ProgrammableMarketing/fifthdoor.com (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 7
  • This quarter · 10
  • Big bets · 3
  • Wildcards · 2
Generated Nov 10, 2026
  1. Programmable
  2. Playbooks
  3. Fifth Door
Fifth Door logo

growth playbook · Nov 10, 2026

fifthdoor.com (opens in a new tab)

Fifth Door growth playbook

Diagnosis

Fifth Door is in public beta with a working loop: 524 public game pages, new launches every day, a weekly jam, an Android app and a publishing path for coding agents. The gap is that these pieces don't link to each other.

Its "games to play" blog posts link to none of the 524 games, game pages are dead ends with no play counts and no "more like this", genre pages show nothing to search engines, and the jam has no public page.

Real visitors also fail Core Web Vitals on the homepage, with a layout shift ten times over the limit. This playbook links the content to the games first, then makes game pages work as landing pages (play counts, related games, "make one like this", score challenges), gives makers and the jam public pages, and then tests the bigger bets: group party play, partners and a paid creator plan.

Growth Score

6/10

Biggest Opportunity

Make every one of the 524 game pages and 32 blog posts a way into play and creation, so search and shared links turn into players and players turn into makers.

Biggest Risk

Rosebud (over 1 million creators, 2.5 million games) and Websim already offer free AI game making with remixing and big galleries, so a small team has to win on the loop between play and create, not on the number of features.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Put real games in the "AI games to play" posts: today they link to none of Fifth Door's 524 game pages

    Pick 6 games for the "Short session / commute energy" list in "Best AI Browser Games 2026" and link them today.

    High impact First result in 2 weeks
  2. 2

    Add a "more like this" shelf to every game page, so a player who finishes one game starts another

    Add a static "More in <genre>" row using the existing genre tags.

    High impact First result in 2 weeks
  3. 3

    Fix the homepage hero image and layout shift that make real visitors fail Core Web Vitals

    Set width and height on the hero and cover images and convert the hero to a resized WebP.

    High impact First result in 1 month

02

Your idea portfolio

22 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 7 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 2Big bets · 9Fill-ins · 8Reconsider · 3
Effort →

Showing 22 of 22 ideas

This week 7

Quick wins with a clear first action.

  • 01

    Fifth Door's sitemap lists 524 public game pages. Its blog posts aimed at players, "Best AI Browser Games 2026" and "AI-Generated Browser Games to Play", link only to the Discover page and the Android app. They don't link to one game page or name a single Fifth Door game. The "Best AI Browser Games" post even has lists such as "Short session / commute energy" and "Thinky breaks" with no games in them. Someone who searches for "AI browser games" lands on an article, then has to find a game on a feed that changes every day. The game pages lose out too: with no links pointing to them, they get little search credit.

    High impactLow effort First result in 2 weeks Content lead

    First action

    Pick 6 games for the "Short session / commute energy" list in "Best AI Browser Games 2026" and link them today.

    How to measure it

    Blog-to-play rate

    Now
    –
    Target
    25% of player-post visits open a game page
    By
    30 days

    Steps

    1. 1

      Pull the shortlist

      Top 60 public games by plays and likes over 30 days, tagged by genre and session length.

      Back-End Engineer 1 day

    2. 2

      Fill the lists

      5-8 games per list in the two player posts, each with cover art, one line and a direct link.

      Content lead 2 days

    3. 3

      Add genre proof

      Three linked games at the end of each "Make a ... with AI" post.

      Content lead 2 days

    4. 4

      Ask the makers

      Tell featured makers in Discord so they can share the post.

      Community manager 1 day

    5. 5

      Link back

      "Featured in" link on each featured game page.

      Front End Engineer 1 day

    6. 6

      Date the lists

      Show "Updated <month>" and set a monthly refresh.

      Content lead monthly

    7. 7

      Measure

      Blog-to-play rate and search clicks on the two posts and the featured game pages.

      Growth lead weekly

    First result
    2 weeks
    Cost
    Staff time only
    Owner
    Content lead

    Smallest test

    Link 6 games in one list of one post and compare clicks to the Discover link for two weeks.

    Why it works

    Fill every list in the two player posts with 5 to 8 real Fifth Door games: a cover image, one line on why it fits ("two-minute run", "turn-based, no reflexes"), and a direct link to /play/<slug>. Choose games by plays and likes, not upvotes on launch day, and refresh the lists every month with a visible "updated" date. Do the same for the genre how-to posts ("Make a Platformer with AI", "Make a Quiz Game with AI", "Make an RPG with AI"): end each one with three games of that genre made on Fifth Door, so readers see what is possible before they press Create. Then add a link from each featured game page back to the list it's in ("Featured in: Best AI browser games").

    Search visitors are one click from playing instead of two or three, which is where browse sessions drop off. Links from the blog are the main way new game pages get found and ranked, and named, linked games give AI assistants something concrete to recommend when people ask for AI games to play.

    Ready to use

    ChecklistGame list entry template

    [ ] Cover image (4:3, title in art)
    [ ] Game name linked to /play/<slug>
    [ ] One line: why it fits this list
    [ ] Session length and controls (tap, keys, turn-based)
    [ ] Maker credit linked to profile
    [ ] "Make one like this" link to Create
    [ ] "Updated <month year>" at the top of the post

    Risks and how to handle them

    • Featured games break or get unpublished

      Check links in the monthly refresh and pick games whose makers agree to keep them public.

    • Makers left out feel ignored

      Rotate the lists and explain the criteria in the post.

    Evidence

    • Your websiteBlog posts "Best AI Browser Games 2026" and "AI-Generated Browser Games to Play": both link to Discover and the Android app, but neither links to a single /play game page or names a Fifth Door game. "Best AI Browser Games" has "Criteria-based lists" such as "Short session / commute energy" and "Thinky breaks" with no games in them.Source
    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
  • 02

    A Fifth Door game page shows the title, a one-line description, the maker, a genre tag, comments and Share, and nothing after that: no related games and no next game. Every game already has genre tags (Board Games, Chess, Dungeon Crawlers, Magic), so the data for a shelf exists.

    High impactLow effort First result in 2 weeks Front End Engineer

    First action

    Add a static "More in <genre>" row using the existing genre tags.

    How to measure it

    Games per session from game-page arrivals

    Now
    –
    Target
    +30%
    By
    45 days

    Steps

    1. 1

      Define matching

      Same genre tag first, then same maker, ranked by plays and likes.

      Product lead 1 day

    2. 2

      Build the query

      Related-games endpoint with caching.

      Back-End Engineer 2 days

    3. 3

      Render the shelf

      Six cards below the game, in server-rendered HTML.

      Front End Engineer 2 days

    4. 4

      Add a next card

      "Next game" prompt on mobile when the player exits.

      Front End Engineer 2 days

    5. 5

      Exclude weak games

      Hide games with broken builds or no cover art.

      Back-End Engineer 1 day

    6. 6

      Measure

      Games per session and second-game rate.

      Growth lead weekly

    First result
    2 weeks
    Cost
    About a week of engineering
    Owner
    Front End Engineer

    Smallest test

    Ship the shelf on the 50 most-visited game pages first.

    Why it works

    Below each game, show six games from the same genre tags, sorted by plays and likes, plus two more from the same maker. On mobile, show a "Next game" card when the player leaves the game frame. Render the shelf in the page HTML so search engines can follow the links.

    Most visitors arrive on one game from a shared link or a search. Giving them a next game is the cheapest way to raise games per session, and it spreads link value across the 524 pages.

    Risks and how to handle them

    • Same few games shown everywhere

      Add a slot for newer games with few plays.

    Evidence

    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
    • Your websiteGenre pages: Discover tags games with genre links such as /browse?category=board, /browse?category=dungeon-crawler and /browse?category=chess. Read without running scripts, /browse?category=board shows only the title "Fifth Door": no heading, text or game list.Source
  • 04

    Maker profiles show total plays and a play count on every game card, but the game page itself, where a stranger decides whether to press play, shows no play count. It shows "Comments 0" twice instead.

    Medium impactLow effort First result in 2 weeks Front End Engineer

    First action

    Reuse the profile's play-count field on the game page template.

    How to measure it

    Game page visit-to-play rate

    Now
    –
    Target
    +10%
    By
    30 days

    Steps

    1. 1

      Set thresholds

      Show plays above 50 and comments above 0.

      Product lead 1 hour

    2. 2

      Expose the data

      Plays and likes on the game page API.

      Back-End Engineer 1 day

    3. 3

      Add to the page

      Under the title, server-rendered.

      Front End Engineer 1 day

    4. 4

      Hide zero comments

      Show "Be the first to comment" instead.

      Front End Engineer 2 hours

    5. 5

      A/B test

      Half of game pages for two weeks.

      Growth lead 2 weeks

    6. 6

      Roll out

      To all pages and genre lists.

      Front End Engineer 1 day

    First result
    2 weeks
    Cost
    About 2 days of engineering
    Owner
    Front End Engineer

    Smallest test

    Show plays on the 20 most-played games for a week.

    Why it works

    Show plays and likes under the game title once a game passes a small threshold (for example 50 plays), and hide the comment count until there is at least one comment. Use the same numbers on Share previews and genre pages.

    Visible popularity is the strongest quick signal for a free game, and the data is already collected. Hiding zeros keeps new games from looking abandoned.

    Risks and how to handle them

    • Low numbers discourage new makers

      Use the threshold and a "New" badge below it.

    Evidence

    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 07

    Real visitors fail all three Core Web Vitals on the homepage: LCP 4.5s, INP 270ms and a layout shift of 1.03, ten times the 0.1 limit. The page weighs 5.4 MB, 4.5 MB of it images, and the slowest element is the hero image. For a product that promises instant play in the browser, the front door is the slowest part.

    High impactMedium effort First result in 1 month Front End Engineer

    First action

    Set width and height on the hero and cover images and convert the hero to a resized WebP.

    How to measure it

    Field CLS and LCP (75th percentile)

    Now
    –
    Target
    CLS under 0.1, LCP under 2.5s
    By
    60 days

    Steps

    1. 1

      Find the shift

      Record which shelves move on load.

      Front End Engineer 1 day

    2. 2

      Reserve space

      Fixed sizes for hero, shelves and cards.

      Front End Engineer 2 days

    3. 3

      Resize images

      WebP/AVIF at display size, lazy below the fold.

      Front End Engineer 3 days

    4. 4

      Cache longer

      Long cache headers on hashed assets and covers.

      Back-End Engineer 1 day

    5. 5

      Allow zoom

      Remove user-scalable=no from the viewport.

      Front End Engineer 1 hour

    6. 6

      Fix console errors

      The 10 errors in the PageSpeed run.

      Front End Engineer 2 days

    7. 7

      Track field data

      Search Console Core Web Vitals report.

      Growth lead monthly

    First result
    1 month
    Cost
    About 2 weeks of engineering
    Owner
    Front End Engineer

    Smallest test

    Fix the hero image and reserved space first and re-run PageSpeed.

    Why it works

    Serve the hero and game covers as resized WebP/AVIF with set width and height, load covers below the fold lazily, give images long cache lifetimes (about 2.9 MB saved on repeat visits), reserve space for the shelves that load late, and remove user-scalable=no from the viewport so mobile players can zoom. Clear the 10 console errors.

    Page content that jumps while you tap leads to wrong taps and bounces on mobile. These fixes are standard and measurable within a month of real-user data.

    Risks and how to handle them

    • Allowing zoom breaks game controls

      Keep zoom off only inside the game frame, not on the site.

    Evidence

    • PageSpeedHomepage, November 10, 2026: mobile score 51, desktop 61. Real visitors fail Core Web Vitals: LCP 4.5s, INP 270ms and layout shift 1.03 (the target is under 0.1). The page weighs 5.4 MB, 4.5 MB of it images, and the largest element is the hero image; the biggest savings are cache lifetimes (about 2.9 MB) and image delivery (about 2.4 MB). The viewport blocks pinch-zoom (user-scalable=no) and 10 errors are logged to the console.Source
    • HiringGem job board (checked November 10, 2026): a Front End Engineer (US remote, first published April 25, 2025) and a Back-End Engineer (US remote, first published May 13, 2025) are still listed. "Fifth Door unlocks the power of imagination by making it easy for anyone to create, share, and play turn-based games"; the back-end engineer works on "our AI-powered game engine and web applications"; the front-end engineer builds responsive React and TypeScript apps and optimizes speed.Source
  • 08

    The homepage and Discover rank today's launches "by upvotes", and on November 10, 2026 each of the four launches showed a bare "0". To a first-time visitor, a row of zeros says nobody is here.

    Medium impactLow effort First result in 2 weeks Product lead

    First action

    Replace 0 with "New" on launches younger than 24 hours.

    How to measure it

    Launch list click-through

    Now
    –
    Target
    +15%
    By
    30 days

    Steps

    1. 1

      Change the label

      "New" instead of 0 for fresh launches.

      Front End Engineer 2 hours

    2. 2

      Blend the ranking

      Early plays plus upvotes.

      Back-End Engineer 1 day

    3. 3

      Ask at the end

      Vote prompt when a player exits a game.

      Front End Engineer 1 day

    4. 4

      Notify makers

      Tell makers when their launch gets its first votes.

      Back-End Engineer 1 day

    5. 5

      Test

      Compare clicks on the launch list.

      Growth lead 2 weeks

    6. 6

      Review

      Check the ranking still favours good games.

      Community manager weekly

    First result
    2 weeks
    Cost
    A few days of engineering
    Owner
    Product lead

    Smallest test

    Change the label only and compare a week of clicks.

    Why it works

    Until a game has a few votes, show "New" and the launch time instead of 0. Rank today's launches by early plays as well as upvotes, and ask players to vote when they finish a game rather than only on the list.

    Empty-room signals hurt both players ("is this any good?") and makers ("nobody will see my game"). Small wording changes remove the signal for free.

    Risks and how to handle them

    • Play-based ranking rewards clickbait titles

      Weight completed sessions, not clicks.

    Evidence

    • Your websiteDiscover: "Launched this week", "New games each day, ranked by upvotes"; on November 10, 2026 the Today group showed "4 launches", each with a bare "0" beside it; Launches, Hot and Top tabs; no search box.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 09

    Fifth Door already picks a Game of the day and new launches every day, but the blog has no email signup and there is no newsletter anywhere on the site. A visitor who doesn't sign up leaves with no reason to come back.

    Medium impactLow effort First result in 3 weeks Growth lead

    First action

    Add a one-field sign-up box to the blog post template.

    How to measure it

    Subscribers who play again within 7 days

    Now
    –
    Target
    30%
    By
    60 days

    Steps

    1. 1

      Pick a tool

      A simple email service with double opt-in.

      Growth lead 1 day

    2. 2

      Add sign-up

      Blog, game end screen and Discover.

      Front End Engineer 2 days

    3. 3

      Template

      Game of the day, three launches, jam theme.

      Content lead 1 day

    4. 4

      Automate

      Pull the picks from Discover data.

      Back-End Engineer 2 days

    5. 5

      Maker digest

      Weekly plays and followers for makers.

      Back-End Engineer 1 week

    6. 6

      Measure

      Opens, clicks and 7-day return.

      Growth lead weekly

    First result
    3 weeks
    Cost
    Email tool fee plus a week of work
    Owner
    Growth lead

    Smallest test

    Send a weekly email by hand to the first 100 sign-ups.

    Why it works

    Offer a short daily or weekly email: the Game of the day, three new launches, the jam theme. Sign-up goes on the blog, the end of each game and the Discover page, with one field and no account needed. Signed-in makers get a weekly note with their plays and followers.

    Fifth Door publishes something new every day; email turns that into a habit at almost no cost, and it is a channel the company owns.

    Risks and how to handle them

    • Too many emails

      Let subscribers pick daily or weekly.

    Evidence

    • Your websiteBlog index: 31 posts listed from September 23 to November 9, 2026, with no author bylines, no categories and no email signup. Several titles overlap: "AI Games Mobile Browser", "One Tap AI Games Mobile Browser" and "Free AI Games Iphone Android No App Store"; "Free AI Games Online No Download" and "AI Game Maker Online No Download"; "Best AI Game Makers 2026" and "How to Choose an AI Game Maker".Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 19

    The blog published 31 posts in about seven weeks, with no author bylines and no categories, and several target the same search: "AI Games Mobile Browser", "One Tap AI Games Mobile Browser" and "Free AI Games iPhone Android No App Store"; "Free AI Games Online No Download" and "AI Game Maker Online No Download"; "Best AI Game Makers 2026" and "How to Choose an AI Game Maker". Near-duplicate posts compete with each other in search and read as mass-produced.

    Medium impactLow effort First result in 1 month Content lead

    First action

    Merge the three mobile-play posts into one and redirect the other two.

    How to measure it

    Organic clicks to merged topics

    Now
    –
    Target
    +30%
    By
    90 days

    Steps

    1. 1

      Group topics

      Cluster the 31 posts by search intent.

      Content lead 1 day

    2. 2

      Pick winners

      One main post per cluster by clicks.

      Growth lead 1 day

    3. 3

      Merge

      Fold the best sections in.

      Content lead 3 days

    4. 4

      Redirect

      301 the merged posts.

      Front End Engineer 2 hours

    5. 5

      Add bylines

      Author name, bio and updated date.

      Front End Engineer 1 day

    6. 6

      Add sections

      Play, Make, Build with agents.

      Front End Engineer 1 day

    7. 7

      Measure

      Clicks per topic in Search Console.

      Growth lead monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Content lead

    Smallest test

    Merge one cluster and watch its clicks for a month.

    Why it works

    Pick one main post for each topic, merge the best parts of the others into it and redirect them. Add a named author with a short bio and real experience (a maker or team member), and group posts into Play, Make and Build-with-agents sections. From now on, check new topics against existing ones before writing.

    One strong page per topic ranks better than three weak ones, and visible authors are what search engines and readers use to judge whether content is trustworthy.

    Risks and how to handle them

    • Short-term ranking dip after redirects

      Redirect each post to its closest replacement and keep the URLs of the main posts.

    Evidence

    • Your websiteBlog index: 31 posts listed from September 23 to November 9, 2026, with no author bylines, no categories and no email signup. Several titles overlap: "AI Games Mobile Browser", "One Tap AI Games Mobile Browser" and "Free AI Games Iphone Android No App Store"; "Free AI Games Online No Download" and "AI Game Maker Online No Download"; "Best AI Game Makers 2026" and "How to Choose an AI Game Maker".Source

This quarter 10

The core plays for the next three months.

  • 03

    Discover tags every game with genres (Board Games, Chess, Dungeon Crawlers, Sims & Management, Magic), but each genre page is a query-string filter like /browse?category=board. Read without running scripts it shows only the title "Fifth Door", and no genre page is in the sitemap. Searches like "AI chess games" or "free browser dungeon crawler" have nothing on Fifth Door to land on.

    High impactMedium effort First result in 6 weeks Front End Engineer

    First action

    List the genres with at least 10 public games.

    How to measure it

    Organic clicks to genre pages

    Now
    –
    Target
    2,000 a month
    By
    90 days

    Steps

    1. 1

      Pick genres

      Genres with 10+ games and real search demand.

      Growth lead 2 days

    2. 2

      Clean URLs

      /games/<genre>, with the query filters redirecting to them.

      Front End Engineer 3 days

    3. 3

      Server-render

      Heading, intro and game list in the HTML.

      Front End Engineer 1 week

    4. 4

      Write intros

      80-120 words per genre, written by a person.

      Content lead 1 week

    5. 5

      Add to sitemap

      With lastmod when games change.

      Back-End Engineer 1 day

    6. 6

      Link in

      From game page tags, blog posts and the footer.

      Front End Engineer 2 days

    7. 7

      Measure

      Search Console by genre page.

      Growth lead monthly

    First result
    6 weeks
    Cost
    2-3 weeks of engineering and writing
    Owner
    Front End Engineer

    Smallest test

    Launch three genre pages and compare search impressions after a month.

    Why it works

    Give each genre with 10 or more games a clean URL (/games/board, /games/dungeon-crawler), a server-rendered heading, a short intro written by a person, the top games by plays with links, and a "Make a <genre> game" button that opens Create with a matching template. Add them to the sitemap and link to them from game pages and blog posts.

    Genre pages match the way players search, and they connect content, game pages and Create in one place: Fifth Door's stated difference that "create sits next to Discover".

    Risks and how to handle them

    • Thin pages for small genres

      Only publish genres with 10+ games; keep small ones out of the index.

    Builds on

    Evidence

    • Your websiteGenre pages: Discover tags games with genre links such as /browse?category=board, /browse?category=dungeon-crawler and /browse?category=chess. Read without running scripts, /browse?category=board shows only the title "Fifth Door": no heading, text or game list.Source
    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
    • Your websiteBlog, "Rosebud AI Alternatives": Fifth Door is "one option in the prompt-to-playable space" and its difference is that "create sits next to Discover so inspiration and shipping stay connected"; it does not claim engine parity or Steam export.Source
  • 05

    Fifth Door says its edge is that "create sits next to Discover", yet game pages have no remix or "make one like this" option. Rosebud lets players copy many games into their own editor and Websim shows remix versions. A player who just enjoyed a game is the warmest possible new maker.

    High impactMedium effort First result in 1 month Product lead

    First action

    Write the prompt template that turns a game's genre and description into a Create starter.

    How to measure it

    Game page to Create start rate

    Now
    –
    Target
    5% of game page visitors
    By
    60 days

    Steps

    1. 1

      Write the template

      Genre + description + "make it your own" prompt.

      Product lead 1 day

    2. 2

      Add the button

      Under Share on game pages.

      Front End Engineer 2 days

    3. 3

      Prefill Create

      Pass the starter prompt to the create session.

      Back-End Engineer 3 days

    4. 4

      Credit the original

      "Inspired by" link on the new game.

      Back-End Engineer 2 days

    5. 5

      Opt-in remix

      Maker setting to allow file-level remixing.

      Back-End Engineer 2 weeks

    6. 6

      Measure

      Create starts and publishes from the button.

      Growth lead weekly

    First result
    1 month
    Cost
    3-4 weeks of engineering
    Owner
    Product lead

    Smallest test

    Add the prompt-only button to 20 popular games.

    Why it works

    Add a "Make one like this" button that opens Create with a prompt built from the game's genre and description ("A roguelike wizard duel, but...") and credits the original. Later, let makers opt in to full remixing of their game's files, with a "remixed from" link back.

    It turns the hardest step for a new maker, the blank prompt, into editing an idea they just played. Credit links send traffic back to the original maker, which gives makers a reason to allow it.

    Risks and how to handle them

    • Makers dislike copies

      Idea-level prompts by default; file remix only when the maker opts in.

    Evidence

    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
    • Your websiteBlog, "Rosebud AI Alternatives": Fifth Door is "one option in the prompt-to-playable space" and its difference is that "create sits next to Discover so inspiration and shipping stay connected"; it does not claim engine parity or Steam export.Source
    • Competitor researchRosebud's homepage: "Create games with AI"; over 1 million creators and 2.5 million games; a free start with 10,000 credits; genre galleries; remixing games into your own editor; multiplayer rooms joined from a link; Steam export on Pro; publishing to Rosebud Play and X Playables.Source
  • 06

    Fifth Door "runs its own weekly jam with a theme and a deadline", but the theme, deadline and entry count are visible only through an MCP tool, list_jams. There is no jam page in the sitemap and no linked entries, so players can't play the jam and makers who don't use a coding agent can't find it.

    Medium impactMedium effort First result in 3 weeks Community manager

    First action

    Publish this week's theme and entries on a static /jam page.

    How to measure it

    Weekly jam entries

    Now
    –
    Target
    3x current entries
    By
    60 days

    Steps

    1. 1

      Build the page

      Theme, deadline, entries from the jam data.

      Front End Engineer 1 week

    2. 2

      Web entry

      Enter the jam from Create, not just the MCP tool.

      Back-End Engineer 1 week

    3. 3

      Archive weeks

      /jam/<week> with winners, in the sitemap.

      Front End Engineer 3 days

    4. 4

      Announce

      Theme post in Discord every week.

      Community manager weekly

    5. 5

      Pick winners

      Plays, likes and a staff pick.

      Community manager weekly

    6. 6

      Measure

      Entries, plays per entry and returning makers.

      Growth lead weekly

    First result
    3 weeks
    Cost
    2 weeks of engineering plus weekly community time
    Owner
    Community manager

    Smallest test

    Post one week's theme and entries on a simple page and share it in Discord.

    Why it works

    Publish /jam with this week's theme, deadline, a countdown, the entries so far and an Enter button that works from the web Create flow too. Keep an archive page per week (/jam/<week>) with winners, and announce each theme in Discord and the Game of the day email.

    A weekly theme gives makers a reason to publish this week and players a fresh, named collection to play. Archive pages pile up into a library of themed games that search engines can find.

    Risks and how to handle them

    • Low-effort entries

      Show staff picks first and keep the rule that entries go public during jam week.

    Evidence

    • Your websiteBlog, "AI Game Jam": "Fifth Door runs its own weekly jam with a theme and a deadline." The theme, deadline and entry count are shown only through the MCP tool list_jams, and games are entered with enter_jam; the post links no jam entries and the sitemap has no jam page.Source
    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
  • 10

    The Fifth Door SDK already provides leaderboards and multiplayer rooms, and the seed thesis is games "you can share with friends, instantly". Yet the only sharing on a game page is a generic Share button that sends a plain link.

    High impactMedium effort First result in 6 weeks Product lead

    First action

    List the games that already use SDK leaderboards and pick five to pilot.

    How to measure it

    Challenge links opened per week

    Now
    –
    Target
    1,000
    By
    90 days

    Steps

    1. 1

      Pick pilot games

      Five games using leaderboards.

      Product lead 1 day

    2. 2

      Challenge API

      Create a link with score and challenger.

      Back-End Engineer 1 week

    3. 3

      End-of-game card

      "Challenge a friend" with native share.

      Front End Engineer 1 week

    4. 4

      Landing state

      The score to beat shown before play.

      Front End Engineer 3 days

    5. 5

      Beaten notice

      Tell the challenger by email or in the feed.

      Back-End Engineer 3 days

    6. 6

      Document it

      SDK docs so agent-built games can use it.

      Back-End Engineer 2 days

    7. 7

      Measure

      Links sent, opened and new sign-ups.

      Growth lead weekly

    First result
    6 weeks
    Cost
    3-4 weeks of engineering
    Owner
    Product lead

    Smallest test

    Add a challenge link to one popular leaderboard game.

    Why it works

    When a player finishes a game that uses a leaderboard, offer "Challenge a friend": a link that opens the game with the player's score and name shown ("Beat 4,210"), and tells the challenger when they've been beaten. Show a challenge card in link previews, and count accepted challenges per game for makers.

    A score challenge gives people a reason to send the link, and the friend who receives it has a goal before they press play. It is the social loop the company was funded to build.

    Risks and how to handle them

    • Cheated scores

      Use server-checked scores where the SDK allows it and cap obvious outliers.

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • PressGamesBeat (November 24, 2025): a $20 million seed round led by investors from Y Combinator, Felicis and Protagonist Ventures; six employees; a focus on social play, starting with simple party games such as a trivia game where AI judges whether answers are right; "Fifth Door lets anyone turn an idea into a game they can share with friends, instantly."Source
    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
  • 12

    Maker profiles already show plays, likes, followers, achievements and a "most played" card, and the homepage features "Top makers this week", but profiles have no bio and no links. Makers have nothing to point their own audience to, and no reason to return when their numbers grow.

    Medium impactLow effort First result in 1 month Community manager

    First action

    Add bio and link fields to the profile settings.

    How to measure it

    Second-game rate within 30 days

    Now
    –
    Target
    +20%
    By
    90 days

    Steps

    1. 1

      Add fields

      Bio, three links, pinned game.

      Front End Engineer 3 days

    2. 2

      Milestones

      100 and 1,000 play notices.

      Back-End Engineer 3 days

    3. 3

      Share cards

      Image card for each milestone.

      Front End Engineer 3 days

    4. 4

      Weekly note

      Plays and followers by email.

      Back-End Engineer 1 week

    5. 5

      Spotlight

      One maker interview a week in Discord and the blog.

      Community manager weekly

    6. 6

      Measure

      Second-game rate and profile visits.

      Growth lead monthly

    First result
    1 month
    Cost
    About 2 weeks of engineering
    Owner
    Community manager

    Smallest test

    Send milestone notes by hand to this week's top makers.

    Why it works

    Add a short bio, up to three links (YouTube, TikTok, itch.io, Discord) and a pinned game to profiles. Send makers a weekly note ("Your games got 340 plays and 6 new followers") and a notice when a game passes 100 or 1,000 plays, with a ready-made share card.

    Makers who see an audience keep making, and makers who promote their profile bring new players. Milestone cards turn growth into posts on other platforms.

    Risks and how to handle them

    • Spam links

      Allow links only after a first public game; review reported profiles.

    Evidence

    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 13

    The sitemap lists 524 game pages but no maker profiles, even though profiles are rich pages with play counts and game grids. Makers who search their own name or handle, and players looking for "more from this maker", have no indexed page to find.

    Medium impactLow effort First result in 6 weeks Back End Engineer

    First action

    Count makers with three or more public games.

    How to measure it

    Organic clicks to maker profiles

    Now
    –
    Target
    500 a month
    By
    90 days

    Steps

    1. 1

      Set the rule

      Three or more public games.

      Product lead 1 hour

    2. 2

      Server-render

      Heading, bio and game list in the HTML.

      Front End Engineer 3 days

    3. 3

      Add to sitemap

      With lastmod on new games.

      Back-End Engineer 1 day

    4. 4

      Noindex thin profiles

      Below the threshold.

      Back-End Engineer 1 day

    5. 5

      Link from games

      Maker name links to the profile on every game page.

      Front End Engineer 1 day

    6. 6

      Measure

      Profile search clicks.

      Growth lead monthly

    First result
    6 weeks
    Cost
    About a week of engineering
    Owner
    Back End Engineer

    Smallest test

    Index the top 50 maker profiles first.

    Why it works

    Include profiles of makers with at least three public games in the sitemap, with a server-rendered heading, bio (from idea 12) and game list. Leave thin or new profiles out of the index.

    Profiles collect links from makers' own social accounts and pass them to their games. They are a second indexable surface that grows with every active maker.

    Risks and how to handle them

    • Makers want privacy

      Add a setting to keep a profile out of search.

    Builds on

    Evidence

    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 14

    Fifth Door's MCP server lets Cursor, Claude Code, Codex "or any MCP client" publish games, and the blog has a guide to making a game with Claude Code and one on publishing a "vibe coded" game. But the guide links no example game made that way, so a developer can't see the result before installing anything.

    Medium impactLow effort First result in 1 month Content lead

    First action

    Pull the list of games published through the MCP server and pick ten.

    How to measure it

    Games published through MCP per week

    Now
    –
    Target
    +50%
    By
    90 days

    Steps

    1. 1

      Tag the source

      Mark games published through MCP or the Partner API.

      Back-End Engineer 1 day

    2. 2

      Pick ten

      Best agent-built games with maker consent.

      Community manager 3 days

    3. 3

      Collect notes

      Tool, prompts and SDK features used.

      Content lead 1 week

    4. 4

      Publish the page

      /built-with-agents in the sitemap.

      Front End Engineer 2 days

    5. 5

      Link from docs

      Docs overview, MCP page and blog guides.

      Content lead 1 day

    6. 6

      Refresh monthly

      Add new standouts.

      Content lead monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Content lead

    Smallest test

    Add three linked example games to the "Make a Game with Claude Code" post.

    Why it works

    Tag games published through the MCP server or Partner API, and publish a "Built with coding agents" page: the best games, the tools used, and a short note from each maker on the prompts and SDK features (leaderboards, rooms, saves) they used. Link it from the docs, the coding-agent blog posts and the MCP install page.

    Developers decide on a platform from working examples. A showcase also gives makers a reason to publish here rather than to a generic host, and it answers "where do I put my vibe-coded game?"

    Risks and how to handle them

    • Few strong examples yet

      Start with five and build a few in-house.

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
  • 15

    Fifth Door has an MCP server installed with one command and an IDE extension, both documented on its own site. Developers who browse MCP directories and extension marketplaces for "publish a game" tools only find it if it's listed there.

    Medium impactLow effort First result in 1 month Back End Engineer

    First action

    List the directories and marketplaces where similar MCP servers appear.

    How to measure it

    New MCP logins per week

    Now
    –
    Target
    2x
    By
    60 days

    Steps

    1. 1

      Map directories

      MCP registries and IDE marketplaces.

      Growth lead 1 day

    2. 2

      Write the listing

      One line, features, demo GIF.

      Content lead 1 day

    3. 3

      Submit

      Each directory's process.

      Back-End Engineer 1 week

    4. 4

      Track source

      Tag installs by directory where possible.

      Back-End Engineer 2 days

    5. 5

      Keep current

      Update listings with each release.

      Back-End Engineer monthly

    6. 6

      Measure

      Logins and first publishes.

      Growth lead monthly

    First result
    1 month
    Cost
    $0
    Owner
    Back End Engineer

    Smallest test

    Submit to two directories and compare installs.

    Why it works

    Submit the MCP server to the main MCP directories and registries and the IDE extension to the extension marketplaces it supports, with a clear one-line description ("Publish browser games from your coding agent, with leaderboards, rooms and saves"), a demo GIF of publish_game and links to the showcase (idea 14). Track installs by source.

    Developers find tools in directories, so a listing puts Fifth Door in front of people who are already building with agents, at no cost.

    Risks and how to handle them

    • Outdated listings

      Add listing updates to the release checklist.

    Builds on

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
  • 17

    Fifth Door's own blog says "Apple App Store listing: not available yet" and tells iPhone players to "add a Home Screen bookmark". Android players get a native app, while iPhone players get a bookmark.

    Medium impactMedium effort First result in 1 month Front End Engineer

    First action

    Check the manifest and start URL so a home-screen launch opens full screen on iPhone.

    How to measure it

    iPhone players returning within 7 days

    Now
    –
    Target
    +20%
    By
    60 days

    Steps

    1. 1

      Audit the manifest

      Name, icons, start URL, display mode.

      Front End Engineer 1 day

    2. 2

      Offline shell

      Service worker for the app shell.

      Front End Engineer 1 week

    3. 3

      Install hint

      After the second game, once only.

      Front End Engineer 3 days

    4. 4

      Update the blog

      iPhone instructions point to install.

      Content lead 1 hour

    5. 5

      Track installs

      Home-screen launches by device.

      Back-End Engineer 2 days

    6. 6

      Measure

      iPhone 7-day return.

      Growth lead monthly

    First result
    1 month
    Cost
    About 2 weeks of engineering
    Owner
    Front End Engineer

    Smallest test

    Show the install hint to 10% of iPhone visitors.

    Why it works

    Complete the web app manifest and offline shell so "Add to Home Screen" opens Fifth Door full screen, then show iPhone visitors a one-time, well-timed install hint after their second game. Point the blog's iPhone instructions at it, and keep the same users when the App Store app ships.

    A home-screen icon is the strongest return trigger on a phone. iPhone users currently get the weakest experience, and this closes the gap without waiting for App Store review.

    Risks and how to handle them

    • Game frames behave differently in home-screen mode

      Test the top 20 games in standalone mode first.

    Builds on

    Evidence

    • Your websiteBlog, "Free AI Games iPhone Android No App Store": "Apple App Store listing: not available yet"; on Android a native app is on Google Play ("Fifthdoor: Infinite Arcade"); iPhone players are told to play in the browser and "add a Home Screen bookmark for faster return".Source
    • PageSpeedHomepage, November 10, 2026: mobile score 51, desktop 61. Real visitors fail Core Web Vitals: LCP 4.5s, INP 270ms and layout shift 1.03 (the target is under 0.1). The page weighs 5.4 MB, 4.5 MB of it images, and the largest element is the hero image; the biggest savings are cache lifetimes (about 2.9 MB) and image delivery (about 2.4 MB). The viewport blocks pinch-zoom (user-scalable=no) and 10 errors are logged to the console.Source
  • 18

    The Fifth Door SDK already supports "daily puzzles", the blog has a guide to making a daily puzzle game, and the homepage has a "Puzzles" template. But there's no place where players find today's puzzles together, with a streak to keep.

    High impactMedium effort First result in 6 weeks Product lead

    First action

    List the games already using the daily puzzle feature.

    How to measure it

    Players with a 3+ day streak

    Now
    –
    Target
    1,000
    By
    90 days

    Steps

    1. 1

      Find the games

      Games using SDK daily puzzles.

      Back-End Engineer 1 day

    2. 2

      Build the shelf

      Homepage row and /daily page.

      Front End Engineer 1 week

    3. 3

      Streaks

      One streak across all daily puzzles.

      Back-End Engineer 1 week

    4. 4

      Share grid

      Spoiler-free result to copy.

      Front End Engineer 3 days

    5. 5

      Seed supply

      A daily-puzzle jam theme.

      Community manager 1 week

    6. 6

      Email it

      In the daily email.

      Growth lead 1 day

    7. 7

      Measure

      Streaks and next-day return.

      Growth lead weekly

    First result
    6 weeks
    Cost
    3 weeks of engineering
    Owner
    Product lead

    Smallest test

    Put three daily puzzles in one homepage row for two weeks.

    Why it works

    Add a "Today's puzzles" shelf on the homepage and a /daily page: every game using the SDK's daily puzzle feature, a streak counter across all of them, and a shareable result grid. Send the shelf in the Game of the day email (idea 9) and invite makers to build daily puzzles through a jam theme.

    A daily puzzle is a habit: one short session a day, a streak to protect and a result to share. Fifth Door can offer many daily puzzles in one place, which a single-puzzle site can't.

    Risks and how to handle them

    • Too few good daily puzzles

      Launch with five strong ones and feature them in rotation.

    Builds on

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • Your websiteSitemap (565 URLs, checked November 10, 2026): 524 game pages under /play (updated August 27 to November 10, 2026), 32 blog URLs, 5 docs pages, plus about, terms and privacy. It has no genre (/browse) pages, no creator profile pages and no jam page.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source

Big bets 3

Larger plays that can change the trajectory.

  • 11

    Fifth Door was funded on social play: "small, shared play experiences" and party games, starting with a trivia game where AI judges answers, so groups can personalize games around their own jokes. Today's homepage is a solo arcade of daily launches, and the SDK's multiplayer rooms aren't shown to players at all.

    High impactHigh effort First result in 2 months Founder & CEO

    First action

    Ship one trivia party template on multiplayer rooms to Discord members.

    How to measure it

    New players joining through room codes

    Now
    –
    Target
    30% of new sign-ups
    By
    2 quarters

    Steps

    1. 1

      Pick the format

      Trivia with personalized questions first.

      Founder & CEO 1 week

    2. 2

      Build the template

      On SDK multiplayer rooms.

      Back-End Engineer 1 month

    3. 3

      Phone join flow

      Room code, no account needed to play.

      Front End Engineer 3 weeks

    4. 4

      Pilot

      Discord game nights.

      Community manager 1 month

    5. 5

      Homepage entry

      "Play with your group" at evenings and weekends.

      Front End Engineer 1 week

    6. 6

      Template library

      Best group-made games as starters.

      Product lead ongoing

    7. 7

      Measure

      Joins per room and new sign-ups.

      Growth lead weekly

    First result
    2 months
    Cost
    1-2 engineers for a quarter
    Owner
    Founder & CEO

    Smallest test

    Run three Discord game nights on one trivia template.

    Why it works

    Build a "Party" entry point: pick a party template (trivia, drawing, guessing), type in the group's in-jokes or topic, get a game in minutes, and share a room code that friends join from their phones. Feature it on the homepage for evenings and weekends, and publish the best group-made templates as starting points.

    Party games spread by being played together; every session brings several new players into a room. It is also where Fifth Door is clearly different from creator-tool rivals.

    Risks and how to handle them

    • Group sessions are hard to get right

      Pilot one format with Discord before a public launch.

    Builds on

    Evidence

    • PressGamesBeat (November 24, 2025): a $20 million seed round led by investors from Y Combinator, Felicis and Protagonist Ventures; six employees; a focus on social play, starting with simple party games such as a trivia game where AI judges whether answers are right; "Fifth Door lets anyone turn an idea into a game they can share with friends, instantly."Source
    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
  • 16

    Fifth Door's Partner API already lets hosted agents "create and iterate games over REST", with OAuth sign-in or API keys, but the docs list no partner terms, showcase or listing of partners. GamesBeat said the seed would fund "onboarding creators and studios".

    High impactHigh effort First result in 1 quarter Founder & CEO

    First action

    List 20 products whose users make small games and have no place to publish them.

    How to measure it

    Partner-published games per month

    Now
    –
    Target
    500
    By
    2 quarters

    Steps

    1. 1

      Shortlist partners

      20 products whose users make games.

      Founder & CEO 1 week

    2. 2

      Write terms

      Attribution, stats and limits.

      Founder & CEO 2 weeks

    3. 3

      Publish button

      Drop-in "Publish to Fifth Door" flow.

      Back-End Engineer 1 month

    4. 4

      Partner page

      Docs, logos and showcase.

      Content lead 2 weeks

    5. 5

      First cohort

      5-10 partners live.

      Founder & CEO 1 quarter

    6. 6

      Shared stats

      Plays per partner dashboard.

      Back-End Engineer 1 month

    7. 7

      Measure

      Partner games and their plays.

      Growth lead monthly

    First result
    1 quarter
    Cost
    Founder time plus a month of engineering
    Owner
    Founder & CEO

    Smallest test

    Sign one partner and publish their users' games for a month.

    Why it works

    Package the API as a program: a partner page, a "Publish to Fifth Door" button partners can add to their own products, a partner badge on games, shared stats on plays, and a first cohort of 5-10 partners (game-making apps, AI tutors, creative coding tools). Every game made in a partner's product becomes a Fifth Door game page with an audience.

    Partners bring makers Fifth Door would never reach on its own, and Fifth Door gives them what they lack: a place to play, share and rank games.

    Risks and how to handle them

    • Low-quality partner games flood Discover

      Partner games start unlisted, as with MCP, and earn their way onto Discover.

    Builds on

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • PressGamesBeat (November 24, 2025): a $20 million seed round led by investors from Y Combinator, Felicis and Protagonist Ventures; six employees; a focus on social play, starting with simple party games such as a trivia game where AI judges whether answers are right; "Fifth Door lets anyone turn an idea into a game they can share with friends, instantly."Source
  • 20

    Fifth Door has no pricing page, but asset generation already has "a daily per-account budget", so there is a natural limit to sell more of. Rosebud starts makers free with 10,000 credits and charges for more credits and features such as Steam export.

    High impactMedium effort First result in 1 quarter Founder & CEO

    First action

    Count how many makers hit the daily budget each week.

    How to measure it

    Conversion of budget-limited makers to paid

    Now
    –
    Target
    5%
    By
    2 quarters

    Steps

    1. 1

      Size demand

      Makers hitting the budget each week.

      Back-End Engineer 2 days

    2. 2

      Interview

      Ten heavy makers on what they would pay for.

      Founder & CEO 2 weeks

    3. 3

      Design the plan

      Budget, limits, stats and price.

      Founder & CEO 1 week

    4. 4

      Waitlist

      In-product offer when the budget runs out.

      Front End Engineer 1 week

    5. 5

      Billing

      Subscription checkout.

      Back-End Engineer 2 weeks

    6. 6

      Founding cohort

      Top makers at a fixed early price.

      Community manager 1 month

    7. 7

      Measure

      Conversion, churn and generation cost per maker.

      Growth lead monthly

    First result
    1 quarter
    Cost
    Billing setup and about a month of engineering
    Owner
    Founder & CEO

    Smallest test

    Show a paid-plan waitlist when the daily budget runs out and count sign-ups.

    Why it works

    Test a creator plan for heavy makers: a larger daily generation budget, more files and storage per game (above 200 files and 300 MB), priority builds, custom game URLs and detailed play stats. Keep playing and basic making free. Start with a waitlist and a fixed founding price for top makers, and decide on the price from how many convert.

    The users who hit the daily budget are the ones who value the product most. Charging them funds generation costs without adding friction for new makers.

    Risks and how to handle them

    • Makers leave for free rivals

      Keep free making generous and charge only for heavy use.

    Evidence

    • Your websiteDocs and "Make a Game with Claude Code": an MCP server for Cursor, Claude Code, Codex "or any MCP client" publishes and updates games, lists them on Discover and records gameplay previews; new games start unlisted; a Partner API lets hosted agents "create and iterate games over REST"; the Platform SDK "provides saved progress, leaderboards, multiplayer rooms, and daily puzzles"; asset generation has "a daily per-account budget". The post links no example game made this way.Source
    • Competitor researchRosebud's homepage: "Create games with AI"; over 1 million creators and 2.5 million games; a free start with 10,000 credits; genre galleries; remixing games into your own editor; multiplayer rooms joined from a link; Steam export on Pro; publishing to Rosebud Play and X Playables.Source
    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source

Wildcards 2

Unexpected experiments, each with a cheap way to test it.

  • 21

    Sharing a Fifth Door game today means pressing Share and sending a link. Rosebud publishes games into X Playables so they play inside the feed. Makers who have their own sites, blogs or newsletters have no way to show a playable game there.

    Medium impactMedium effort First result in 2 months Front End Engineer

    First action

    Build an embed URL that shows one game with credit and a link back.

    How to measure it

    Monthly plays on embedded games

    Now
    –
    Target
    10,000
    By
    2 quarters

    Steps

    1. 1

      Embed route

      Minimal player page for iframes.

      Front End Engineer 1 week

    2. 2

      Credit bar

      Maker credit and "Play more" link.

      Front End Engineer 2 days

    3. 3

      Embed button

      Copy snippet next to Share.

      Front End Engineer 2 days

    4. 4

      Track

      Plays per embedding site.

      Back-End Engineer 3 days

    5. 5

      Opt-out

      Makers can turn embedding off.

      Back-End Engineer 1 day

    6. 6

      Promote

      Tell makers in Discord and email.

      Community manager 1 week

    First result
    2 months
    Cost
    About 2 weeks of engineering
    Owner
    Front End Engineer

    Smallest test

    Give embed snippets to 20 active makers and count plays.

    Why it works

    Offer an "Embed" option next to Share: a small iframe snippet with the game, the maker's credit and a "Play more on Fifth Door" link. Track plays from embeds per site, and feature the best embedding sites in the maker newsletter.

    Every embed is a permanent, playable ad on someone else's audience, with a link back. It also gives teachers, writers and community sites a reason to use Fifth Door games.

    Risks and how to handle them

    • Embeds on unsuitable sites

      Let makers block domains and remove embeds on report.

    Evidence

    • Competitor researchRosebud's homepage: "Create games with AI"; over 1 million creators and 2.5 million games; a free start with 10,000 credits; genre galleries; remixing games into your own editor; multiplayer rooms joined from a link; Steam export on Pro; publishing to Rosebud Play and X Playables.Source
    • Your websiteGame pages (Hexmoor Academy and Walkies Inc., checked November 10, 2026): a title, a one-line description, the maker and their follower count, a genre tag, "Comments 0" and a Share button, in about 130-150 words of readable text, mostly navigation and the cookie notice. No play count, no "more like this" games and no remix option.Source
  • 22

    The homepage invites new makers to start from five templates (Arcade, Platformer, Adventure, Puzzle and Random). Rosebud and Websim offer the same free promise. What decides who wins makers is how many first prompts end in a game the maker is proud enough to publish, and Fifth Door doesn't report that number anywhere public.

    High impactMedium effort First result in 1 month Product lead

    First action

    Instrument the five funnel steps and pull last month's numbers.

    How to measure it

    New makers who publish within 24 hours

    Now
    –
    Target
    +25%
    By
    90 days

    Steps

    1. 1

      Define steps

      Prompt, build, follow-up, publish, share.

      Product lead 1 day

    2. 2

      Instrument

      Events for each step.

      Back-End Engineer 3 days

    3. 3

      Review sessions

      50 failed first sessions a week.

      Product lead weekly

    4. 4

      In-product prompts

      Suggested follow-up prompts in Create.

      Front End Engineer 1 week

    5. 5

      Tune templates

      Rewrite the weakest template starters.

      Product lead 1 week

    6. 6

      Report

      Weekly funnel to the team.

      Growth lead weekly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Product lead

    Smallest test

    Review 20 failed first sessions this week and fix one cause.

    Why it works

    Define the funnel (first prompt, first playable build, first follow-up prompt, first publish, first public share) and review 50 failed first sessions by hand each week. Fix the top cause each week: better template prompts, suggested follow-up prompts from the blog's "follow-up prompts" guide shown inside Create, or faster first builds.

    Every acquisition idea in this plan pays off more when more new makers get a working game. A weekly review of failed sessions finds fixes that dashboards miss.

    Risks and how to handle them

    • Reviewing sessions raises privacy concerns

      Review prompts and build outcomes only, under the privacy policy, without personal data.

    Evidence

    • Your websiteHomepage (checked November 10, 2026): fifthdoor.com redirects to play.fifthdoor.com, titled "Fifth Door: make and play AI games in your browser", with the banner "Welcome to Fifth Door public beta! Build games with AI and share them with friends."; a Game of the day, "Launched today" games ranked by upvotes, "Top makers this week" and templates Arcade, Platformer, Adventure, Puzzle and Random. Creator profiles linked from it show total plays, likes, followers, achievements and a play count on each game card, but no bio. There is no pricing page.Source
    • Competitor researchWebsim's homepage: "Create games and apps for free and help make the internet fun again"; remix versions of projects; Hot and Top sorting by today, week, month and all time; projects built around comments.Source
    • Competitor researchRosebud's homepage: "Create games with AI"; over 1 million creators and 2.5 million games; a free start with 10,000 credits; genre galleries; remixing games into your own editor; multiplayer rooms joined from a link; Steam export on Pro; publishing to Rosebud Play and X Playables.Source
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Samlino Group growth strategyA European operational investment group that launches, buys and runs consumer comparison platforms for banking, insurance and utilities…22 growth ideas
  • RiseGuide growth strategyAn expert-powered self-improvement and microlearning app: daily 15-minute lessons in Communication Mastery, Intelligence Training and…22 growth ideas
  • refurbed growth strategyA Vienna-based online marketplace for refurbished products (phones, laptops, tablets, household appliances, sport, kids' products and…22 growth ideas
  • Humaans growth strategyAn HRIS for fast-growing global companies, plus Athena, AI agents that answer employee requests in Slack and Teams and carry out People…22 growth ideas
  • Huscarl growth strategyAn AI-native actuarial advisory for corporate self-insurance: captive feasibility studies, pricing and underwriting for hard-to-price…22 growth ideas
  • JumpCloud growth strategyAn AI-powered unified IT platform for identity, device and access management, covering SSO, MFA, MDM for Windows, Mac, Linux and Android…22 growth ideas
  • Vesence growth strategyAn agent-native computer: a cloud workspace where many agents work in parallel across mail, calendar, files, a browser, Office documents…23 growth ideas
  • Alloy River growth strategyAcquires and partners with established bookkeeping, accounting and fractional finance firms, keeps their brands and founders, and runs…22 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 10, 2026, and cites them throughout. Fifth Door did not commission, review or endorse it, and Programmable is not affiliated with Fifth Door. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Fifth Door runs its business.

Something wrong, or do you represent Fifth Door? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 307 ideas

    Put real games in the blog, add "more like this" and play counts to game pages, fix homepage speed and layout shift, stop showing zero votes, start a Game of the day email and merge overlapping posts.

  2. Days 31 to 6010 ideas

    Indexable genre pages, "make one like this", a public jam page, score challenges, maker bios and profiles, the coding-agent showcase and MCP listings, iPhone home-screen install and a daily puzzle shelf.

  3. Days 61 to 905 ideas

    The big bets: group party mode, partner integrations and a paid creator plan; plus the embeddable player and the first-game success funnel.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value turning an idea into a playable game without code. The most cited problem: finding good games among hundreds of new ones.

  • Playing short games in the browser with no download
  • Not knowing what to make first
  • In their words: “make a game with AI”

Market

Competitive landscape

3 competitors compared: Rosebud, Websim and itch.io (AI-tagged browser games).

  • Gap to exploit: A play-first destination for AI-made games with clear quality signals
  • Rosebud: Creator-first; play is a showcase rather than the core
  • Websim: Games and apps mixed together, less of a game destination

Hiring signals

Fifth Door's job board still lists two US-remote engineering roles, a Front End Engineer and a Back-End Engineer, first published in spring 2025.

  • Front End Engineer: Responsive React and TypeScript apps and speed. Ideas 2, 3, 4, 7, 17, 21.
  • Back-End Engineer: Backend for the AI-powered game engine. Ideas 6, 10, 16, 20.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks first (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence