Store page 0/16 The page goes up the day the trailer exists, not the day the game is done. Every devlog, post and pitch points at it.
Capsule / header image readable at 231×87 px. Test it shrunk before you test it big Title legible without the logo’s flourishes: shrink to a thumbnail and squint Key art signals the real genre and tone. If testers ask “is this a horror game?” and it isn’t, change the art Short description under 30 words: genre + verb + hook, in that order Genre tags match what the game actually is, not what sells First paragraph of the long description works alone; most readers stop there Feature list is concrete: “7 areas, 3–4 hours, no combat, no fail state” Say what the game is NOT (“no combat, no timers you can fail”). It filters bad refunds Honest length estimate. Overstating it is the fastest route to bad reviews No hype words: revolutionary, stunning, unique, immersive Accessibility section lists only what ships (rebinding, subtitles, colour-blind options, no timed fails), never what you plan Minimum and recommended specs tested on a real low-end machine, not guessed Supported languages listed separately for interface, audio and subtitles; content rating questionnaire answered Open the page on a phone. Fix what breaks Ask one person who has never seen the game what it is. If they are wrong, the page is wrong Every claim on the page is true of the current build All done in this section.
Trailer 0/12 Target 75 seconds, no narration. Every shot answers “what do I do?” or “how does it feel?”; a shot that answers neither is cut, however pretty.
All done in this section.
Screenshots & GIFs 0/13 Screenshots and GIFs are the product for everyone who has not played it. Capture on purpose, not from whatever happened to be on screen.
Capture from a release build: no debug overlay, no fps counter, no console, no version watermark Placeholder art out of frame; settings at the values a normal player would have Stills at native resolution, 1920×1080 minimum; keep a 3840×2160 set for the store page PNG for stills, never a re-compressed JPEG. Hide the cursor 6–8 store screenshots, ordered: 1 the hook, 2–3 the core loop mid-action, 4–5 variety, 6 something warm with a character in it No letterboxing, no text overlays added in an editor Compose every still: one readable landmark, no empty half of the frame; six keepers per area, you will use one GIFs 3–6 seconds, loop without a visible seam, start on motion (the first frame is the thumbnail) One idea per clip; no UI transitions, pause menus or loading screens Under 5 MB for forums, under 15 MB for a store video source; 30 fps, 800–1000 px wide for social No music bed over a mechanic clip; caption anything explained by sound Name files YYYY-MM-DD-area-subject.png, store them outside the project folder, backed up, raw and export both kept Log every image you use publicly so you know which one moved the numbers All done in this section.
Press kit & outreach 0/12 A list of 40 people who might actually care beats 400 addresses scraped off a page. No blast: every message names one specific thing that person covered.
Press kit: transparent PNG logo, 6 screenshots, trailer file, 3-paragraph factsheet Studio name, contact email, release date, price and platforms in one place One zip under 100 MB at a permanent URL, linked from the store page Licence line in the kit: “press and creators may use these assets freely” List 10 released games genuinely near yours in tone and scope. Peers, not hits For each, find who covered it. Keep only people who covered two or more of your ten Use each contact’s stated route (public email, submission form) and obey their stated rules exactly Add one researched contact per working day; 40 contacts costs eight weeks of five minutes Six-line pitch, no attachment: subject with genre and build state · the thing they covered · genre + verb + hook · build state, window, platforms · trailer, kit and store links · “key on request, no embargo, no obligation” Never send the same text twice; one follow-up after 7 days, then stop forever Never buy or scrape a list. Never pitch anyone who publishes “no unsolicited pitches” Keys generated in small batches, each logged against a name and date; never post an unclaimed key publicly All done in this section.
Pricing 0/12 One decision, made once, defended for a year. Price is a positioning signal before it is a revenue lever.
List 10 released peers with your length, scope and audience. Write down each one’s launch price and current price Take the median launch price. That is your band Move up one step only for something peers lack (full voice, big art budget, long content tail); down one step only if your length is at the bottom of the set Never price below the band to “get started”. You cannot raise a price later without looking like a rug-pull Launch discount decided: 10% for the first 7 days, or none. Announced as “launch week only” and ends when you said No discount beyond launch week in the first 30 days; never two discounts inside 30 days Regional pricing on storefront defaults, reviewed once. Do not hand-tune 40 regions solo No fake original price, no countdown pressure copy, no “price rising soon” unless it is scheduled, public and true Demo decision made by T-60: yes or no. A bad demo costs more than no demo Tax forms filled before launch, not after the first payout Runway assumes roughly a third gone to platform cut and tax, and 30–60 days from sale to bank Refunds: honour the storefront policy without argument. A refund is cheaper than a bad review All done in this section.
T-90 to T-30: build the evidence 0/13 T-90 Store page live at T-90, long-lead press at T-60. Wishlists start at zero; that is expected, not a failure, and it is why the page must not slip.
Release week checked for major festivals and storefront sale events. Move if it collides Store page live with trailer, 6 screenshots and full description Announced with a date window (“March”), not a day Wishlist baseline recorded the week the page goes live Trailer footage captured every sprint, not in one panic week Devlog every second Friday without exception. A devlog with a number in it outperforms one without Press list at 40 researched contacts T-60: personal pitches to written press and curators, one at a time T-60: demo decision made. If yes, it ships at gold quality or not at all T-60: subtitles and accessibility settings final Weekly tracker filled every Friday: wishlists, followers, devlog views, the delta and the one thing that drove it At 20% of your wishlist target at the halfway point, the plan is wrong, not slow. Change it Never buy followers or wishlists. It poisons every number you would use to decide All done in this section.
T-30 to T-7: lock it down 0/12 T-30 Announce the exact day at T-30. Nothing in the last 14 days is new work; T-14 onward is execution only.
Exact launch day announced. After this it does not move for anything short of a crash bug Video creators get keys: no embargo, no obligation 20 social posts queued, each rewritten per platform Store page re-read end to end against the real build; release date field now shows the day Support email, refund policy and a public bug form exist and are tested T-14: gold build archived with a version tag and checksum, offsite T-14: feature freeze. Blockers only, and each fix re-runs the three-machine clean test T-14: launch-day posts written and scheduled, including the “it is out” post T-14: one short follow-up to every unanswered pitch, then stop Licences and credits final; every asset in the credits has a licence you can show Slip rule: if the build cannot pass gold by T-14, move the date before T-30, publicly, once. Never silently, never twice Sleep schedule fixed. A tired dev on launch day is the biggest risk left All done in this section.
Launch week: rehearsal and freeze 0/12 T-7 Rehearse launch day at T-7 on a clean machine. Everything below is execution; every decision was made already.
Exact launch build uploaded to the storefront’s private or unlisted channel Downloaded as a customer would, on a clean machine; first 20 minutes played Build launches with no console window, no debug overlay, no placeholder text; version number visible on the title screen Runs on the minimum-spec machine at the target frame rate; controls work on keyboard and on a controller Quit from any screen, restart, continue: save intact Zero blocker bugs, zero majors on the critical path Price, regional pricing, store page, trailer playback and refund copy verified Key redemption flow tested with one key you generated yourself FAQ answers written, ready to paste on launch day Launch posts scheduled, assets attached, links tested Phone charged, laptop charged, food in the house The night before: sleep. Nothing you fix at 02:00 will be a net gain All done in this section.
Launch day runbook 0/16 T-0 You are on support duty, not marketing duty. The posts are already written. Nothing on this list is a decision.
T-2 h: final download test on the clean machine, first 20 minutes played T-1 h: price, discount, date and platform list confirmed on the live page T-0: flip to public. Verify the buy button in a private browser window T+5 min: buy a copy yourself. Confirm receipt, download, install, launch T+15 min: post the launch posts, one per platform T+30 min: email everyone who replied to a pitch: it is live, here is the link T+1 h: first support sweep: email, store discussions, social replies T+2 h: first crash-report check. Triage only; do not fix yet T+4 h: second sweep. Pin a “known issues” note if two people hit the same thing T+6 h: eat. Walk outside. The numbers will still be there T+8 h: third sweep. Decide: day-one patch tonight or tomorrow morning T+10 h: record the numbers. Stop for the day Patch today only for crash, save corruption, progression block or a wrong price/store claim; never after 22:00 A review calls out a real flaw: thank them, log it, do not argue. Never reply to a review while annoyed Do not announce a roadmap, a sequel, a sale or a date for anything. Do not ask anyone for a positive review Sales far below forecast: change nothing today. Decisions on bad data at hour four are always wrong All done in this section.
First 30 days after 0/11 T+1 Answer everything for 72 hours, ship patch 1 by T+7, hold the price through the first quarter.
Support sweep morning and evening; every report triaged into one tracker Every review and message answered for the first 72 hours T+1: one honest “day one” note: thank you, known issues, what is next Every patch re-runs the clean-machine test and gets a one-line public changelog the same day T+7: patch 1 ships with the top 3 reported issues, no scope creep T+7: thank every creator who covered it, no ask attached Actual numbers logged against the forecast: wishlists, sales, refund rate, median playtime Launch post-mortem written while it still hurts: what moved the numbers, what did not, what you would cut next time Refund rate over 10% means the page is overpromising. Fix the page, not the price No discount beyond launch week; no roadmap promises until patch 1 is out and stable Reprice only when two are true: refunds over 10%, median playtime under half the advertised length, three months near-zero at full price. Then once, publicly, with a patch All done in this section.