TL;DR: We followed 2,910 title edits on US App Store apps and the 10,669 edit-keyword pairs that each edit made newly possible. Within 28 days, 16.2% of those pairs entered an app's top 250. A control group of pairs the old metadata already covered, with no position yet, entered at 10.2%. That is a lift of 6.0 points (95% range 4.8 to 7.1). The ratio between the two groups stayed between 1.4 and 1.7 on every day we measured after day 1. The same test on subtitle-only edits shows no lift (0.1 points). The gain is in the top 50. The top 10 saw nothing.
The data: from February 19, 2026 our crawler stores each US app's name, subtitle and release date for every version, and every change in an app's position on a keyword. Between February 19 and October 14 we found 478,622 name or subtitle changes across 361,565 apps. We kept the ones where a new word appeared in the name or subtitle, then followed the keywords that word completed for 28 days. This post tests one piece of ASO folklore, taken from a Telegram chat of ASO practitioners that we labeled (26,975 threads, 2019 to 2026): that a metadata change shows up in rankings within hours to three days. About 18 threads in our sample said so, and the answers inside them disagreed: "a few hours", "a day", "three days", "weeks for all keys". The chat contradicts itself on 13 of the 17 App Store rules we pulled out of it, and this is one of them.
Keywords completed by a new title word enter the top 250 at 1.6 times the control rate
For each edit, we took keywords whose words are all in the new name and subtitle, at least one of which is a word that was not in the old metadata. We call these the new keywords. The app had to have no recorded position on them since February 19. The control is keywords that the old metadata already covered and that had no position yet. If new keywords rise faster than the control, the edit is the likely reason, because both groups share the same app, the same release and the same hidden keyword field.
| Days after release | New title keywords in top 250 | Control keywords in top 250 | Lift (points) | 95% range | Ratio |
|---|---|---|---|---|---|
| 1 | 0.74% | 0.50% | +0.24 | -0.02 to +0.49 | 1.49 |
| 3 | 2.27% | 1.33% | +0.94 | +0.47 to +1.37 | 1.71 |
| 7 | 4.45% | 2.84% | +1.61 | +0.99 to +2.27 | 1.57 |
| 14 | 8.74% | 5.62% | +3.12 | +2.24 to +3.97 | 1.55 |
| 28 | 16.19% | 10.23% | +5.96 | +4.83 to +7.11 | 1.58 |
The control is not at zero, which matters. Apps enter a top-250 list for a keyword all the time without touching their metadata. Roughly 10% of control keywords did so in four weeks. About six in ten of the rankings that follow a title edit would have appeared anyway.
The lift shows from about day 3 and holds the same ratio to the end
The lift in points keeps growing. It is 0.9 points at day 3, 3.1 at day 14 and 6.0 at day 28. The ratio does not grow. It sits between 1.4 and 1.7 for the whole month after day 1.
If Apple needed one to two weeks to apply a title change, the ratio would start near 1.0 and climb. It starts at 1.4 on day 2 (range 1.03 to 2.07), is 1.7 on day 3 and stays near 1.5 afterwards. By day 7, 27% of the 28-day lift is visible, by day 14 it is 52%. We read the rest as our own crawler catching up, not as Apple's delay, for the reason in the next section.
Our crawler cannot see hours, and that caps every number above
The position log records a row only when a position changes, and an app that drops out of a result list leaves no row. Each keyword is crawled on a cycle. For the keywords in this study, the last crawl before our check was a median of 4.7 days earlier. Of these keywords, 29% had been crawled in the last 3 days, 70% in the last 7 and 87% in the last 14. During the study only the keywords that AppSigma users track were crawled daily. The rest of the 1.5 million scored keywords waited longer, some for months. Since October 22, 2026 we also crawl a fixed panel of about 7,500 top search terms every day, so the next studies in this series can work at day resolution.
So a keyword that gained a ranking on day 1 can show up on day 9. The cumulative curve above is Apple's timing plus our revisit delay. What survives that blur is the ratio. It is already above 1 at day 2 and flat after, so we see no sign of a delayed effect. We also cannot say the lift appears within hours. We have nothing finer than a day, and the day-1 lift has a range that includes zero.
Keywords crawled in the last 7 days gave a 28-day lift of 6.3 points. Keywords crawled earlier than that gave 5.0 points, with a wider range (2.5 to 7.4). Both agree with the headline.
A word added to the subtitle shows no lift
Not every new word is a title word. In 4,756 edits, the new words went into the subtitle only. We ran the same test on those.
| Day | Subtitle keywords ranked | Control ranked | Lift (pts) | Range (95%) |
|---|---|---|---|---|
| 7 | 2.13% | 2.54% | -0.41 | -1.09 to +0.27 |
| 14 | 4.21% | 4.33% | -0.12 | -1.04 to +0.72 |
| 28 | 8.22% | 8.11% | +0.11 | -1.24 to +1.25 |
The curve is flat and its range straddles zero on every day. A subtitle word did not raise the share of keywords an app entered at the top 250 in our data. That does not prove the subtitle carries no weight. Subtitle edits are smaller and the keywords they complete are rarer, so our test has less power there. It does show that a title word's lift is not just the effect of "any new word in the metadata". The chat has a rule that the title outweighs the subtitle, and some of its members say the gap has shrunk. A fair test of that rule moves one word between the two fields and changes nothing else. That is the next post in this series.
The lift buys the top 50, not the top 10
| By day 28 | Title keywords | Control | Lift in points | Range |
|---|---|---|---|---|
| Reached top 50 | 6.46% | 4.47% | +1.98 | +1.18 to +2.75 |
| Reached top 10 | 1.64% | 1.83% | -0.19 | -0.72 to +0.29 |
A new title word moved more keywords into the top 50. It did not move more of them into the top 10. When a keyword does enter, the first position we see has a median near #75 for both groups. The first-ever placement of a new word is usually deep in the list. Climbing from there is a different problem, and a title edit alone did not solve it in four weeks.
The result holds in every cut we tried
| Cut (title edits) | New keywords | Control | Lift at day 28 | Range (95%) |
|---|---|---|---|---|
| Keywords created before Feb 19 | 15.45% | 9.66% | +5.79 | +4.48 to +7.07 |
| Snapshot captured within 1 day of release | 17.00% | 11.41% | +5.59 | +3.59 to +7.34 |
| Keyword last crawled 7 days ago or less | 17.45% | 11.20% | +6.26 | +4.92 to +7.71 |
| Keyword last crawled more than 7 days ago | 13.47% | 8.51% | +4.96 | +2.48 to +7.37 |
| Keyword sits inside the app name | 17.95% | 11.81% | +6.14 | +4.61 to +7.74 |
| Keyword mixes name and other words | 12.55% | 5.04% | +7.51 | +5.93 to +8.97 |
| Competitiveness score 25 or below | 22.55% | 16.24% | +6.31 | +3.57 to +8.81 |
| Competitiveness score above 25 | 12.06% | 7.05% | +5.01 | +3.88 to +6.22 |
The lift is between 5 and 7.5 points in every row. It is a little larger on keywords with a lower competitiveness score, as you would expect, and the ranges overlap.
What this means for a title edit
- Judge a title edit at four weeks, not at three days. About half of the 28-day lift was visible by day 14. Our own crawl delay explains part of that.
- Expect most of the new keywords to stay unranked. Even for title words, only 16 of 100 new keywords reached the top 250 in four weeks. The control got 10 of 100 with no new word at all.
- Do not expect the top 10 from the title edit alone. The gain was in the top 50. Anything above that came from other levers in the same release.
- Do not rely on the subtitle to carry a new keyword. We found no lift from subtitle-only words. Test it before you give up a title slot for it.
Track which new words each edit unlocks and where you land on them with AppSigma keyword tracking. The earlier Keiki case study shows what a larger metadata and ads push looked like on one app.
Data quality: what broke, and the clean-panel result
The position log is append-on-change, so a quiet day can mean "nothing moved" or "we did not look". We checked the log before trusting it. We counted rows per day for 300 of the sampled apps from February 19 to October 14.
- Collection gaps. There are no rows at all from March 14 to 24, and rows fall below 40% of the median on 18 more days (April 28 to 30, May 15 to 25, June 24 to 26, July 4). Row counts then roughly triple from September, as crawl capacity grew. The search proxies were also drained from September 6 to 10.
- Late detection of edits. The median gap between a release date and our snapshot is 7.2 days. We anchor on the release date and keep only snapshots within 3 days of it. A separate cut on snapshots taken within a day of release shows the same lift.
- Known traps. Person-name keywords add noise to ranking studies. Our keywords must be built from words in the app's own name and subtitle, and the same filter applies to the control. We drop edits where more than 60% of the metadata words are new, which removes full rewrites and locale flips.
For the clean panel we dropped every edit whose 28-day follow-up touched one of those 34 degraded days. That leaves 897 title edits (3,432 new-keyword pairs, 5,428 control pairs).
| Day | New title keywords | Control | Lift (pts) | Range (95%) | Ratio |
|---|---|---|---|---|---|
| 3 | 2.74% | 1.86% | +0.88 | +0.01 to +1.68 | 1.48 |
| 7 | 4.92% | 3.36% | +1.57 | +0.41 to +2.70 | 1.47 |
| 14 | 8.86% | 5.81% | +3.04 | +1.53 to +4.48 | 1.52 |
| 28 | 17.83% | 11.72% | +6.11 | +4.08 to +8.13 | 1.52 |
The lift at day 28 is 6.1 points, against 6.0 on all edits, and the ratio stays near 1.5. Before day 7 the ranges are wide: at day 3 the lower end of the range is barely above zero, so we call the effect visible from about day 3, not day 2. The subtitle result also holds on the clean panel (+1.0 points at day 28, range -0.8 to +2.8). We did not build a panel of keywords crawled every day, because at the time only user-tracked keywords had a daily crawl. Keeping only keywords with a crawl in the three days before our check leaves 595 edits and a lift of 5.7 points at day 28 (range 2.2 to 9.1).
Caveats
- Our first control was wrong. We first used all keywords an app already matched. About 59% of those were inside the app's own name, and 86% of the control keywords that appeared in the top 10 were name keywords, the kind that rank the first time our crawler reaches them. That inflated the control, and it made subtitle edits look 6.7 points worse than the control, which is impossible. We now compare within groups of the same competitiveness score and the same "keyword inside the app name" flag. The numbers above use that version.
- We cannot prove the edit is the cause. A title edit often ships with a new version, new screenshots or a change to the hidden keyword field. The control shares those changes only in part. The result is a lift that goes with a title edit, not a measured effect of the word alone.
- We see entries, not exits. The log has no row when an app drops out of a list, so we measure the first time an app appears. We cannot measure how long a ranking lasts.
- Snapshots lag releases. Only 49% of snapshots are taken within 7 days of release. We kept edits captured within 3 days of release (134,201 of 478,622) and a random 8,000 of the 48,574 that passed our other filters. The cut with snapshots inside 1 day gives the same result.
- One storefront, one sample. US only, February to September 2026, one edit per app. Competitiveness score is our own measure of how strong the apps are that already rank on a keyword.
Frequently asked questions
How long does a new word in the app title take to show up in rankings?
The lift is visible from about day 3 and keeps growing for the 28 days we followed. Keywords completed by a new title word were 1.5 to 1.7 times as likely to enter the top 250 as a control from day 3 on, but our crawler revisits a keyword only every 10 days or so, so we cannot separate Apple's delay from our own.
Does a word added to the subtitle rank like a word added to the title?
Not in our data. After 28 days, keywords completed by a new subtitle word entered the top 250 at 8.2%, against 8.1% for the control, a difference of 0.1 points with a range from -1.2 to +1.3. Title words showed a lift of 6.0 points.
Will a new title word get my app into the top 10?
Rarely. A new title word raised the share of keywords that reached the top 50 by 2.0 points, from 4.5% to 6.5%. The share that reached the top 10 did not change: 1.6% against 1.8% for the control.
Can you see a ranking change within hours of a metadata update?
We cannot. Our crawler checks a typical keyword every 10 days or so, and during the study only the few keywords that users track were crawled daily. Our shortest window is one day, and at one day the lift is small and its range includes zero.