Every tool is also a cost
The Shopify App Store makes adding a tool feel free. Most have a free tier, installation takes a minute, and each one solves a real problem you actually have. So stores accumulate, and the accumulation is rational at every individual step.
The cost that does not appear on any invoice is that every tool is another place to look. A team with nine apps is not paying nine subscriptions, it is paying attention nine times, and attention is the scarcer resource in a store with three people in it.
This is why the question is not which app is best at a job, but whether a job needs its own app at all.
The test
Before adding anything, three questions. A tool should pass all three.
1. Does this replace something, or add to it? Tools that replace are nearly always worth it. Tools that add need a much higher bar, because the cost is permanent and the benefit is usually specific to the situation you are in right now.
2. Will somebody open this weekly without being reminded? If the honest answer is no, it will be abandoned, and an abandoned tool is worse than none because it holds stale data somebody may act on.
3. What happens if we do not? Often the answer is that a job stays slightly annoying, which is an acceptable outcome. Slightly annoying beats another destination to check.
What a small store genuinely needs
There is a small set of jobs where not having a tool actually costs money rather than convenience.
- Somewhere the team coordinates. Not optional past one person. It can be a group chat, and for many stores that is enough.
- Something that tells you when stock is running out. Manual checking works until it does not, and the failure is expensive.
- Something that flags risky orders. Shopify does a version of this already. The question is whether the flag reaches a person who can act before dispatch.
- A place for procedures. Only once somebody other than the owner does the work. Before that it lives in one head and that is fine.
- A place for tasks with owners and dates. Once "I thought you were doing that" has happened more than once.
That is five jobs, and importantly it is not five apps. Several overlap, and the useful question is how few tools cover the jobs you actually have.
The overlap nobody checks
The most common cause of an oversized stack is buying twice for jobs that one tool already covers, because the second purchase was triggered by a specific frustration rather than a review of what you had.
Worth checking whether you have two of any of these:
- Two places messages happen. Extremely common, usually a general chat app plus something store-specific, and it means every conversation starts with a decision about where to have it.
- Two places tasks live. A project tool plus a task feature inside something else. One of them is stale and people do not always know which.
- Two alerting paths for the same event. An app sending low stock to email and another sending it to chat. The second one taught everybody to ignore the first.
- A tool doing a job your existing tools already do adequately, bought because it does that job slightly better. This is the hardest one to resist and the most common source of stack creep.
A reasonable shape by team size
Rough, and the sizes matter more than the specifics.
One or two people. Shopify plus a group chat, plus stock alerts if you carry inventory. Almost anything else is anticipating a problem you do not have yet.
Three to five. Add somewhere tasks have owners and dates, because this is the size at which things start being dropped between people. This is also where a store-aware coordination tool starts earning its place over a general chat app, because the volume of "which order do you mean" has become a real daily cost.
Six to ten. Add a place for procedures, because you are now hiring often enough that explaining verbally does not scale, and access control becomes something you have to think about rather than something that takes care of itself.
Beyond that you are no longer a small team and most of this advice stops applying.
Removing one
Worth doing once a year, and it is easier than it sounds. For each app, ask when somebody last opened it deliberately, as opposed to receiving a notification from it. Anything nobody has opened in a month is a candidate.
The one to look hardest at is whichever tool duplicates a job another tool already does. Not because the weaker one is bad, but because the duplication itself is the cost: two places for one job means every person makes a small decision every time, and sometimes they make it differently.
The free tier is not free
Most stack creep comes in through free plans, because a free plan removes the only checkpoint most stores have. Nobody convenes a discussion about whether to install something that costs nothing, so nobody applies the three questions, so it gets installed.
The costs a free plan does not remove are the ones that matter. It is still a destination to check, still a place data lives, still something holding permissions to your store, and still something that will need removing later by somebody who has to work out whether anyone depends on it.
The free tier also has a specific failure pattern worth knowing. Free plans are usually limited in a way that bites once the tool is embedded, and by then switching costs something. That is not a criticism of the model, it is how it is supposed to work, but it means the decision to install should be made as if you were on the paid tier, because that is where you will end up if the tool succeeds.
Trial hygiene
If you are going to try something, decide two things before you install it.
What would make this a yes. One sentence, written down. Not "see if it helps", which is unfalsifiable and means the trial ends by being forgotten. Something you could actually check: orders stop being missed, nobody asks which order that is, the Monday count takes half as long.
When you will decide. A date, in the calendar, with the uninstall as the default. A trial with no end date does not end; the app simply becomes part of the stack by attrition, which is exactly how you got nine.
Trialling one thing at a time matters more than it sounds. Two simultaneous trials mean you cannot attribute an improvement to either, so both tend to stay, and you have doubled the stack to solve one problem. The same logic applies to trialling during an unusual week: a tool tested during peak tells you how it behaves at peak and nothing about a normal Tuesday.
Uninstalling properly
Removing an app is not always just removing an app. Some write data into your store, adding tags, metafields or theme code, and uninstalling can leave that behind or take something with it that you wanted.
Two minutes of care: check whether anything you rely on came from that app, export anything worth keeping, and look at the theme afterwards if the app touched the storefront. It is a small job that occasionally saves a large one, and it is the reason a stack should be pruned deliberately once a year rather than in a panic.
Related reading: our comparison of the apps in this category covers which tool for which job, and whether you need a collaboration app at all is the version of this question for our own product.
Common questions
How many apps should a small Shopify store have?
Fewer than it has, usually. The better measure is jobs rather than count: a store needs coordination, stock warning, risky-order flagging, and later procedures and owned tasks. How many apps that takes depends on overlap.
What is the real cost of adding another tool?
Another place to look. The subscription is visible and usually small; the attention is invisible and is the scarcer resource in a three-person team.
When does a small team need a task tool?
Around three to five people, or as soon as "I thought you were doing that" has happened more than once. Below that, coordination is nearly free and a task tool is overhead.
How do I know if a tool is worth keeping?
Ask when somebody last opened it deliberately rather than clicking a notification from it. Anything nobody has opened in a month is a candidate for removal, and duplicated jobs are the first place to look.
Is it better to have one tool that does several jobs adequately?
Usually, at small team sizes. A tool that is the best at one job but is a ninth destination often loses to a tool that is second best at three and is somewhere people already are.