Skip to content

invade.io — Play Store refresh and publishing guide

Source reference: version 0.5.0, commit 05eec6f. This is not a statement of the version currently on Google Play.

Documentation · Android publishing · Release process

Current source target: 0.5.0 release candidate. Store copy was prepared for the 0.3.2 refresh and updated for 0.5.0; graphics were recaptured and revalidated on 9 November 2026.

This document records the implemented marketing work, the files to upload, and the checks the owner must make in Play Console. Nothing in this workflow publishes an app, changes the live listing, or changes the game’s telemetry defaults.

  1. Review build/store_capture/contact_en.png and contact_es.png, then the other language contact sheets. These are review sheets, not upload screenshots.
  2. Read the English or Spanish listing in docs/store_listing/.
  3. Confirm that the production release matches this source version, particularly its telemetry behavior. Do not copy the current privacy paragraph onto an older release.
  4. Have native speakers review translated copy, especially the 13 non-EN/ES full descriptions, which began as machine-translated drafts. Short graphic captions have been edited for game terminology and layout, but that is not native-speaker approval.
  5. Build the upload ZIP with the commands below. Extract it locally; paste the text and upload the individual PNG files in Play Console. The ZIP is not a Console import format.
Terminal window
# Text, locale, field-length and alt-text checks: offline, no writes.
python3 tools/build_store_listings.py --check
# Fresh real-scene captures and final graphics. Requires Godot 4.7.2 and a display.
# On a desktop, omit xvfb-run. Do not use Godot --headless for image generation.
xvfb-run -a godot --audio-driver Dummy --path . res://scenes/tests/store_assets.tscn -- --no-telemetry
# Validate all graphics, caption provenance and hashes, then package the material.
python3 tools/build_store_listings.py --package

Delivery: build/play-store-listing.zip. It contains copy and images grouped by Play Console locale, the existing 512 px icon, capture manifest and this guide. build/ is intentionally ignored by Git; the reproducible source and final PNGs live in docs/. Rebuild the ZIP after changing any text, image or this guide.

CI also generates this material and exposes a play-store-listing review artifact in GitHub Actions for seven days, including the ZIP and language contact sheets. It does not upload anything to Play Console or modify signing/release permissions.

BeforeImplemented improvementPurpose
Brand-only titlesLocalized brand + genre/territory-conquest titles, within 30 charactersExplain the game to unfamiliar visitors
Outdated two-language/15-achievement descriptionConsistent copy for 15 languages, 20 achievements, 36 regions and 108 frontsMatch the current game and show its depth
A menu opens the galleryThe first three images show real attacks on distinct mapsMake the core interaction understandable before asking visitors to read
Uncaptioned UI-test screenshotsEight images per language with one benefit-led headline and short supporting lineExplain the experience at thumbnail size
Mostly empty progressA representative mid-game fixture with 51 cities, earned upgrades and achievementsShow what continuing to play can unlock
Cosmetics/atlas below the visible areaReal scrolling to troop styles/base shapes/map themes and the populated atlasMake those features visible rather than merely claiming them
Repeated icon + logo feature graphicA gameplay-led map confrontation, cropped below the HUD, with a smaller wordmarkCommunicate the game beyond the icon
Screenshots depend on a prior --ui runOne self-contained Godot capture sceneAvoid stale or missing source screenshots
Different themes by localeA consistent light presentation, a genuine dark appearance screen and dark-blue captionsKeep markets coherent while demonstrating both UI themes
PNG screenshots retain alphaExplicit RGB8 final PNGs at the target dimensionsMatch the documented Play upload format
No large-screen materialFour uncaptioned 1920×1080 tablet captures per languageShow actual landscape layouts without adding phone-marketing overlays
Manual, disconnected filesOffline validation, hashes, ordered manifest, localized alt text and upload ZIPReduce publishing mistakes

The icon is deliberately unchanged. There are no invented troops, fake controls, device frames, testimonials, ranking badges, install calls-to-action or competitor branding. No gameplay feature, monetization system, analytics service or account system was added.

Core promise: approachable territory strategy on real-world maps, with tactical decisions, progression, offline play and no advertising interruptions.

  1. What the player does: drag between bases, send troops, capture territory.
  2. Why decisions matter: factories, fortresses, capitals, bosses and continent rules.
  3. Why there is more to discover: campaign, invasions, expeditions and daily challenges.
  4. Why keep playing: upgrades, cosmetic rewards, atlas, ranks and achievements.
  5. Why it fits the player: offline, no ads or purchases, 15 languages and accessibility options.

Returning players get a clearer view of content and goals already present. New players see the interaction before the menus. These are marketing hypotheses, not measured acquisition or retention improvements; use the measurement plan in §10.

Do not advertise multiplayer, live PvP, online rankings, cloud saves, all cities of the world as campaign objectives, or guaranteed battle lengths. The daily challenge runs locally and has comparable starting forces; sharing copies a result card.

Each docs/store_listing/<game-locale>.md has exactly three fenced blocks:

  1. App title, at most 30 characters.
  2. Short description, at most 80 characters.
  3. Full description, at most 4,000 characters.

Only paste the block contents, not Markdown headings or fences. The ZIP already separates them into title.txt, short_description.txt and full_description.txt.

English title: invade.io: Offline Strategy (27 characters). Spanish title: invade.io: Estrategia offline (29 characters).

Titles change only the listing; the installed game’s name and permanent package identifier macarronesc.invadeio remain unchanged.

Final source: docs/store_assets/screenshots_<game-locale>/. ZIP: <play-locale>/phone/. Upload only the eight PNGs, in numeric order.

FileEnglish headlineSource/proof
1_conquer.pngDrag. Send. Conquer.Athens → Istanbul troop dispatch, europe_4
2_tactics.pngEvery base changes the battleEuropean factory/fortress/boss map, europe_5
3_invasions.png36 regions. 108 fronts.Iberia’s capital front, inv_iberia_3
4_campaign.pngSix continents. More to conquer.Campaign map, stars and expedition entry
5_daily.pngA different challenge every dayFixed real daily-challenge seed, with its rule/objective
6_upgrades.pngBuild a stronger armyReal upgrade cards and next upgrade effects
7_cosmetics.pngMake every conquest yoursReal appearance screen, scrolled to troop/base/map options
8_atlas.pngTrack every conquestReal atlas with conquered cities and achievement progress

All final phone images are 1080×1920, 24-bit RGB PNG. A 320 px caption band occupies one-sixth of the height. The remaining 1080×1600 area is a real game viewport rendered at that supported size: no stretching, fictional UI or handset frame.

The fixture represents achievable mid-game progress, not the initial rewards of a new installation. It is isolated from player saves. Battle production and AI advance in fixed simulation steps before a valid attack is dispatched; combat rules and army counts are not inflated for the image. No fabricated victory screen is used.

Every PNG has a matching .alt.txt in the ZIP. Paste that localized description into the image’s alt-text field where Console offers it; do not upload the text file as an image. Alt text describes the image rather than repeating the headline, and is ≤140 characters.

Source: docs/store_assets/tablet_<game-locale>/. ZIP: <play-locale>/tablet/. The four PNGs show campaign, an invasion battle, appearance options and the atlas. They are 1920×1080 RGB, rendered from the actual landscape layout, without additional marketing text. Use them in the appropriate tablet screenshot sections after checking the installed release on the corresponding devices. Reusing the set for both tablet sizes is appropriate only if the real layouts match.

  • docs/store_assets/feature-graphic-1024x500.png: 1024×500 RGB, no alpha. The ZIP includes the same brand-only graphic in each locale directory plus localized alt text.
  • assets/brand/play-store-icon-512.png: existing 512×512 RGBA icon, packaged as icon-512.png. Do not replace Android launcher assets with the feature graphic.
  • Contact sheets and raw captures stay in build/store_capture/; do not upload them.

The promo video is rendered from the real game, like the screenshots: scripted battles driven by PlayerBot, the shared StoreFixture save, and captions from docs/store_listing/graphics.json.

Terminal window
scripts/render_trailer.sh en # any game locale; needs godot, xvfb-run and ffmpeg

The output is build/trailer/invadeio-trailer-<lang>.mp4 (1920×1080, about 35 s, the game’s own audio). Upload it to YouTube (public or unlisted, ads off, captions on) and paste the URL in Play Console → Store listing → Video. It is not committed. The video opens on gameplay, never on a logo, and ends on the browser-demo URL.

Game locales are not always the same identifiers as Console locales. Packaging uses the following mapping; select the matching language, not a guessed country.

Game/sourcePlay Console/ZIPLanguage
enen-USEnglish (United States)
eses-ESSpanish (Spain)
pt_BRpt-BRPortuguese (Brazil)
frfr-FRFrench
dede-DEGerman
itit-ITItalian
ruru-RURussian
trtr-TRTurkish
ararArabic
hihi-INHindi
ididIndonesian
viviVietnamese
zh_CNzh-CNSimplified Chinese
jaja-JPJapanese
koko-KRKorean

Arabic captions and the game’s interface use RTL; geographic maps and the brand are not mirrored. Per-script Noto subsets include the new caption characters. Language support does not mean the app is distributed in every country; review country availability separately. No Traditional Chinese or other unsupported translation is promised.

6. Update the published listing in Play Console

Section titled “6. Update the published listing in Play Console”

Console navigation labels can change. Work in the existing application; do not create a new app or change its package identifier.

  1. Keep a copy of the current text, image order and screenshots before replacing them.
  2. Select the default listing language and paste its title, short and full description.
  3. Upload the eight phone PNGs in order. Remove the old phone images so the gallery does not mix the two campaigns or exceed eight slots.
  4. Replace the feature graphic; retain the icon for this first iteration.
  5. Add the tablet set in the matching device sections, not the phone section.
  6. Add localized alt text to phone images, tablet images and the feature graphic.
  7. Add/update each of the other supported language listings with its own text and graphics. Check previews; do not assume English screenshots automatically become localized.
  8. Preview the result on a phone-sized display: the first image must explain the game, headings must be readable, and the first three images must show gameplay.

Main-listing assets can appear across test tracks. An internal testing track is not a private alternate main listing. Use Console previews or a suitable custom listing for creative preparation, and review submission/publishing status carefully.

  • Recommended type/category: Game → Strategy. The public listing inspected on 28 October used Simulation. This is a recommendation, not a change already made in Console.
  • Select up to five genuinely relevant tags from Console’s available choices. Strategy, single-player and offline characteristics are relevant; do not invent tags, select multiplayer, or add accessibility tags before real-device verification.
  • Keep support email alegandro2507@gmail.com unless the owner chooses another.
  • Confirm the developer contact information and country availability.
  • Review supported form factors and Play Games on PC availability. Generated tablet images are not proof that all devices or PC distribution have been tested.
  • Leave external marketing settings to the owner’s distribution preference; there is no automatic opt-in or advertising campaign in this change.

7. Mandatory privacy, audience and release gate

Section titled “7. Mandatory privacy, audience and release gate”

The live English/Spanish listing inspected on 28 October said two languages, 15 achievements, no network permission and no data collection. Its release notes advertised 15 languages and telemetry off by default. The public privacy page dated 22 October also described opt-in telemetry. Those statements were not changed remotely.

The checked-out 0.5.0 code has share_analytics=true and share_crashes=true as defaults for new/missing settings; saved false choices are preserved. It includes Aptabase and Sentry and requests network permissions. The local privacy-policy source already describes this behavior. This marketing work does not change it.

Before submitting:

  • Confirm the actual production AAB/version and the commit used for captures. The manifest records the base commit and whether the working tree was dirty. These prepared assets include local changes until you commit them; do not mistake the recorded base commit for an already released build.
  • Match the full-description privacy paragraph to the shipped build. If the production app still has opt-in defaults, use accurate wording for that build, not the supplied defaults paragraph. Preserve honesty in every translation.
  • Review consent/disclosure requirements before shipping enabled-by-default analytics; a toggle or store description is not itself prior consent.
  • Publish the matching policy at https://macarronesc.github.io/invade.io/privacy-policy.html and verify the public contents. The existing Pages workflow deploys from main; merely editing a local HTML file does not update the public page.
  • Review Data safety for all currently distributed versions/tracks that are in scope. Optional telemetry is still collection when data leaves the device. Check app activity, crash/diagnostic data, IP-derived location and session identifiers against the SDKs and provider configuration.
  • Determine whether the service-provider sharing exception applies; do not mark all third-party processing as sharing or as no collection by assumption.
  • Keep the no-ads/no-purchases and no-account access declarations only while they remain true. No credentials are needed to review this game.
  • Confirm target audience and Families settings. The public page displayed a Families commitment, while the repository policy says the game is not directed at under-13s. Resolve this intentionally; PEGI 3 is not a target-audience choice.
  • Keep IARC/content-rating answers truthful. Do not manufacture a new rating based on a marketing preference.
  • For an app release, use an unused increasing versionCode, follow docs/RELEASES.md, and run the existing Android bundle verification. No signing credentials or release codes were changed for this store refresh.

Data safety worksheet for the owner (verify with the SDK/provider settings, not automatic form answers):

Data to investigateLikely Console category/purposeRequired confirmation
Level outcomes, time, tabs, upgradesApp activity; analyticsWhat actually leaves the device and whether every user can opt out
Crash traces, device/OS and performance contextCrash logs/diagnostics; analyticsNative report contents, retention and any attached identifiers
Country inferred from a connection IPApproximate location; analyticsWhether the provider performs/stores that inference
Temporary session IDs and SDK-generated identifiersReview Device or other IDs definitionsDo not assume a temporary identifier never needs declaration
Encrypted requests and deletion/contact processSecurity practicesVerify actual transport and what deletion requests can be fulfilled; deleting a local save is not deletion of provider data

A listing-only update need not invent an application release. If shipping the pending 0.5.0 build too, write accurate release notes from docs/store_listing/release_notes.md; do not describe captioned store screenshots as newly implemented in-game functionality. The ZIP includes that document as release-notes-reference.md; use only the matching release section, not the historical notes for another version.

8. Regeneration, validation and provenance

Section titled “8. Regeneration, validation and provenance”
  • Titles/short/full descriptions: docs/store_listing/*.md (three blocks per locale).
  • Headlines/subtitles/alt text: docs/store_listing/graphics.json (one entry per locale).
  • Scenarios, fixture, artwork layout and capture order: scripts/tests/store_assets.gd.
  • Locale mapping, upload validation and ZIP creation: tools/build_store_listings.py.
  • Fonts: tools/build_locale_fonts.py now also reads the marketing captions.

For new caption characters, regenerate the existing per-script font subsets from the licensed Noto source fonts, then let Godot import the changed resources:

Terminal window
python3 tools/build_locale_fonts.py /path/to/noto-source-fonts
godot --headless --path . --editor --import

--translate is an explicit authoring operation: it replaces non-EN/ES full description drafts and fills missing graphics locales through the existing online translator. It preserves titles/short descriptions and existing graphics edits. Do not run it over native-reviewed full descriptions without backing up those edits. Normal checking, capture and packaging do not call a translation service.

Review one locale without replacing the final set

Section titled “Review one locale without replacing the final set”
Terminal window
xvfb-run -a godot --audio-driver Dummy --path . res://scenes/tests/store_assets.tscn -- --lang=ar --no-telemetry

This writes only build/store_capture/ preview material. It does not publish a partial locale set into docs/store_assets/ or replace its manifest. Run without --lang after approval to publish the complete generated set locally. Run only one capture process per checkout; staging filenames and the synthetic save are shared. Use separate workspaces if captures need to run concurrently.

The generator resets a separate fixture for every image/language, uses fixed seeds, checks real dispatches and scroll positions, validates font coverage and caption bounds, rejects non-rendering headless use, and checks that the real save hash did not change. Telemetry is suppressed by the test-scene guard and --no-telemetry. The synthetic save path is selected before fixture interactions; production tests and capture scripts remain excluded by export_presets.cfg.

All renders must finish successfully before final files are replaced. Each file is staged and renamed; the manifest is written last. This is not a transactional filesystem-wide rollback if a disk error interrupts publication. Packaging fails closed on mismatched hashes/missing files; regenerate after repairing the error. Only the explicitly named old gallery PNGs are retired; unrelated files are not deleted.

docs/store_assets/manifest.json records version, Godot version, commit/dirty flag, fixed seed/day, scenario, raw-source path, localized copy and SHA-256 for each output. The package validator checks image sizes/color type, complete locale sets, file order, alt text and hashes. Changed captions require a new capture, not reuse of stale images. Exact pixels can vary between GPU/font-renderer versions; deterministic scenario state is not a cross-platform pixel-hash guarantee.

Terminal window
python3 -m unittest discover -s tools/tests -v
python3 tools/check_repository.py
godot --headless --path . res://scenes/tests/test_runner.tscn -- --quick
python3 tools/build_store_listings.py --package

Review real-size images and thumbnails, not only passing tests. Inspect Arabic shaping/direction, Hindi marks, CJK glyphs and long German/French captions. Also compare representative screenshots against the actual Android release on a phone and tablet. Desktop rendering does not prove touch ergonomics, TalkBack, haptics, safe areas, performance or the native SDKs’ data behavior on those devices.

PhaseReview and resulting correction
MessagingChecked feature counts against code and text lengths for all 15 locales. Removed obsolete no-network/no-collection promises; made daily play explicitly local
English creative proofReviewed first-three captures, full images and contact sheet. Changed attack timing to capture troops before arrival; fixed atlas/cosmetic scrolling; used the real dark theme so white base-shape frames remain visible; removed the banner’s clipped HUD and moved branding away from the action
LocalizationAdded caption glyphs to Noto subsets; shortened the Russian campaign headline, fixed RTL artwork placement, and rewrote incorrect/awkward translated game terminology and expedition rules. The generator checks every rendered locale for fit/glyphs; native-speaker approval remains a publication gate
Packaging and QAAdded offline field/image/hash checks, regression tests, explicit locale mapping, tablet outputs, alt text and the self-contained ZIP
Measurement preparationDefined a baseline and controlled test process below. No Console experiment or acquisition result is claimed

The operational validation result for the generated delivery is recorded in docs/store_assets/validation.md. Its hardware/native-language/Console limitations remain applicable even when automated checks pass.

  1. Record the publication date and current store visitors, acquisitions and listing conversion using Play Console’s definitions. Save the original gallery as the baseline.
  2. Segment by listing language/country and traffic source where reports support it; do not mix an advertising spike with organic conversion and call it a creative win.
  3. At sufficient traffic, test the first screenshot or first-three sequence as one variable. Keep the title, icon and other images fixed during that experiment.
  4. Choose the decision metric and stopping rule before starting; use Console’s experiment confidence reporting rather than picking a winner after a few installs. With very low traffic, collect qualitative feedback instead of declaring significance.
  5. Ask unfamiliar players what the game is, how to send troops and what they would unlock next after viewing the gallery. Correct misleading expectations first.
  6. Monitor available retention/crash reports after release. Better store conversion is not evidence of better retention. Existing aggregate session analytics do not provide persistent-user retention cohorts; no new tracking identifier was added.
  7. Test a simplified icon or a different headline only after the gallery has a useful baseline. A short real-gameplay video is a later creative project, not a requirement to use this package. No video, paid campaign or automated publishing was added.