Apple Changed Search Algorithm for App Store

I've also noticed screwy behavior. My app is called "Tea" and it's a fairly well regarded app for tea drinkers.

For the search term "tea", the new algorithm places my app at spot 59 (before it had top billing), and places other irrelevant items, like political tea party stuff, games, and general purpose timers, ahead. If the purpose of the new ranking methodology was to place higher weight on downloads and less on name matching, then there's an unintended consequence of certain niche search terms getting crowded out by tangentially related popular apps. How many people searching "tea" are interested in "M.A.S.H" (result 6) or "Proud Republican"(result 8)? These apps didn't make their bones on tea, but somehow the new algorithm is acting like they did.

This is not to mention that it makes it extraordinarily difficult for folks to find my app if I tell them its name [1]. Here's hoping that this system has more under the hood than meets the eye.

1) https://twitter.com/Lady_KatieJane1/statuses/216444225909301...

OP is my business partner. Building on top of someone elses (Apple's) platform is getting scarier and scarier. They are constantly making sweeping changes that effect our bottom line and provide no transparency, or analytics of almost any kinds, for devs. If I didn't know better I'd say apple hates devs and thinks of us only as a necessary evil... Just throwing us enough of a bone to not jump ship. Obviously, we plan to spread onto other platforms in the meantime we serve at the pleasure of the king, it seems. We are the new generation of "Serf Devs".

Yeah I noticed this yesterday and didn't get time to really research it. I thought I might be crazy, but sure enough, now app name and keywords are searched independently. You can search for "Instagram camera" and not find Instagram. That's crazy.

I need to scramble to get some app updates done to change keywords. I wish Apple would make some sort of statement about this. I also wish they'd let us update keywords without a binary change, even if it's a one-time "free" update to deal with this change.

Great post. Another important change: Apple used to give a lot of weight to titles that exactly matched the search phrase. If you wanted to rank highly for a certain search term, e.g. "flashlight", you could change your title to "flashlight" or "flashlight." or "flashlight+" to show up at the top of results. This trick no longer works... one of our apps went from result #3 to result #37 for the search that matches its title.

Personally, though, I love sudden upheavals like this. If you're paying attention and move fast, it gives you an edge against competitors that have become fat and happy.

The other big problem is that Apple provide zero analytics. I have no idea how people find my app. Apple don't tell you what searches people used so it's impossible to optimize or effectively test keywords. Plus you can only change your keywords with a binary update.

I've stuffed keywords in app titles, and it works well. It's rampant in the navigation category.

A couple releases ago, Apple rejected my app on the grounds we were doing it too much. So, I changed the title and resubmitted.

While I was gearing up for next release, I sent several emails to the review team complaining it wasn't fair, because my precise competitors were doing it, and I was getting killed because my app didn't rank for the search "topo maps" as well as it used to.

I never got an affirmation back that they thought I was in the right, but I changed my title back to a longer version next release and the reviewer didn't reject it this time. Hard to tell if it's just luck of the reviewer draw.

From what we can see, the number of downloads is now a more dominant factor in rank order. Searching for "wedding" a lot of PlayFirst games show up ahead of actual wedding apps because they cross-promote Wedding Dash. Meanwhile WeddingHappy, our wedding planning app, is down to number 74 and downloads have decreased 30%, even though we fare better with more specific search terms.

This makes it much more difficult for new entrants to get into the market without established apps to cross-promote them, and for small developers to stay afloat with a handful of niche apps.

I feel like ranking more popular apps higher in search results hurts app discovery, which I can't imagine is their goal.

The other major problem with app search that I didn't mention in the article is the auto-complete. For some unknown reason rather than having an auto-complete based on what the user is typing, they have an auto-complete that puts up full app names. In my little niche, if you start typing "debt..." you are probably thinking "debt free" or "debt snowball". Instead you are greeted with "Debt free - pay off your debts with debt snowball method." Yep, that is what I was planning on typing.

It starts with "debt free" so a lot of people probably click on that. There are tons of apps that should be coming up for this search, but before this weekend it was 3. that app and 2 others that have all of those keywords he stuffed in his title. Now only his app shows up, as the other 2 apps were the keyword list along with having "Debt Snowball" in the title.

Surely having just 1 app come up in the search isn't the best user experience.

"The Web is dead. Long live the Web."

I may be wrong but it looks like Google's fear of the mobile ecosystem - that apps cannot be crawled, indexed and searched for may be an overreaction. It is precisely because apps cannot be crawled, indexed and searched for, the mobile application ecosystem's expansion will be limited - mobile app discovery will continue to rely on the web. Apps are isolated and cannot link to a particular state of another app. No links, no network effects. The expansion of the app ecosystem is currently limited by the number of apps in the market; With each new app in the market users will require more time to search for a suitable app. More websites, however, actually encourages even more websites, due to linking and references between all the websites; More websites enable users to discover faster the website they derive the most value out of (due to linking). More mobile applications, on the other hand, do not. It's almost like binary tree search O(log(n)) vs iterative search O(n).

These are just my opinions, however, so take it with a grain of salt.

The last time I put something on the app store (~1 yr ago), you could enter up to 100 characters of keywords, which would have prevented the author's initial problem. Did Apple change keywords in the upload process?

I have to admit that I have little sympathy for developers can't speak to whether or not this specific move makes is a better solution or not, but search optimization is a silly little game to take advantage of inefficient systems in the first place. Damn the marketplace provider if they try to optimize to help users find the right products for their needs, rather than those who can throw the most resources behind gaming the system.