Microsoft Keeps Pulling Users Back to Edge—and That Is a Web App Distribution Problem
An installable web app can have a flawless manifest, a resilient service worker and an interface that feels every bit as deliberate as native software. None of that guarantees a fair route to the user. Distribution begins one step earlier: will the operating system let people reach and keep the browser they chose?
A newly published Mozilla Research investigation argues that Windows repeatedly fails that test. Its findings matter beyond the old Edge-versus-Firefox rivalry. The browser is where people discover web apps, grant permissions, encounter installation controls and often run the installed product. When an operating system redirects that journey, it is influencing the web app market before a developer’s product even appears.
What the researchers found
Over the Edge 2.0 is a Mozilla-commissioned follow-up by independent researchers Harry Brignull and Cennydd Bowles. They examined important browser-choice journeys on Windows 10 and Windows 11 in the United States, United Kingdom, India and Germany, with Germany representing the European Economic Area.
The authors say they encountered techniques including obstruction, preselection, nagging, visual interference and misleading wording. Their conclusion is carefully framed as an expert assessment: Microsoft does not let people download an alternative browser, make it the default and continue using it without what the researchers classify as harmful interference.
Several findings are especially relevant to web distribution:
- Downloading a rival browser can trigger retention tactics. The researchers documented Edge or Bing presenting prominent messaging intended to keep users with Edge, including browser-injected messaging on Chrome’s download page outside the EEA.
- A Windows migration can undo an existing choice. In the tested Windows 10-to-Windows 11 backup and restore journey, the alternative browser was not transferred and Edge became the default.
- Copilot can bypass the selected default. Links opened from Microsoft’s assistant went to Edge in the researchers’ tests rather than to the browser the user had chosen.
- Geography changes the experience. The researchers found fewer of these patterns in the EEA and attribute the difference to regulatory pressure, although they say some problems remain there too.
The evidence needs the right caveats. Mozilla competes directly in the browser market and commissioned the work, even though the authors state that the findings are their own. The research documents selected interface journeys rather than measuring the behavior of every Windows user, and browser experiences can change through updates, experiments and regional configuration. The Register said Microsoft did not respond to its request for comment.
Why this is a web app distribution story
No service worker API changed this week. No manifest field was added or removed. That does not make the report peripheral to installable web apps. It exposes a distribution layer that technical PWA checklists usually ignore.
The browser is part storefront, part installer and part runtime. It decides how installation is surfaced, how an app is represented in browser-managed app lists and how links, permissions and updates connect back to the web. Support and presentation vary by browser and platform, even when two browsers share substantial engine technology.
That means a default-browser decision can affect three points in the product funnel. First comes discovery: which browser receives a link from search, an assistant or another application. Second comes conversion: whether the user sees understandable installation controls and trusts the surrounding interface. Third comes retention: whether later links and operating-system actions return the user to the environment where the app was installed and understood.
Microsoft is not blocking standards-compliant web apps in the behavior described by this report. The problem is subtler. If Windows repeatedly routes attention toward Edge, Edge gains more opportunities to define how the web-to-app transition feels. Rival browsers receive fewer opportunities to compete on installation, privacy, permissions or app management. A theoretically open distribution channel can therefore acquire a powerful practical gatekeeper.
Who should care
Founders and product leaders should treat browser and operating-system context as acquisition infrastructure. A visit arriving through Edge on a new Windows installation may have a different history from one arriving through a deliberately selected browser. Browser share is not always a clean expression of preference; defaults, migrations and system-owned entry points can shape it.
Developers should test more than the installed app window. The important journey begins with an ordinary link and includes first launch, installation, deep links, default-browser changes and device migration. Installation guidance should remain accurate across browsers and should not assume that everyone has arrived through the browser they normally use.
Growth teams should segment installation and repeat-use data by browser, Windows version and region where privacy rules permit it. A sudden change in conversion may come from the surrounding distribution surface rather than from a manifest deployment or application release.
Investors and distribution partners should recognize browser choice as platform risk. The EEA comparison is particularly revealing: one vendor can ship materially different choice architecture in different jurisdictions without changing the web application itself. Market access can therefore depend on regulation and operating-system design as much as on browser capability tables.
What changes for web apps now
The immediate answer is strategic, not technical. Teams do not need to rewrite their service workers because of this report. They do need to stop treating the default browser as neutral plumbing.
A resilient web app distribution plan should:
- measure the full route from external link to repeat standalone use;
- test installation language and controls in each browser that materially serves the audience;
- verify behavior after Windows upgrades, device replacement and restored backups;
- avoid browser-specific claims unless the experience has actually been detected;
- keep a normal, high-quality browser experience available when installation is hidden or declined.
This is also a reminder of the web’s central advantage. A web app remains reachable through a URL even when the installation path is imperfect. That fallback preserves access—but it does not erase the commercial value of a smooth, trusted route into an app-like experience.
What IndApp watches next
The most important signal will be whether Microsoft answers the specific findings and changes the documented journeys. Four tests matter: whether Copilot begins honoring the default browser, whether Windows migration preserves the user’s browser choice, whether browser-download interventions disappear, and whether protections already visible in the EEA spread to other markets.
Regulators outside Europe are another watch point. The report’s regional comparison suggests that policy pressure can change interface behavior. If other authorities pursue equivalent remedies, browser distribution on Windows could become more consistent across markets.
Installable web apps do not need a single corporate store to reach users. But they do need operating systems to respect the route users select. The open web promise is not only that any developer can publish. It is also that no hidden hand should decide which browser gets to deliver the app.
Further reading
Recent news
