• 2× more pages visited per session, per user
• 74% increase in time spent per session

Flipkart Lite

E-commerce · India

• 3× more time on site — 3.5 minutes vs 70 seconds
• 40% higher re-engagement rate; 3× lower data usage
• 63% of Flipkart Lite users reached the site via 2G
• 60% of all visits came from the home-screen icon; service workers let shoppers browse categories, searches and product pages offline

Lancôme

Luxury beauty e-commerce · L'Oréal · launched Oct 2016

• 51% increase in mobile sessions overall; 53% on iOS
• 15% decrease in bounce rates
• Push notifications: 18% open rate; over 18,000 shoppers signed up for alerts

Carry the caveats: the recovered-cart push figure conflicts between sources — PWA Stats states a “12% increase in recovered carts” and “8% of people responding to a notification make a purchase”; Retail Dive (24 May 2017) states push “has driven an 8% rise in conversion rates on recovered carts”. Also: most of Lancôme's 2017 PWA users were on iPhones, which did not then support service workers — so the 17% lift is largely a fast-mobile-site-rebuild effect.

BookMyShow

Entertainment ticketing · India

• PWA 54× smaller than the Android app and 180× smaller than the iOS app — about 440 KB
• Subsequent loads in less than 2.94 seconds; checkout within 30 seconds
• Service workers “to deliver reliable performance on slow or unreliable networks”

MakeMyTrip

Online travel · India

• 38% improvement in page load times; 160% increase in shopper sessions
• Bounce rate lowered by 20% vs the previous mobile site
• Reached users in 24% more cities; works offline from cached data; no install required

Housing.com

Online real estate · India · “Housing Go”

• 10% longer average session
• Page-load performance “increased by over 30 percent”
• Built on Service Workers, IndexedDB, Add to Home Screen and Push Notifications; users can review previous searches while offline

Rakuten 24

E-commerce — grocery / everyday goods · Japan · results over a 1-month timeframe

• 450% jump in visitor retention vs the previous mobile web flow
• Visit frequency per user up 310% compared to the rest of their web users
• Credited driver: installability. Workbox service worker (cache-first CSS/JS, stale-while-revalidate images) for offline / bad network is secondary
• Results achieved “without spending significant time and money building platform-specific apps for iOS and Android”

Read carefully: the 310% compares people who installed against “the rest of their web users” — a self-selected installed cohort, which inflates the apparent effect.

George (ASDA)

Clothing retail · UK · George.com

• 2× lower bounce rate; 20% more page views per visit
• 28% longer average time on site for visits from the Home screen
• Also used push notifications and offline page access

Debenhams

Department-store retail · UK · PWA built in 4 months on Mobify

• User journey from browsing to purchase 2× to 4× faster than the previous mobile site
• During peak trading 2017, mobile revenue overtook desktop revenue

West Elm

Home-furnishings retail · Williams-Sonoma · US · beta test

• 15% increase in average time spent on site
• Trade-press telling: a “9 percent” mobile conversion boost “driven largely by revenue-per-visitor gains” — i.e. the 9% is revenue/conversion, not two separate 9% gains

Petlove

Pet-supplies e-commerce · Brazil

• 2× more users go to checkout already signed in (simplified signup / auto sign-back-in)
• Not attributed to a single capability in the source — PWA plus credential / auto-sign-in

Small conflict: contemporaneous Chrome Dev Summit notes put time spent at 2.7×, not 2.8×. Conversion is 2.8× in both tellings.

Treebo

Budget hotels · India · PWA 2017, results published Feb 2018

• Conversion rates for repeat users: 3× increase
• Median interactive time on mobile dropped to 1.5 s

BMW

Automotive · BMW.com

• 50% growth in mobile users; 49% more site visits vs the old site
• Page load 4× faster than the old site

Do not credit the PWA alone: the project was AMP + PWA combined (AMP content pages feeding a PWA). Quote the 4× as an AMP+PWA result.

5miles

Local buy/sell marketplace · US

• 50% decrease in bounce rates
• 30% increase in time spent on site
• Push notifications also added for Android mobile-web users

Flagged headline: web.dev's own Summary headline says the PWA “increases conversions by 60%” — but its Results list contains no 60% figure. The two do not reconcile; this card quotes the Results list, not the headline.

Garbarino

Electronics / home retail · Argentina · sourced but not counted in the 30 — aggregator-only

• Returning visitors +13%; pageviews +35%; bounce rate −9%

OLX

Online classifieds · India · launched Nov 2016

• 146% higher CTR on Ads; 80% lower bounce rates
• 23% less time until page is interactive
• Over 600,000 users signed up for notifications since the site launched in November 2016
• Speed drove the ad-CTR / bounce gains — “ads loading faster, the clickthrough rate (CTR) rising 146%”

United eXtra Electronics

Consumer-electronics retail · Saudi Arabia

• 4× increase in re-engagement
• Opted-in users returned 4× more often and spent twice as much time on the site (within six weeks)
• Web push came to exceed email as eXtra.com's best-performing retention-marketing channel

Scope note: this is a web-push implementation on the mobile site (via Mobify), published by Google alongside the PWA showcases — not a full PWA rebuild story. The credited capability is web push only.

Trivago

Hotel search / travel metasearch · PWA in 33 languages / 55 countries

• Engagement +150% for people who add the PWA to the home screen
• Repeat visits rose to an average of 2 vs 0.8 on the previous mobile site
• More than 500,000 people added the icon
• 67% of users who go offline “continue to browse the site when they come back online”

Twitter Lite

Social media · became the default mobile-web experience globally Apr 2017

• Takes up less than 1 MB on device; launches up to 30% faster
• Data Saver “can reduce your data usage by up to 70 percent” (Twitter's own launch post)
• Secondary summaries of the web.dev case add: 250,000 daily unique users launching from the home screen an average of 4 times a day, and 10 million push notifications sent daily

Ola

Ride-hailing · India

• 68% increase in mobile traffic in Tier 2 and 3 cities
• In Tier 3 cities, conversion rates are 30% higher with the PWA (Tier 2: same as native app)
• Native sizes for scale: Android app 60 MB, iOS app 100 MB
• 20% of users who book in the PWA had previously uninstalled the app
• Repeat visits use as little as 10 KB; 3.4-second first visit and under a second for repeat visits on 2G/3G

Konga

E-commerce · Nigeria

• 92% less data for initial load, vs native app
• 82% less data to complete first transaction, vs native app
• 63% less data for initial load, vs previous mobile web experience
• 84% less data to complete first transaction, vs previous mobile web experience

Framing note: the often-quoted “−92% data usage” is only correct in the vs-native-app, initial-load framing. Google's own PWA explainer credits service workers directly: “Service workers enabled Konga to send 63% less data for initial page loads, and 84% less data to complete the first transaction!”

Jumia Travel

Travel / hotel booking · Africa · Jumia Group · case study released 13 Jun 2017

• 50% lower bounce rate
• 5× less data used; 25× less device storage required
• PWA uses 80% less data than the native app to complete the first transaction
• Release also credits increased conversions through push notifications

Uber — m.uber

Ride-hailing · global · Uber Engineering, Jun 2017

• Designed so anyone can request a ride on any device / network

No financial figure exists: no revenue, conversion or booking figure was published for m.uber. This is a reach / enablement case only — do not attach a financial % to it.

Starbucks

Coffee / food ordering · PWA ordering launched 2017; builder case study 2018

• Daily active users doubled (builder's account — see flag)
• Desktop users ordering at about the same rate as mobile app users
• Offline-first ordering: browse the menu, customise orders and add items to cart without a connection, with a service worker keeping core JavaScript/CSS available offline

Flag: the builder-side account is most often retold as daily active users doubling; other retellings phrase it as Starbucks doubling “the number of web users who place orders each day”. No Google/web.dev Starbucks case-study page with a results table was located — this case rests on builder (Formidable) and press accounts, unlike most entries here.

Tinder Online

Dating / social

• Core experience delivered at 10% of the data-investment costs for a user in a data-costly market (2.8 MB)
• Business signals in the primary study are qualitative: users “swipe more”, “message more”, “purchase on par with native apps”, longer sessions vs native
• Service workers “for network resilience”; push notifications for chat engagement; long-term asset caching via service worker / Workbox

No % revenue figure exists in the primary study — and a vendor page's “+137% engagement” claim for Tinder Online is not in it. Do not repeat it.

Pinterest

Social / visual discovery · two different measurement periods — do not merge

• 2017 case: time spent +40% vs the old mobile web experience; core engagements +60%; time-to-interactive fell from 23 seconds to 5.6 seconds on average Android hardware over slow 3G
• 2018 one-year retrospective: weekly active users on mobile web +103% year-over-year (Brazil 156%, India 312%)
• Session length +296%; Pins seen +401%; logins +370%; new signups +843% year-over-year
• 800,000 weekly home-screen users in under 6 months
• Service workers “for network resilience” alongside the speed work

Nikkei

News / financial media · Japan · PWA launched Nov 2017

• 2.3× organic traffic; 49% more daily active users
• 2× page views per session; 2× better Speed Index
• 14 seconds faster time-to-interactive; 75% faster loading with prefetch; Lighthouse score 23 → 82
• Credited directly: “By adding a service worker… top articles are always available and loaded almost immediately because they're stored using Cache Storage” — cache-first also gives offline / flaky-network reliability; installable via manifest

Forbes

Business media · US

• Secondary tellings also credit push notifications and Add to Home Screen

Weakest sourcing in this list: aggregator only. The same aggregator also carries a different Forbes set — “2x increase in average user session length, 6x completion rate, and 20% more impressions. Loads in 0.8s down from 3 to 12s” and “Forbes' PWA loads in 2.5 seconds… Impressions per visit are up 10%”. The aggregator does not separate their provenance; the figures may come from different Forbes tests/talks.

The Weather Channel

Weather media · weather.com · global · push test 2016; international PWA rollout reported 2017

• 80% improvement in load time
• PWA “now available in 62 languages and 178 countries”
• Push = severe-weather alerts: “they used a technology called service workers to provide offline processing and deliver messages quickly”
• One code base supporting 60+ languages is also a cost / reach point — see Lever 5

Selio

Buy / sell / chat marketplace app

• Conference-talk notes put figures on it: CPA €3–5 for native apps, and €0.30–0.40 for PWA users
• Users spent an average of 6+ minutes in the app
• The saving is acquisition cost (link-based, no app-store friction) — not a measured dev-cost saving

Cost avoidance, stated without a figure

Inside three cases above — honest about what is and is not measured

• Rakuten 24 (№ 08) “was able to drive these results without spending significant time and money building platform-specific apps for iOS and Android” — it had judged itself “not in a position to invest the time and cost in developing a platform-specific app”
• MakeMyTrip (№ 06) — operating on any OS is “non-trivial in terms of development time and expense” saved
• The Weather Channel (№ 29) — 62 languages / 178 countries served by enabling languages “using one code base”

Open point, stated plainly: no case in the verified set publishes an audited development-cost saving in currency. Treat any specific “$ saved by building a PWA instead of native” figure circulating in blog posts as unsourced — see the Alibaba flag below.