If your Android app's search rankings jumped around in September, it probably wasn't your listing. Google Play shipped a search ranking update on Monday 14 September, a second wave two days later, and a third on Wednesday 23 September, each time in nearly every country at once. Google announced none of it.
I build Applyra, an ASO tool that measures keyword rankings on Google Play and the App Store every day, and every number below comes from its data.
TL;DR
- Three store-wide waves in ten days: 14, 16 and 23 September 2026.
- On the 23rd, US search results moved 9.6 times more than on an ordinary day. The Netherlands reached 18x, Germany 17x.
- The App Store stayed flat in the same markets on the same days. The cause is Google, not your listing.
- It was not a rollback: after the 23rd, 40% of US positions still sat 10 places or more from their rank before the 14th.
- Ranks 11 to 100 and long-tail searches took the biggest hit, which is exactly where most indie apps live.
Three waves in ten days
Every figure below compares a market with its own ordinary day. 1x is a normal day, 10x is ten times more movement than usual.
| Wave | Markets moved | United States | Peak (Netherlands) |
|---|---|---|---|
| Mon 14 Sep | 15 | 5.4x | 10.1x |
| Wed 16 Sep | 14 | 5.9x | 13.1x |
| Wed 23 Sep | 15 | 9.6x | 18.0x |
The Netherlands was hit hardest every time. To put the last line in plain terms: on 23 September, 48% of US searches saw their apps move 10 places or more in a single day. The typical keyword moved 9 places, when it normally moves 1.
Google Play in the United States. Each red dot is a store-wide update: 20 August, then 14, 16 and 23 September.
23 September, market by market
The latest wave, with the App Store in the same markets on the same day:
| Market | Google Play | App Store |
|---|---|---|
| Netherlands | 18.0x | 0.9x |
| Germany | 17.0x | 1.0x |
| India | 10.9x | 0.7x |
| Poland | 10.7x | 1.5x |
| Australia | 10.6x | 1.8x |
| Japan | 9.7x | 1.5x |
| Italy | 9.6x | 1.2x |
| United States | 9.6x | 0.9x |
| United Kingdom | 8.5x | 0.9x |
| Indonesia | 8.0x | 0.8x |
| Canada | 7.8x | 0.7x |
| Turkey | 7.1x | 1.0x |
| France | 3.3x | 1.0x |
| Mexico | 2.8x | 1.2x |
| Spain | 2.3x | 1.0x |
| Brazil | 1.9x | 1.0x |
12 markets ran above 7 times their ordinary day. Brazil barely moved, on this wave or the two before, which is what a staged rollout looks like: the same update reaches some countries before others.
The App Store didn't move, and that's the diagnosis
Look at the right-hand column. On 23 September, not one of these markets reached 2x on the App Store, and the United States read 0.9x.
That column is what turns a spike into a diagnosis. Both stores were measured side by side, in the same markets, on the same days. One rewrote its results three times in ten days. The other didn't notice. Whatever shipped, it shipped at Google.
It's also the check worth running on your own app the next time your rankings fall. The first question is never "how do I fix this". It's "did I break this, or did the store?" Both look identical in a dashboard and call for opposite responses. The fastest way to tell them apart is a control: the same app, the same days, on a store you didn't touch. Both stores down means the cause is yours. One store down and the other flat means the cause is that store, and nothing you write in your listing will bring it back.
Did Google roll it back? No.
Two big moves nine days apart look like a mistake being undone. Following every US position through both waves says otherwise:
- Of the positions that moved 10 places or more on both the 14th and the 23rd, 2 in 3 went back the other way. That part was a correction.
- But nearly 4 in 10 of the positions the 23rd moved hadn't moved much on the 14th. That part was new.
- The net result: after the 23rd, 40% of US positions still sat 10 places or more from where they were before 14 September.
Google adjusted its update. It didn't take it back. If you lost ground on the 14th and got it back on the 23rd, you were in the part that got corrected. If you didn't, stop waiting for a rollback.
Who got hit: the part of Google Play where indie apps live
The top of the results held. Almost everything below it was rewritten. Here is how many US positions kept their exact rank through 23 September, against an ordinary day:
| Rank before the update | Kept its position on 23 Sep | On an ordinary day |
|---|---|---|
| 1 to 3 | 64.6% | 85.8% |
| 4 to 10 | 22.4% | 64.3% |
| 11 to 25 | 7.0% | 40.8% |
| 26 to 50 | 2.8% | 25.8% |
| 51 to 100 | 1.7% | 16.0% |
The 14th had the same shape. Three things stand out if your app isn't a household name:
- The long tail moved most. 51% of low-volume searches moved 10 places or more on the 23rd, against 41% of the biggest head terms. The specific multi-word phrases small apps live on were hit hardest.
- Ranks 11 to 50 lost ground on both days, 7 to 9 places on average on the 23rd. That is the band where a good listing turns into installs.
- New apps broke in. On 14 September, 1 in 8 US positions went to an app that wasn't in the top 100 for that search the day before. On an ordinary day it's 1 in 90.
In other words, Google rebuilt the exact part of the store where indie apps get discovered, then rebuilt it again. That cuts both ways: the same updates opened searches to thousands of apps that had never ranked there.
The new rankings are here to stay
24 hours after each wave, most of the new positions were still there: roughly 9 in 10 after the 14th, and 8 in 10 in the United States and 9 in 10 in Germany after the 23rd. A glitch reverts overnight. These didn't.
Whatever ranking you woke up to on 24 September is your position now, and any keyword strategy you built before the 14th was calibrated against a store that no longer works the same way.
What to do this week
- Don't react with a release. A panic metadata change re-triggers indexing at the worst possible moment, adds your own variable on top of Google's, and destroys the only clean before-and-after you had. Read the new positions for two or three days first, then decide.
- Confirm the date before you touch anything. If your Google Play rankings moved between 14 and 24 September, this is why. Check whether your App Store rankings held over the same days.
- Look at your keywords ranked 11 to 100, not your top 3. Your strongest terms mostly survived. The movement, in both directions, is in the band below. That is where you gained and where you lost.
- Re-read which keywords you can win. Every keyword difficulty score is computed from who currently ranks. Google just changed who ranks, everywhere, three times. Any list of "keywords I can realistically win" built before 14 September is now stale.
- Mark the seam in your own history. Any before-and-after you run across 14 to 24 September is measuring Google, not your work. If you changed your short description on 8 September and judged it on 20 September, you learned nothing about your short description. Re-run that test on clean dates.
Google will do this again
Four store-wide updates in five weeks: 20 August, then 14, 16 and 23 September. None announced, none confirmed.
The next one is coming. The only question is whether you'll recognise it on the day, or spend a week rewriting a listing that was never the problem.
Data: Applyra's algorithm change detector, which publishes this measurement every day for both stores. Here is the 23 September page for the United States.
Top comments (0)