All 107 of them. The most common complaint is not a bug — it is nobody answering.
We are about to ask apparel merchants to install a ninth inventory app. Before doing that it seemed worth finding out what the people who installed the other eight were angry about — not the average, which is mostly people who are content, but the tail, which is where a product's actual failure modes are written down by the people they happened to.
So: a census, not a sample. Every one-star review on the public Shopify App Store listing of eight apps, read in full, tagged one by one.
The frame is apps a merchant installs because a size sold out, or because they cannot decide what to reorder. That is the shelf Sizecurve is asking to sit on. It was written down in apps.txt before a single review was fetched, because a list of competitors assembled after skimming the complaints is a list assembled to support whatever the complaints happened to say.
Deliberately excluded, and named so you can disagree: stock-sync and report-pundit (ERP plumbing and general reporting — a different job), peel-insights and lifetimely (customer analytics). Their one-star reviews would be about someone else's product.
| App | rating | reviews | one-star | share |
|---|---|---|---|---|
| Stocky — Shopify's own | 3.0 | 211 | 59 | 28.0% |
| Back in Stock | 4.9 | 940 | 21 | 2.2% |
| Inventory Planner | 4.5 | 152 | 12 | 7.9% |
| Assisty | 4.8 | 358 | 7 | 2.0% |
| Fabrikator | 4.8 | 112 | 3 | 2.7% |
| Prediko | 4.9 | 249 | 3 | 1.2% |
| Inventora | 3.8 | 8 | 2 | 25.0% |
| Cogsy | 4.9 | 13 | 0 | 0.0% |
The share column is there because the count on its own is a popularity ranking wearing a lab coat. 21 one-stars out of 940 and 59 out of 211 are not the same fact about a product.
A ninth slot was held for a pre-order app — what a merchant does when a size is gone — and it is empty. The obvious candidate is delisted and we could not find a live one on the shelf without guessing at slugs, so the row was dropped rather than filled with something invented.
Each review carries one or more tags. The categories were written after the first read-through and applied on a second pass — the wrong order for a hypothesis and the right order for a description. Every call is recorded against a review id, so any single one can be disputed without re-reading the corpus.
| Complaint | all 8 apps | excluding Stocky |
|---|---|---|
| nobody answered, or the answer was useless | 39.3% | 56.2% |
| it did not work, or stopped working | 35.5% | 35.4% |
| it did not do what the listing said | 21.5% | 16.7% |
| the price rose, or the contract closed | 14.0% | 31.2% |
| it ran, and the figures were wrong | 13.1% | 8.3% |
| visibly abandoned by whoever owns it now | 10.3% | 4.2% |
| something they relied on was removed | 8.4% | 4.2% |
| locked behind another paid tier | 7.5% | — |
| too slow or laborious to use in a working day | 6.5% | 8.3% |
| uninstalling did not uninstall it | 3.7% | 6.2% |
| never got it running at all | 2.8% | 4.2% |
Stocky is 55% of the corpus and it drags the whole table, so the second column drops it. Stocky's complaints are unusual in a specific way: 8 of its 59 one-stars are about being gated behind a $89/month POS Pro subscription, which is a pricing decision inside Shopify and has no equivalent anywhere else on the shelf. Take it out and the shape changes — price more than doubles, and support becomes a majority complaint.
More than half of the one-star reviews outside Stocky say some version of the same sentence: I wrote, and nothing came back.
"I have sent two emails regarding an issue with the app and never got a response back from them."
— NoticeMe, Netherlands, over a year using the app
"If you need help you are on your own, they recommend that you contact a developer — it is not fair to call it support, sorry."
— BODYOUTLET, Sweden, over a year using the app
"There is no way to contact support."
— SWWSupplies, United Kingdom, four months using the app
This is not what we expected to find. An inventory app is a forecasting product, and the obvious guess is that its one-star reviews are about bad forecasts. Wrong numbers are there — 13.1% — but they are the fifth complaint, behind an unanswered email, a thing that broke, a listing that overpromised, and a bill that went up.
One review in the corpus describes the failure that costs the most, and it is four sentences long:
"stopped working: collecting and sending notifications six months ago. Only realized it today… they were billing us for 6 months while the app didn't do anything!"
— iLevel Lab, United States, almost six years using the app
Six years of trust, six months of nothing, and the only reason it ended is that somebody happened to look. An app that breaks loudly gets uninstalled on Tuesday. An app that breaks quietly gets paid.
That review changed our code this week. See below.
Excluding Stocky, price is the third complaint at 31.2% — and almost none of it is "too expensive". It is about the price moving after the merchant had migrated their data in.
"Held hostage on a 12 month contract… was paying about $300/month and they jacked it up over $600/month and required us to sign a 12 month contract. All this during our busiest time of year here in Q4."
— Parker Clay, United States, about four years using the app
"I've also heard sketchy sales practices on raising prices drastically once you are on. (As they know it's hard to stop using it once you have migrated all orders, data, etc.)"
— Amberjack, United States, five months using the app
And one merchant was told they had become too small to keep:
"I was told that my company was too small for their service… I was advised to contact them again once I reach more than 1 million dollars in annual revenue. They should clearly state these conditions on the App Store so smaller businesses don't waste their time."
— ergoPouch Chile, Chile, seven months using the app
Four reviews, and the smallest category that still matters, because it is the one a merchant discovers when they are already leaving:
"When I uninstalled the app, there was still code in my theme… This code prevented my theme's back in stock functionality from working, so now I have to go through my code to figure out what is going on."
— Nootkas, United States, two months using the app
"Although uninstalled, it still adds stock quantities under product 'Incoming' and messes with our inventory… It's like Hotel California, you can get in anytime you want, but you can never leave."
— MT Health GmbH, Germany, twelve months using the app
The cliché about one-star reviews is that they are written in the first ten minutes by someone who never read the setup guide. On this shelf they are not.
Three quarters of these people used the product for months or years before giving up on it. Sixteen went back and edited a review they had already written. The longest is 294 words. They span 21 countries and eight years, and 56 of the 107 were written since the start of 2024.
A competitor reading competitors' reviews and concluding that it would never do any of that is worthless. So here it is split three ways: what is structurally impossible, what is a choice already made, and what was actually broken in our code until this week.
Leftover code. Sizecurve injects nothing into your theme — no script tag, no snippet, no app block, no widget. It cannot leave code behind because it never put any there. Uninstalling deletes the token, the cached analysis and the change history on Shopify's app/uninstalled webhook. Wrong stock counts. It holds no write scope of any kind, so it cannot touch your inventory, correctly or otherwise.
Two plans, said plainly. Free to find broken size runs and returns on every style. $29 a month adds size curves and what to reorder, in which sizes, with a 14-day free trial, billed by Shopify on your existing invoice. No annual contract, no minimum revenue. There is no migration — it reads what Shopify already has — so there is no moment where we know you cannot leave.
Support is one person and the listing says so. We are not going to promise a one-hour SLA we cannot keep at this size; 56% of these reviews are what happens when someone promises and then does not. Replies to the weekly email go to a human, and the honest commitment is that you always get an answer, not that it is instant.
The iLevel Lab review sent us to look at our own Monday job, and it had exactly that failure mode. If reading a store threw — an expired permission, a Shopify outage, anything — the sweep logged a line to a console nobody reads, moved on to the next shop, and the merchant heard nothing. Their dashboard would sit there showing last week's numbers. The $29 would keep going out.
Now a failed sweep is counted against the shop, and on the second consecutive failure Sizecurve emails the merchant to say it cannot read their store — what it is reporting, when it last worked, and this, verbatim:
If we cannot fix it, uninstall: if you are on Standard, Shopify stops the $29 the moment the app is removed, and we would rather that than bill you for a week we did not work.
It sends on the second failure rather than the first because one bad Monday is a network or a rate limit, and two is us. A read that succeeds but finds no sized styles counts as a failure too: the app produced nothing that week and the merchant is still paying, which is the same fact wearing a better coat.
One-star reviews are the complaint tail, by construction. Back in Stock has 940 reviews and a 4.9 average; the 21 people quoted here are 2.2% of them, and the other 919 are not in this piece. Nothing above is a verdict on whether any of these apps is good. It is a description of how they fail when they fail, which is a different and more useful question when you are deciding what to build.
Two more limits worth stating. We are a competitor, which is a conflict of interest — the mitigations are that the app list was frozen before reading and that every tag is published against a review id, so you can check the framing against the corpus rather than trusting it. And these are the reviews people wrote down: a merchant who quietly uninstalled on day two is not here at all.
Public listing pages only. robots.txt fetched and obeyed, a user-agent naming this app and a contact address, one request at a time with a six-second floor, no account and no login. Every page of HTML was cached to disk before anything was parsed out of it, so re-reading the corpus with a fixed parser costs Shopify nothing. A 429 is waited out, not recorded as an answer — it is a statement about our rate, not about whether we may read the page.
And the parser was wrong the first time. Our date pattern did not allow for the word Edited, which Shopify prefixes onto any review the merchant went back and changed — so 16 of 107 came out with no date, and we nearly published "16 undated" as a fact about Shopify's markup rather than about our own regex. Fixed off the cache, without re-fetching anything. Whether a review was edited turns out to be worth keeping on its own.