Adoption and evidence
The evidence here is sharply asymmetric. What browsers can do is highly verifiable and is set out below with primary sources. What share of event attendees actually use an event app is not independently known by anybody, and the figures in circulation do not survive being checked.
Every figure on this page carries a primary source and a date. Where a widely quoted number could not be traced to one, it is named in the second half as something not to repeat, rather than quietly omitted.
What can be verified
Service workers are supported by browsers representing over 96% of global usage
The feature that makes criterion 2 possible is effectively universal. Published as “over 96%” deliberately: the live figure and the underlying dataset disagreed in the first decimal place on the day this was checked, so a decimal would imply precision that is not there. Can I use, service workers, usage data August 2026, checked 2026-09-09.
Safari on iOS has supported service workers since iOS 11.3, in March 2018
Continuously, to the current release. This is the strongest fact on the page: two independent primary datasets agree exactly, and it retires the most common technical objection to browser-based event apps, which is usually stated as though iOS were still the blocker it was a decade ago.MDN browser-compat-dataand caniuse, both checked 2026-09-09.
Since iOS 26, Safari imposes no technical requirements for installability at all
WebKit, in September 2025: “there are now zero requirements for ‘installability’ in Safari.” Any site added to the Home Screen opens as a web app. The same post confirms Home Screen web apps on iOS have never required a service worker. This is recent, material, and not yet widely reflected in the compatibility tables people quote.WebKit Features in Safari 26.0, published 2025-09-15, checked 2026-09-09.
On iOS, web push requires the user to have added the app to the Home Screen
A real limitation, stated here rather than glossed over. Push does not work in a Safari tab; it requires Home Screen installation, from iOS 16.4 onwards. This is the clearest capability gap between the browser and a native app, and it is the one legitimate reason an organiser might still choose to ship one.Web Push for Web Apps on iOS and iPadOS, published 2023-02-16, checked 2026-09-09.
Safari clears service worker caches after seven days of non-interaction, and exempts Home Screen web apps
Registrations and caches are removed after seven days of Safari use without interacting with the site. Home Screen web apps have their own counter and are exempt. For an event companion this is a good rule rather than an obstacle: it is a documented reason the software does not outlive the event, which is criterion 4, and a sourced engineering argument for offering Home Screen installation without requiring it, which is criterion 5.WebKit, full third-party cookie blocking and more, published 2020-03-24, checked 2026-09-09.
53% of mobile site visits are abandoned if a page takes more than three seconds to load
Read the label carefully. This is a page-load statistic, not an app-install statistic, and it is routinely relaundered as one. It says nothing about how many people abandon a download. It is included because delay tolerance on mobile is relevant and this is the one figure in the area with a stated sample and method.Google, The Need for Mobile Speed, data collected March 2016, n=3,700, checked 2026-09-09.
What cannot be verified, and why that matters
This is the more useful half of the page. Buyers are routinely quoted adoption figures during procurement. None of the ones in circulation can be checked, and one of the most-repeated has a dead source.
No independently verified measurement exists of event app adoption
Searching for the underlying study behind the commonly quoted rates produces vendor blog posts citing other vendor blog posts. The most-repeated figure, an “average event app adoption rate” of 63%, is published by a company that sells event apps to the events it is measuring, with no locatable report, no sample size, no date, and no definition of what “adoption” counts. It is named here as an example rather than repeated as a number. Since this was first written the page carrying it has been withdrawn: the URL returned the article on 2026-09-04 and from 2026-09-07 it returns a redirect to the blog index, still redirecting on 2026-09-09, so the most-quoted adoption figure in event technology now has no live source at all. The archived copy still carries the sentence, attributing the figure to a benchmark report that is itself not linked.The archived page carrying it, archived 2025-11-17, original URL re-checked and still redirecting 2026-09-09.
Event app vendors cannot measure a population adoption rate, even in principle
This one needs no statistic, only the method. A vendor's telemetry covers events that bought an event app. Events that did not buy one, and attendees at events using something else, are invisible to it. The denominator is missing by construction, so a rate derived from that data is not a population rate however carefully it is computed. Any figure of this kind, from any vendor including the publisher of this site, should be read as a statement about that vendor's customers.
The “a quarter of apps are used only once” figure has a dead primary source
It appears constantly in event technology marketing. It traces to a single 2016 analytics vendor report, and both of that report's primary pages now return 404. The secondary coverage survives; the underlying data does not. It is included here as a demonstration of how this literature works rather than as a fact.The surviving secondary report, published 2016-05-31; both primary URLs re-checked and still 404 on 2026-09-09.
No independent data exists on how often venue wifi fails
The mechanism is well understood and uncontroversial: dense concurrent connections in a confined space contend for finite spectrum, and demand peaks at doors and session changeovers. The rate is not known. No independent study measuring event wifi failure was found, only vendor success stories. This site therefore explains why dense venues strain wifi and makes no claim about how often it fails. For scale, one of the few artefacts with disclosed numbers records 80,880 devices and 22,126 peak concurrent connections across four days at a large trade show. The network operator’s own release, published 2014-03-12, figures re-read at source 2026-09-09. Vendor-reported, and twelve years old: use it for scale only.
What was deliberately left out
Several figures that would have supported the case were excluded because they did not survive checking: a browser-support percentage for Add to Home Screen that is technically sourced but substantively misleading about current iOS; three sets of app retention benchmarks that are vendor-reported with no published method; a widely circulated cross-vendor retention benchmark that appears not to correspond to any real publication; and several install-abandonment statistics with no traceable original. Two academic papers that might contain a genuine independent adoption figure are cited nowhere on this site because they have not been read.
The case for this category is architectural and checkable rather than statistical. If a genuinely independent measurement of attendee adoption is published, it belongs here, whichever direction it points.
Last updated 2026-09-09. Corrections, including sources for anything listed above as unverifiable, are welcome: see the editorial policy.