sonarReelMeal is my iPhone app that turns TikTok, Instagram and YouTube cooking videos into recipes. Until late August 2026 its App Store name was "ReelMeal: Recipe Keeper". On paper that was a sensible choice, because "recipe keeper" is one of the most searched terms in the category. In practice it brought the wrong visitors.
This is the before and after, with the numbers from App Store Connect and Sonar, including the part that went wrong.
The short version
- Before: "ReelMeal: Recipe Keeper", anchored on a keyword with an Apple popularity of 55.
- After (version 1.3, live on August 27, 2026): "Video to Recipe AI: ReelMeal", with the subtitle "Recipe Saver & Grocery List".
- US conversion (first-time downloads per impression) went from 0.19% to 0.45%.
- "video to recipe" moved from #15–23 to #10.
- One loss: "cooking video to recipe" fell from #21 to outside the top 100 when "cooking" left the keyword field.
The numbers are small and the test wasn't clean. Both are covered below.
The problem: impressions without downloads
When I looked at App Store Connect on August 26, the US product page had 4,469 impressions and 7 downloads over the previous 30 days, about 0.16%. On August 18 alone it got 3,490 impressions and no downloads at all.
The same app converted at 2–9% in markets where the listing led with video, Germany at about 9%. So the product page wasn't the problem. The US listing was attracting the wrong searches.
My read on why: someone typing "recipe keeper" wants a box for recipes they already have, and many of them are looking for Recipe Keeper, an established app with that exact name. ReelMeal's pitch is getting recipes out of videos. The title matched the search, then the screenshots told a different story.
What the keyword data said
US App Store, numbers Sonar returned on October 7, 2026. "Proxy" is Sonar's own popularity estimate, shown where Apple reports its floor of 5.
| Keyword | Apple popularity | Difficulty | ReelMeal rank, early August | October 7 |
|---|---|---|---|---|
| recipe keeper | 55 | 56 | not tracked | not found |
| video to recipe | 5 (proxy 24) | 26 | #15–23 | #10 |
| video recipe | 5 | 21 | #28–124 | #34 |
| cooking video to recipe | 5 (proxy 8) | 38 | #20–21 (late August) | not found |
"Recipe keeper" is the only term here with real reported demand, and it comes with a difficulty of 56 and a strong navigational pull toward a competitor.
"Video to recipe" reads as Apple's floor, but Sonar's proxy puts demand at 24, the difficulty was 26, and it describes exactly what the app does. The detail that decided it: ReelMeal was already climbing for it without having it in the name. Sonar's rank history showed #15–23 through mid-August and #13–15 by August 20, a week before the new title went live. Putting the phrase in the name was a bet on a term Apple already considered relevant, not a cold start.
The change
| Field | Before (version 1.2) | After (version 1.3) |
|---|---|---|
| Name | ReelMeal: Recipe Keeper | Video to Recipe AI: ReelMeal |
| Subtitle | not recorded | Recipe Saver & Grocery List |
| Keywords | keeper, cooking, ingredient, instruction, import, extract, organizer, saver, book, grocery, meal, planner, vault | tiktok, instagram, reels, youtube, import, scanner, keeper, manager, organizer, cookbook, extract, save, box |
Three decisions behind it:
- Anchor first, brand last. Nobody searches for a new brand name, so the 30 characters went to the phrase people do search for. There is more on this trade-off in how long an iOS app title should be.
- Keep the old term reachable. "keeper" stayed in the keyword field. Apple combines tokens across the name, subtitle and keyword field, so "recipe" in the name plus "keeper" in the field still covers "recipe keeper", just without the title's weight.
- Spend the keyword field on sources. TikTok, Instagram, Reels and YouTube describe where users find the recipes, which matches the new anchor. The keyword field guide covers the formatting rules.
What happened
US product page, App Store Connect. Impressions come from the Analytics Reports API and first-time downloads from the Sales Reports API.
| Window | Impressions | First-time downloads | Conversion |
|---|---|---|---|
| August 1–23 (old title) | 4,225 | 8 | 0.19% |
| September 1–23 (new title) | 2,875 | 13 | 0.45% |
Over 30-day windows on either side of the release, US first-time downloads went from 10 to 19, and worldwide from 71 to 94.
Impressions fell by about a third while downloads rose. That is the trade I was after: fewer people seeing the listing, more of the right ones.
Before you take the 0.45% to the bank, the caveats:
- The numbers are small. Thirteen downloads is not a sample you can run statistics on.
- The "before" window is derived. The daily analytics export I have starts on August 30, so August 1–23 is the August total minus the last week. It also contains the August 18 spike.
- It wasn't a clean test. I changed the name, subtitle and keyword field together, and added 15 Apple Ads keywords to the US campaigns on August 26. The sales report can't separate paid installs, so some of the lift may be paid.
The mistake: dropping "cooking"
"cooking" was in the old keyword field and appears nowhere in the new metadata. ReelMeal ranked #20–21 for "cooking video to recipe" from August 23 to 27. From September 4 it ranked #117–172, and from September 27 Sonar no longer found it at all.
It's the flip side of Apple combining tokens across fields. Every phrase that needed "cooking" lost it at once, and nothing in App Store Connect tells you which ranked phrases depend on a token before you remove it. The check is quick: before you submit, list the keywords you rank for in Sonar and look for any that use a word you're about to delete.
What I'm still watching
On September 27–28, several long-tail ranks disappeared at once, including "recipe from video", whose words are all still in the metadata, and US downloads slowed to 2 in the following ten days. "Video to recipe" held at #10. I don't know the cause yet, and I'd rather say so than invent one. I'll update this post when I do.
What I'd tell another indie developer
- Judge a title keyword by conversion, not impressions. Split App Store Connect by country and look for the market where your listing converts. That listing's angle is your real anchor. Conversion rate benchmarks help you tell bad from normal.
- Exact beats popular for a small app. A floored keyword that describes what you do can convert better than a popular one that half-describes it. Use a proxy estimate to see whether the floor hides real demand, as in finding low-competition keywords.
- Look for terms you're already climbing. A keyword where you rank #15 without trying is cheaper to win than one where you're absent.
- Diff your keywords before you submit. Check which ranked phrases depend on any token you remove.
- Change one thing at a time if you want to learn from it. I didn't, so I can tell you what happened but only part of why.
Questions
Should I put the most popular keyword in my App Store title?
Only if your app is what those searchers want. Popularity tells you how many people search, not whether they will download. ReelMeal's US page got plenty of impressions from "recipe keeper" and converted at 0.19%.
How quickly does a title change show up in rankings?
The new title went live on August 27, 2026, and Sonar picked up the release about a day later. "Video to recipe" settled in the #10–16 range within days, but it had been climbing for weeks before, so the change consolidated a trend rather than creating one.
Does Apple combine keywords across the name, subtitle and keyword field?
Yes. That's why ReelMeal still covers "recipe keeper" through "recipe" in the name and "keeper" in the keyword field. It cuts both ways: removing a single token can take down every phrase that relied on it, as "cooking" did here.
