Marketing Your App More Efficiently with Apple Search Ads — Dilip Reddy, Search Ads Optimization

Run brand keywords if your app is easily replaced by a competitor (skip if network effects make you irreplaceable). Then test lower brand bids and track impression share, which Apple reports per search term per country. If impression share holds (e.g. ~90%), you have roughly halved the cost; some companies accept ~70% impression share for a ~70% cost cut. Apple charges per tap and favors the relevant brand app, so you can often win your brand term below competitors' bids, but a competitor with much better screenshots, messaging or ratings can take share, so keep monitoring. Move the savings to generic high-intent keywords that bring new users.

Instead of only tier-1 countries, add as many countries as possible: a handful of localized keyword sets (Spanish, Portuguese, German, French) covers most of Latin America and Europe, and elsewhere Apple defaults to English. Copying campaigns takes minutes. Test newly opened storefronts (in 2024, Brazil and about six other Latin American countries). In any new storefront, start with bids so low they get no impressions and raise gradually, because a high start (e.g. $10) may pay double for ~90% of the same installs.

Keep one relevant keyword list, classified into groups (brand, high-intent generic, low-intent, competitor, broad) with different bids: e.g. for a yoga app, 'yoga' is high intent, 'pilates' low intent and bid lower. Pick keywords yourself (ChatGPT only for ideas). Run each core term as both exact and broad match; broad match catches changing long-tail variants that competitors often don't bid on. Check the search terms report occasionally to see where spend goes; when broad match surfaces a relevant term (e.g. 'weather alerts' from 'weather'), add it as its own exact + broad keyword. Scale by adding relevant keywords, broad match and countries, with a simple campaign structure rather than hundreds of campaigns per country.

When a keyword or matched search term underperforms (e.g. a yoga ad showing on 'Peloton'), find the keyword causing it and bid it down so low that installs are very cheap or Apple stops showing the ad, then retry in two or three months. Only add a negative keyword when the term is consistently bad and clearly never a fit. Avoid the common advice of removing any keyword that doesn't make revenue in the first week, since competitors' bids change week to week.

Brand keywords have the highest ROAS because other channels already drove the user to the App Store. Treat brand as its own entity: never compare it with generic, competitor or broad match keywords, and don't mix its data into the overall Apple Search Ads numbers. Be skeptical of case studies that don't say whether ROAS is brand or generic.

Don't pre-adjust bids for seasonality. Check App Store Connect's last two years of organic traffic to know what to expect (search ads trends correlate), then review campaigns weekly and bid up or down per keyword or country ad group based on the past 7 or 15 days. Expect CPAs to rise around Thanksgiving/Black Friday in the US even for non-shopping apps; February to summer is a quieter, cheaper time to start testing ASA.

ASA doesn't report revenue, so link keywords and ad groups to revenue via an MMP or RevenueCat. Don't set a hard revenue target and kill the channel if missed; the e-commerce 400% ROAS rule doesn't fit subscriptions. Track time to break even: 40-50% ROAS in the first week is fine while you add keywords and tune, since a one-time install cost keeps earning renewals. Optimize for revenue and ROAS together: a competitor spending twice as much at lower ROAS may grow bigger and outbid you. Get install-to-trial and trial-to-paid conversion right first.

An app with strong organic traffic throwing off cash set its paid acquisition bar at 100% ROAS by the first annual renewal (365 days) instead of 7-, 30- or 90-day ROAS, relying on renewals in later years to make each cohort profitable.

Once generic impression share is maxed out, use keyword-specific custom product pages, e.g. a Black Friday page for US Black Friday keywords. For A/B tests, run the same keywords with the default product page in half of US states and the new custom product page in the other half, then compare. Splitting by geography is preferred over ad scheduling by time of day, since intent varies by time (a yoga app in the morning).

For new apps, run Apple Search Ads on high-intent keywords with a small daily budget ($10-20/day for about 5-6 installs a day) as customer discovery, and use those users to improve the app. Also useful when better-ranked competitors make it nearly impossible to rank organically for a keyword.

Host plans to start Apple Search Ads for his weather app on terms like 'weather' and 'weather radar' as broad match. Search terms for features the app lacks, like 'weather alerts', would get their own exact match at a very low bid (they bring installs that leave bad reviews or don't convert), with bids raised once the feature ships. The guest adds that a consistently bad term can be made a negative keyword.