The threshold that determines this is frequency. Applications earn their place when somebody uses them weekly or more, because the convenience of an icon on a home screen compounds across many visits. For anything used a few times a year, the installation is a larger cost than every visit it saves, and most people simply decline.
The abandonment problem is worse than the installation one. Even where somebody installs, applications are used a handful of times and then forgotten, and a business that spent months building one frequently finds its actual usage is a small fraction of its website traffic. That outcome is common enough to expect rather than to guard against.
The cost is the other half of the argument and it is routinely underestimated. Building for two platforms, submitting to two stores, maintaining compatibility as operating systems update, and pushing changes through a review process is a permanent commitment rather than a project. A business that builds one and stops maintaining it has something visibly broken with its name on it.
What people usually want when they ask for this is achievable without it. A site that works properly on a phone, loads quickly, and lets somebody complete the thing they came for. A progressive web application, which can be added to a home screen without a store, covers the remaining convenience for a fraction of the cost.
Where an application genuinely makes sense is recognisable. Frequent repeat use, a need for genuine offline function, hardware access the browser cannot provide, or a loyalty mechanism where the notification is the product. Those are real cases and they are narrower than the enthusiasm suggests.
Notifications are the reason most businesses actually want one, and it is worth being honest about that. They are a direct channel to a customer's attention, they are also the fastest route to being uninstalled, and browsers and messaging now provide most of the same reach without the barrier.
Ask your customers rather than assuming either way, since the answer is specific to your business. Five conversations establishes whether anybody would install it and what they would use it for, and the common finding is that they want something the website should already do.
Then fix the mobile experience first regardless of what you decide, because that is where the majority of your visitors already are. A business considering an application while its site is slow and awkward on a phone is proposing to build a second thing rather than fix the first.
Check whether your competitors have one and whether anybody uses it, since the review counts and update dates on the store listing tell you honestly what the category actually supports.
Consider what the maintenance costs annually before building anything, since operating system updates alone require periodic work and an application nobody maintains is removed from the stores eventually.
Check your existing mobile traffic and what those visitors do, since a business whose phone visitors already complete what they came for has answered this question without building anything.
Ask what problem the application would solve rather than whether it is expected, since the useful answer is usually a specific frustration with the current experience that can be addressed directly.
Build the web version properly first even if you eventually want an application, since the site is what somebody sees before they would ever install anything and it has to work regardless.