Mirakle Research · September 12, 2026
When an app leaves the Shopify App Store, does it come back?
Since 4 August our scraper has kept every verdict a Shopify App Store app page gives it. In the 38 days since, an app that had been listed was replaced on its own page by a notice that it was not currently available, or by a missing page, 495 times, across 493 apps. That is 15.3 departures a month for every 1,000 apps listed. By the end of the window 153 of them were listed again (which is all this report means by coming back), a median of 7.4 days after they left, and 342 were not. This report is about which ones came back. Before that question can be asked, the 495 have to survive a duller one: whether they are departures at all, or readings of ours.
The first suspect
What our own scraper looks like when it is wrong
A departure is our reading of a page, so the reading has to be ruled out first. Three Shopify surfaces can tell us an app is gone: the app's own page, its review page, and the sitemap that lists every app. We read all three, and we can misread all three.
We recorded 499 absences that only the sitemap saw. 474 of them ended, and every one ended between 1.3 and 2.8 days after the last listed reading. 59 were seen only on the review page, and the 43 that ended did so within 1.6 days. Nothing about an app leaving the store would make it come back after the same two days. Our schedule would. An app is read again a day or two after the reading that missed it, and found listed. Two surfaces with the same fault leave the same mark, and these two do, so the fault is in how we read them.
The absences Shopify's app page reported do not have that shape. The shortest ended 1.7 days after the last listed reading, and they are spread across the whole window. A middle tier of 174 absences, seen by the review page and the sitemap together but never checked on the app page, cannot be sorted: 30 of its 90 returns fall inside our schedule's window. It is left out, and including it would move the share listed again from 30.9% to 36.3%.
That is the licence for everything below. From here, a departure means the app page said so: 486 times "not currently available" and 9 times a missing page. The sitemap agrees with the page once it is read on its own terms. Of the 330 departures still open that a fresh sitemap has been read across, 320 (97.0%) have been missing from every one of those reads.
A page that says an app is not currently available rules out our scraper. A missing page does not quite, because it leaves a duller explanation standing.
The second suspect
A missing page, and a new address
An app that changes its web address looks, from the old address, exactly like an app that left. 50 absences were missing pages. For 41 of them (82.0%), the same developer's name appeared on an app we first saw within 14 days, or an app with the same name is live now. The test is loose, so it was run on the 486 "not currently available" departures as well, where a move is not in question. It finds a new address for 3.7% of them. The 41 are moves, and they are left out of everything below.
That leaves departures in which the listing itself went away. When one of those comes back, the next thing to know is what comes back with it.
What a return brings back
The review count on either side of the gap
Shopify prints a review count on every listing. For the 153 departures that ended, we have that count from the last listed reading before the app left and from the first reading after it came back. 120 were identical, 26 were higher and 7 were lower. Added up, the apps left with 19,656 reviews and returned with 19,768; the reductions come to 48 reviews in all.
A departure, on this evidence, is not a deletion. Whatever takes the listing off the store leaves its reviews where they were, and the listing comes back whole. So any app could in principle return, and the question worth asking is which ones do.
Cut one · by reviews
Which apps are listed again
The first cut is by the one number every listing prints. 74.7% of departures were apps with no reviews, against 57.3% of the apps listed today. They are also the ones least often listed again. Within 14 days 21.3% of the unreviewed apps were back, against 54.6% of apps with at least one review. The curves keep separating after that: by 28 days the estimate for apps with 100 or more reviews is 87.1%. By then only a handful of those 33 apps are still being followed, so we lean on the 14-day reading.
The rival · age
An app with no reviews is usually a new app, and a new app is the kind most likely to be abandoned, so age could be drawing the same line. The grid holds age still. Among apps more than two years old, 18.4% of unreviewed departures are listed again against 50.0% of reviewed ones. Among apps under six months old it is 23.4% against 37.5%. Age does not account for the gap.
Something separates the apps that come back from the ones that do not, and it follows the review count more closely than it follows age. The next question is whether the thing being separated is the app at all.
Cut two · by developer
Apps that leave together
If the app is not the unit, the nearest other unit is the developer, and developers leave in groups. 105 departures, 21.2% of the total, belonged to 29 groups in which one developer lost two or more apps on the same day. We call a group like that a shelf. The largest held 13 apps.
If the apps on a shelf came back or stayed away independently, at the rate of single departures from the same fortnight, about 15.4 of the 29 shelves would have split, with some apps back and some not. None did. 11 came back whole, 10 of them on a single day, and 18 have not come back at all. A shelf is defined by its apps leaving together, which says nothing about whether they return together, so that part is not built into the count. The remaining rival is time: a shelf that left last week has not had long to return. That explains why a shelf might still be gone. It does not explain why no shelf is part gone.
The shelves also bear on the first cut. 78.1% of shelf departures were apps with no reviews, so shelves staying away together could have been doing some of the work of the review gradient. Among the 390 single departures alone, 20.7% of unreviewed apps were back within 14 days against 49.1% of reviewed ones. The gradient does not come from the shelves.
For these apps, the thing that returns is the shelf and not the app. That describes an outcome. It does not say who decided it, and nothing we can see does. It also cannot say for how long a shelf that stays gone will stay gone: 38 days is enough to watch one come back, and not enough to wait out one that does not.
Cut three · by time
What the longer record shows
A longer wait needs an older and weaker instrument. Our daily snapshot log goes back to 27 April. It was not built to record departures: it writes a row when a listing is read successfully, so a departure shows up as missing rows, and so does a day our scraper did not reach an app.
Before using it we scored it against the app page over the weeks both exist. Of 159 gaps it records in that period, 76.1% match a departure the app page reported and 99.4% match an absence some Shopify surface recorded. It finds 79.1% of the returns the app page saw. It is a usable instrument with a known error, and nothing below leans on it harder than that.
Over 137 days it records 1,202 departures, after the same new-address rule removes 278. The share listed again rises quickly and then flattens: 35.3% by 30 days, 42.8% by 60 and 49.6% by 120. The longest absence that ended lasted 100 days. And it puts the review bands in the same order the app page did: by 60 days, 33.0% of unreviewed apps against 91.4% of apps with 100 or more reviews.
Two instruments built in different ways, over windows of different lengths, give the same ordering. That is one finding measured twice, not two findings, and it is sturdier for it.
The log also answers whether August and September were an episode. In every month it holds with at least 20 usable scan days, departures ran between 9.0 and 16.7 a month per 1,000 apps. Leaving and coming back is a standing condition of the store, not something that happened to it this summer.
What survived
What we can say, and where it stops
An app leaving the Shopify App Store is ordinary and often temporary. 15.3 a month leave for every 1,000 listed, while in the same weeks we saw 2,518 apps for the first time, 5.1 for every departure. In the longer record 49.6% are back within four months, and the returns we could compare came back with their reviews. Which ones come back is not even. Apps with reviews come back far more often than apps without, at every age we could measure, and apps that leave as a shelf come back as a shelf or not at all.
"Not currently available" does not say who made the app unavailable, a developer or Shopify, or why, and a return does not say what changed. What we have instead is a shape: listings kept whole while they are away, reviewed apps brought back more often than unreviewed ones, shelves that return on a single day or not at all. A shelf of 13 apps from one developer coming back on a single day is the kind of trace a decision about an account would leave, though we cannot see the account or the decision. The shape is not standing in for a reason we failed to find; it is the only part of the reason that reaches a public page.
Method
How this was measured
The window runs from 5 August 2026 to extraction. Our scraper records every verdict an app's page returns: listed, not currently available, or gone (a 404). An absence is a stretch between a listed reading and the next listed reading; it has ended when a later reading is listed and is open otherwise. A departure is an absence during which Shopify's own app page said the app was not currently available, or during which the page was gone and no new address was found. A 404 counts as a new address when the same developer's name appeared on an app we first saw within 14 days of it, or an app of the same name is live. Absences seen only on Shopify's sitemap, only on an app's review page, or on those two together are reported but are not departures. A sitemap read that repeats the exact contents of an earlier read is treated as a stale copy and ignored. Time until return is measured from the first reading that found the app gone. Return shares are Kaplan-Meier estimates, so absences still open at extraction count as unfinished rather than as permanent. The longer history uses the daily snapshot log from 27 April 2026: a gap of 4 or more usable scan days between two readings, or a final gap of that length for an app not live now, with new addresses removed by the same rule. A usable scan day is one on which at least 92% of the fortnight's peak number of apps was read. Review bands use the count Shopify printed on the last listed reading before the absence.
Limits. Everything here is our reading of public pages, not of Shopify's systems. We do not see who took a listing down or why, whether merchants could still use an app while its listing was gone, or how many had it installed. Our scraper stops reading an app's page once it has left both the sitemap and the live store, so an app whose page came back without its sitemap entry would be counted as still gone, and a departure short enough to fall between two of our readings is not counted at all. Both push the return figures down. The 15.3 departures a month per 1,000 apps is measured against the 25,521 apps our scraper read as listed on an average full day of the window. Shelves are grouped by the developer name on the listing, and two developers with the same name would be counted as one.
Every number comes from the snapshot behind this page, published as raw JSON. No app, developer or reviewer is named in it.