I'm Cristobal, and I build Dropr, a Shopify recommendations app. I wanted to know why store owners give up on apps like mine, so I collected the one-to-three-star reviews of the apps in the same category with a script and had an AI model sort them by main complaint. This post is what they say. I don't name apps or publish rates per app, because the samples per app are small and capped, and this isn't a ranking.
The complaints, by share
Of the 879 reviews, 241 (27%) say the app didn't work or didn't show on the store. Price comes next with 144 (16%), then support with 83 (9%), setup being too hard with 50 (6%), the store getting slower with 37 (4%) and design with 15 (2%). The remaining 309, about a third, have no clear main complaint: short reviews like "not for me", feature requests, or a mix of small problems.
| Main complaint | Reviews | Share of 879 |
|---|---|---|
| Didn't work or didn't show | 241 | 27% |
| Price or billing | 144 | 16% |
| Support | 83 | 9% |
| Too hard to set up | 50 | 6% |
| Slowed the store | 37 | 4% |
| Design | 15 | 2% |
| No clear main complaint | 309 | About a third |
Most of these are one-star reviews: 508 of the 879 (58%) gave one star, 164 gave two and 207 gave three. Price complaints were the harshest, with 111 of 144 (77%) at one star.
How long merchants used the app before reviewing it
Next to every App Store review, Shopify prints how long the merchant had been using the app, for example "About 1 hour using the app". That field tells you when things went wrong.
212 of the 879 reviews (24%) were written less than a day after the merchant started using the app, and 301 (34%) within the first week. At the other end, 265 (30%) came after six months or more.
The two ends tell different stories. Setup complaints come early: 27 of the 50 (54%) were written in the first week, often after a few hours of trying. "Didn't work" complaints are split: a third (33%) in the first week, and 37% after six months or more. Many of those late reviews describe an app that used to work and then stopped: after a Shopify or app update, after a theme change, or when settings reset without warning.
What the merchants wrote
The "didn't work" reviews describe recommendations or bundles that never appeared on the store, prices that didn't load, discounts that didn't apply, and conflicts with a specific theme or cart. One merchant, two hours in: "Tried everything to get the upsell to work but nada every time." (If that's where you are, the recommendations not showing checklist covers the usual causes.) 57 of the 241 also mention support, usually because the problem stayed unsolved.
Setup reviews are about how much there is to configure before anything shows. "The whole thing is overly complicated when it doesn't have to be," wrote one merchant after about 12 hours with an app. Another, three hours in: "Nothing makes any sense and all options are hidden."
Price reviews are mostly about charges the merchant didn't expect rather than the list price: being billed while not using the app or after uninstalling, a "free plan" that turned out to be a trial or to leave out the feature they wanted, usage fees, or fees counted on all store orders instead of the orders the app helped. Across all 879 reviews, 82 mention a charge, a bill or a refund. The pricing models post shows how the main apps charge.
Support reviews are about waiting: days between replies, or a fix promised and not delivered. Speed reviews describe product pages that got slow or glitchy; three of the 37 describe a checkout that stopped working until the app was removed. The speed post explains how to measure an app's effect before and after you install it.
What I take from it as a store owner's checklist
If you're about to install an upsell or cross-sell app, test for the problems people actually hit. Check that the app shows on your own theme on the first day, including on mobile. Work out the price at your order count in your busiest month. Send support a question during the trial and time the reply. Count how many settings you have to fill in before anything appears. And check that the app uses Shopify's app blocks or app embeds, so removing it removes everything. The guide to choosing an upsell app turns these into steps on the App Store listing.
Dropr already sets up its recommendations on install with nothing to configure, uses only app blocks and an app embed, and has one paid price. These reviews are a list of ways it can still fail. It went live in July 2026 and has no reviews of its own yet.
How the reviews were collected and sorted
I started from 8 well-known upsell apps and added the apps listed in the Shopify App Store's upsell and cross-sell category, then a script read their public review pages in October 2026. For each app it took the newest 100 reviews rated 1 to 3 stars, or all of them if the app had fewer, so the sample leans recent. I left out 22 apps the script flagged as off-topic for this question (page builders, shipping protection, chatbots, cash-on-delivery order forms, gift wrap and similar), which leaves 879 reviews of 58 upsell, cross-sell and bundle apps.
An AI model read each review and picked one main complaint from a fixed list, or "no clear category". A model can misfile a review, especially a short or mixed one, so treat the shares as rough. 86% of the reviews are in English, and 70% are dated 2025 or 2026 (the oldest is from 2013). Two free apps made by Shopify, Search & Discovery and Shopify Bundles, contribute 100 reviews each; without them the order stays the same (didn't work first, then price, then support), and price rises to 21% because nobody complains about the price of a free app.
Bad reviews are a self-selected group. These numbers say what unhappy merchants write about, not how often apps fail.
Frequently asked questions
Which upsell app has the fewest complaints?
I don't rank apps from this data. With at most 100 bad reviews per app and very different install counts, a per-app rate would mislead. Look at the newest one-to-three-star reviews of the apps you're considering instead.
Are upsell apps generally unreliable?
This data can't say. It only contains bad reviews. It shows what goes wrong when something does, which is useful for deciding what to test during a trial.