7 min read ·
How to Market an iOS App: A Weekly Routine
Quick answer
A weekly routine for marketing an iOS app on your own: one number to watch, one listing change, every review answered and one channel worked every week.
You shipped an iOS app, launch week is over, and downloads have settled into a number you check more often than you would admit. Most guides to marketing an iOS app hand you twenty tactics at once. One person cannot run twenty tactics. What a solo maker can run is a short routine on the same days every week, measured against one number.
This is that routine. Once you have done it twice it takes about 90 minutes a week, and the order matters. The listing comes first, because every other channel sends people to it. The reviews come second, because they sit on the listing. One channel comes third. The number opens the week and closes it.
Pick one number and write it down every Monday
Open App Store Connect, go to Analytics, and read two figures for the last seven days: impressions, and the product page conversion rate for App Store Search traffic. Impressions tell you how often people saw the app. Conversion tells you how many of the people who searched and saw it went on to download.
Downloads on their own mix two problems that have different fixes. Reading the pair tells you which one you have:
| Impressions | Conversion rate | What it usually means | Where to spend this week |
|---|---|---|---|
| Low | Steady | Few people find you | Keyword field, subtitle, a channel that sends people to search for you |
| Steady | Low | People find you and leave | Screenshots, the first lines of the page, ratings |
| Low | Low | Both | The listing first, because a channel only sends people to a page that loses them |
| Rising | Holding | It is working | Keep the routine and make small changes |
What counts as a good conversion rate depends on the category, so compare the app with itself week on week. A spreadsheet with three columns is enough: the date, impressions, conversion rate. Add a fourth for the one change you made that week. After six weeks you can see which changes moved anything.
App Analytics also splits traffic by source: App Store Search, App Store Browse, App Referrer and Web Referrer. When you post somewhere and downloads jump, the jump shows under Web Referrer or App Referrer. That is how you tell whether a post sent anyone.
Monday: make one listing change, and only one
Change the subtitle and the screenshots in the same week and you will never know which one moved the number. One change a week is slow. It is also the only pace at which a solo maker learns what works on their own listing.
Each field has a limit, and most of them only change with a new version:
| Field | Limit | Changes without a new version |
|---|---|---|
| App name | 30 characters | No |
| Subtitle | 30 characters | No |
| Keyword field | 100 characters | No |
| Promotional text | 170 characters | Yes |
| Description | 4,000 characters | No |
| Screenshots | Up to 10 per device size | No |
Apple ranks search on the name, the subtitle and the keyword field, so those three carry your keywords. In the keyword field, separate words with commas and no spaces, and leave out any word already in the name or subtitle, since Apple counts it once already. Spend the freed characters on words people type that your title does not cover.
Promotional text is the one field you can change any week without shipping a build. Use it for what is true this week: the feature in the latest release, a seasonal use, or one line answering the question that keeps coming up in the reviews. It sits above the description, so it is the first copy many people read.
For the fields that need a version, keep a running note of the change you want and send it with the build you were shipping anyway. The routine still holds in a week with no release: that week's change is the promotional text.
Tuesday: answer every review that asks or complains
Reply from App Store Connect, under Ratings and Reviews. The reviewer is notified of your response and can update their review, which is the one way a two-star rating ever becomes a four.
Answer in this order:
- Questions. Someone asking whether the iPad version is coming is telling you what the page fails to say. Answer them, then consider whether the promotional text should say it.
- Bugs you have fixed. Name the version that fixed it, so they know what to update to.
- Low ratings with a specific complaint. Say what you changed, or say plainly that you will not change it and why.
- Kind five-star reviews, when you have time. A short thank you is enough.
We wrote templates for each of these in How to Reply to App Store Reviews.
Ratings themselves come from inside the app. iOS shows the system rating prompt at most three times a year per user, however often your code asks for it, so spend those three on moments when someone has just finished the thing your app is for: the workout logged, the file exported, the streak kept. First launch is the worst moment to ask.
Wednesday and Thursday: work one channel properly
Pick one place where the people who would use your app already talk, and stay there for six weeks before you judge it. That might be the subreddit for the hobby your app serves, X if your users are developers or makers, short video if the app is visual and consumer, or a newsletter in your niche. Spreading the same post across five places gets you five half-read posts and no idea which place works.
Each week on that channel:
- Answer the threads where someone is asking for what your app does. We answer only threads under a day old, because a reply to an older thread lands after the asker has moved on.
- Post one thing of your own that shows the app doing one specific job. A screen recording of the job beats a feature list.
- Check Web Referrer and App Referrer the day after. Views on the post tell you about the post. Downloads under those sources tell you about the app.
If six weeks of honest work on one channel sends no one, move to the next one on your list and keep the routine. Our article on getting your first 100 users for an app goes further on choosing that first channel from the store console reports.
Friday: write the release notes and close the week
If a build goes out this week, write the What's New in plain words, naming what the user can now do. It is read by the people deciding whether to update and by the people deciding whether to download. Send the queued listing change with it.
Then open the spreadsheet again. Write the week's impressions and conversion rate beside Monday's figures, and one line on what you changed. That line is the whole point of the routine: in two months you will have a record of what this listing responds to.
What to leave until the routine holds
Some work pays only once the basics are done:
- Apple Ads. Ads send people to the same product page. If it converts poorly, you pay for every tap that leaves. When the page holds, start with a small monthly budget on exact-match keywords you already rank for, and read cost per install against what those installs pay you.
- Custom product pages and product page optimisation tests. Both need enough traffic to read a result. On a few hundred impressions a week, a test runs for months.
- Localised listings. Worth doing for the countries already sending you downloads. Translate the keyword field properly rather than word for word.
- Press and launch platforms. Useful once, around a real release. They do not replace the weekly work.
Running the routine with Kwern
We built Kwern's ASO agent to do the listing and review part of this week for you to approve. It runs weekly, reads the keyword field, the product page analytics by source and the reviews that came in, then drafts the replies and the listing changes worth making. A review reply and the promotional text go to App Store Connect when you approve them, because those are the two things Apple lets you change outside a release. Every other field arrives as a value you paste in with your next build, or as a pull request when your metadata lives in a repository you connected.
When you get to ads, the Apple Ads agent proposes bids, keywords and pauses against the monthly budget you set, and with RevenueCat connected the weekly report shows revenue and return on ad spend for each campaign. Nothing is published or spent until you approve it.
The channel days are yours to choose. Kwern has agents for X, Reddit and Hacker News that find threads and draft replies, and it has no video agent, so a short-video channel stays with you. Plans start at $29 a month with every agent included, and the model runs on your own provider key. The details are on the pricing page.