We build our own software, so people expect us to argue for building. We mostly do not, and the reasoning is worth setting out because getting this wrong is expensive in both directions.
What an app actually costs
Not the monthly fee, or not only. Three things:
The subscription, compounding. Fifteen apps at fifteen to fifty dollars a month is a real line item, and it grows because apps get added and almost never removed. Nobody remembers to uninstall the one that was replaced.
Front-end weight. Many apps inject scripts into your storefront. Each one is another request, another thing parsed before the page becomes usable, and it is cumulative in a way nobody notices because it happens one install at a time. The store that feels slow rarely got slow from one decision.
A dependency you do not control. Apps get deprecated, sold, repriced, or broken by a Shopify platform change. When that happens the problem is yours and the fix is on someone else's schedule.
What custom actually costs
More than the build. Custom code has no vendor maintaining it against platform updates, which means it is your job forever, and the person who understands it may not be around in two years. A custom solution nobody can maintain is worse than an app you resent paying for.
The build is also usually the cheap part. The expensive part is every edge case discovered afterwards, all of which an established app already handles because thousands of stores found them first.
How we actually decide
One question: is this problem specific to how this business works, or is it a problem thousands of stores have?
If thousands of stores have it, an app exists, it is better than what you would build, and it is cheaper. Reviews, subscriptions, upsells, loyalty, currency, basic bundles. Building these is a way of paying more for less.
If the problem is specific to your operation, no app fits properly, and you will end up with two apps and a spreadsheet bridging them. That is the signal to build.
The tell is not complexity. It is whether the workflow is a competitive advantage or just plumbing. Plumbing should be bought.
The middle option people forget
A lot of what gets solved with an app is a theme edit. A trust badge, a delivery estimate, a size guide, a custom section: these are template work, not features, and they add no subscription, no external script and no dependency.
Somewhere between a quarter and a third of the app installs we find on stores we take over could have been thirty lines in the theme. Nobody set out to do it that way. Each one was the fastest fix available in that moment.
The audit worth doing once a year
List every installed app, its monthly cost, when it was last opened, and what would break if it were removed. Anything nobody can answer the last question about should be removed on a quiet Tuesday, having taken a theme backup first.
On stores we have taken over, this exercise has consistently removed apps that were still being paid for and no longer did anything, usually because they were superseded by something else and never uninstalled.
We do this as part of taking on a store, and it is the rare piece of work that pays for itself in the first month. It is covered under Shopify development, along with the theme work that replaces some of what comes off.
