ASO Keyword Research Guide 2026: App Store Connect’s 100-Byte Field

Why the 100-byte keyword field still decides discovery

Apple’s App Store search is still heavily metadata-driven. Per App Store Connect reference docs, the Keywords field accepts up to 100 bytes of content—comma-separated terms that help determine where your app can appear in search. Waste those bytes on duplicates, category words, or brand vanity, and you silently forfeit rankings competitors will take.

This 2026 guide walks through a practical research workflow for indie developers and growth marketers: seed generation, competitor mining, byte budgeting, metadata mapping, and ongoing refresh cadence.

Know the metadata map (before you research)

Apple indexes more than the keyword box. Align research with every indexed surface:

Apple’s public guidance: you are already searchable by app name and company name, so do not waste keyword bytes repeating them. Do not use other apps’ or companies’ names. Irrelevant, trademarked, celebrity, or offensive terms are common rejection causes.

Byte math: characters are not always bytes

For English ASO, one character ≈ one byte for basic Latin letters. Non-Latin scripts and some symbols can consume multiple bytes—always count in App Store Connect’s editor or a UTF-8 byte counter when localizing to Japanese, Korean, Chinese, Arabic, etc.

Packing rules that free budget

Apple’s product-page guidance recommends maximizing unique words by avoiding:

• Plurals when the singular is already present (Apple’s matching is forgiving on many English plurals)
• The word “app” and category names you already inhabit
• Duplicate tokens across name, subtitle, and keywords
• Special characters (#, @) unless they are truly part of the brand
• Spaces after commas (run,jog not run, jog) — spaces inside multi-word phrases are fine (Real Estate)

Example compression:

Bad:  fitness, fitness app, workouts, workout, gym, fitness tracker
Better: fitness,workout,gym,tracker,hiit,steps,cardio

A research workflow that fits a two-person team

Step 1 — Define intents, not just words

Group opportunities into intents:

1. Core job — what the user hires the app to do (invoice, interval timer)
2. Audience — who they are (freelancer, postpartum)
3. Situation — when/where (offline maps, travel sim)
4. Feature — concrete capabilities (ocr, csv export)
5. Competitor-adjacent generics — category terms users type after seeing rivals (not rival brand names)

Write 3–5 “jobs to be done” sentences first. Keywords without intents become random lists.

Step 2 — Build a seed list from real language

Sources that stay grounded:

• Support tickets and onboarding surveys (“I needed something that…”)
• App Store search suggestions while typing your seeds
• Autocomplete on Google/YouTube for the same problem space (intent discovery, not Play ranking)
• Reviews of your app and category peers (phrasing users already use)
• Your analytics: branded vs non-branded search terms if available

Aim for 150–300 raw candidates before scoring.

Step 3 — Competitor mining (ethical and guideline-safe)

You can learn from competitors without copying their trademarks into your keyword field.

Do:

• Install top 10 category apps; note how name + subtitle encode keywords
• Record which generic phrases appear repeatedly in their creatives and reviews
• Track their screenshot headlines as clues to converting intents
• Watch which terms they emphasize after major updates

Don’t:

• Put competing app names or company names in keywords (Apple forbids this)
• Stuff irrelevant high-volume terms to hijack traffic
• Clone an entire metadata string—Apple reviewers and algorithms both notice junk relevance

Build a spreadsheet with columns: term, intent bucket, evidence (review quote / autocomplete), estimated competition, relevance (1–5), decision (name / subtitle / keywords / CPP / reject).

Step 4 — Score with a simple indie rubric

You do not need enterprise tooling on day one. Score each term 1–5 on:

• Relevance — would a user who searches this be happy after install?
• Specificity — head terms (photo) vs mid-tail (raw photo editor)
• Business value — closer to paid conversion or retention?
• Coverage — already covered by name/subtitle?

Prioritize mid-tail terms you can win and convert. A #50 rank on a huge head term often loses to a #5 rank on a precise phrase that downloads.

Step 5 — Allocate to fields

1. Put the highest-value, brand-safe phrases in name/subtitle when they read naturally.
2. Pour remaining unique tokens into the 100-byte keyword field.
3. Reserve distinct intents for Custom Product Pages keyword assignment so search can show the matching creative set.
4. Use promotional text for launches and offers—not keyword overflow.

Localization is not translation

A 100-byte English list rarely ports 1:1. For each locale:

• Research local autocomplete and competitor subtitles natively
• Re-count UTF-8 bytes
• Drop calques that nobody searches
• Keep brand name consistent; adapt generics

DE compound nouns, JP loanwords, and PT-BR variants can change byte cost dramatically—budget per locale, not by copying EN.

Refresh cadence for 2026

Avoid weekly thrash. Metadata changes need time to settle; change when evidence says so.

Tie keywords to screenshots

Discovery without conversion wastes spend. For every priority keyword cluster:

• Ensure the first screenshot visually answers that query
• Align subtitle promise with slide-one headline
• Use CPPs when two intents need different creative stories

ASO is a loop: keywords earn impressions; screenshots earn installs; retention and ratings earn more rankings.

Common 100-byte mistakes

1. Repeating the app name three times
2. Leaving unused bytes (“we’ll fill later”)
3. Comma + space habits that burn 10–15 bytes
4. Competitor brand stuffing
5. Ignoring subtitle as a keyword surface
6. English-only thinking on JP/KR storefronts
7. Never revisiting after product-market fit shifts

Mini template you can copy

Name:      [Brand] [Primary job]
Subtitle:  [Audience] · [Differentiator]
Keywords:  term1,term2,term3,...   # exactly 100 bytes, no trailing waste
CPP A:     intent cluster "budget couples" → matching screenshots
CPP B:     intent cluster "receipt scan" → matching screenshots

Bottom line

Treat the App Store Connect keyword field like a scarce inventory slot: research intents, mine competitors for generics (not brands), pack 100 bytes without vanity duplicates, and wire winning terms to screenshots and Custom Product Pages. That disciplined loop—not a longer description—is what moves indie apps in 2026 search.