Wishlist conversion by sub-genre. The 15% Week-1 conversion figure that circulated in 2020–2023 is no longer the median — it described a much smaller, more curated Steam catalog. Steam now adds roughly 25k new titles per year, wishlist quality has degraded sharply (free-add behavior, list bloat, less aggressive notification reach), and developer-shared data through 2025 puts realistic Week-1 conversion in the 1–3% range for most indies, with ~5% representing the high end of "very good." Anything above 5% is either a strong genre fit, an above-median review tier, or buzz the team actually earned pre-launch. The viral exceptions — co-op horror (Phasmophobia, Lethal Company), online co-op breakouts (PEAK, Mage Arena, R.E.P.O.) — still clear 12–25% W1, but these are the rare outliers that justify their own sub-genre tier, not a planning baseline. At the bottom, traditional visual novels, 2D platformers, and match-3 sit in the 1–2% band. The forecaster carries a separate Week-1 and Year-1 benchmark for each of 42 sub-genres across 12 parent families: pick the closest fit for your main, optionally blend in a secondary at 70/30 for hybrids. The expected review tier then applies a multiplier: Overwhelmingly Positive = 1.4×, Very Positive = 1.15×, Mostly Positive = 1.0×, Mixed = 0.55×, Negative = 0.25×. Review score gates discoverability at the algorithm level, so its effect compounds across the year. Sources: GameDiscoverCo (Simon Carless), Chris Zukowski indie post-mortems, public developer disclosures, and our own client launch data 2023–2026.

Wishlist age, velocity, and pre-orders. Not all wishlists are equal. Intent decays with age: the model weights cohorts at ~1.0 for adds under 3 months, 0.95 to 6 months, 0.80 to 12 months, 0.60 to 24 months, and 0.45 beyond — so a 50k base built slowly over four years can convert worse than 15k built in the last six months. Because the sub-genre benchmarks already assume a typical age mix (cohort weight ≈ 0.78, ~35% of the base added in the last six months), the correction is expressed as deviation from that typical base and is opt-in: with the toggle off the model is unchanged. With it on, two Steamworks numbers — adds in the last 6 months and page age — are enough to infer the spread. Freshness scales wishlist conversion (clamped 0.55×–1.28×). Recent velocity is a different mechanism: Steam's algorithm rewards momentum with Popular Upcoming placement, so it scales organic discovery (0.80×–1.25×), not conversion — keeping the two from double-counting. Pre-orders are treated as what they are: paid, guaranteed Week-1 sales at the pre-order price (Steam's standard incentive is 10%), removed from the convertible wishlist pool, and counted toward launch-day ranking so they seed organic. They are a known quantity, so they appear in the forecast but not in the tornado. Sources: Chris Zukowski's wishlist-cohort and Next-Fest velocity analyses; GameDiscoverCo on Popular Upcoming mechanics; ZR client Steamworks data 2023–2026.

Wishlist conversion — best-practice context. The Wishlist Conversion Check panel surfaces three things publishers and partners ask about constantly: the effective Week-1 rate the model is using, where it sits versus the sub-genre median, and the realistic 80% band around it (≈ ±50% — outliers blow through both ends). Three factors dominate everything else and are worth checking against your own situation: (1) wishlist freshness — adds older than ~12 months convert 30–50% worse than fresh demo/Next-Fest cohorts, so banked wishlists are not all equal in quality and a 50k count that grew slowly over four years can convert worse than a 15k count built in the last six; (2) review-tier inflection — the algorithm gates discoverability sharply at the "Very Positive" threshold, so a single tier change is the biggest conversion lever short of changing the game; (3) launch context stacking — a playable demo at launch reliably lifts conversion ~8%, Next Fest concentrates revenue into the launch month, and the two effects stack. The "Net per wishlist" readout on the panel is the single number most negotiation conversations land on — translate it for a publisher who isn't fluent in conversion percentages.

Regional revenue split. Year-1 indie revenue concentrates in NA + EU (~55–65% combined for most genres), with the remainder distributed across China, Asia ex-China (JP/KR/SEA), and a long tail (LATAM, RU/CIS, ANZ, MENA, India). Genre shifts this materially: cozy and narrative skew Western, soulslike + JRPG over-index in Asia, sim and racing are EU-heavy (DE/UK driving sim, FR + IT football sims), horror gets an outsized LATAM boost from the YouTube/TikTok creator economy. The shares applied are revenue shares — Steam's regional pricing is already baked in. Unit shares look different: emerging markets carry more players than dollars, which matters for community size, modding ecosystems, and audience for your next title, but should not be conflated with revenue weight in publisher conversations. The tilt selector models a deliberate localization or marketing push (a localized EU campaign, a CN partnership, a LATAM/RU push) that overrides the genre default. Sub-genre overrides handle the obvious outliers (soulslike, JRPG, co-op horror, racing sim, etc.) so the auto profile reflects sub-genre reality, not just parent family. Defaults are sourced from GameDiscoverCo regional analyses, Steam Hardware Survey language splits, Chris Zukowski sub-genre breakdowns, and published indie publisher post-mortems. Directional, not predictive — real launches deviate ±30–50% per region.

Pre-launch buzz. Buzz is the multiplier you've earned before the algorithm sees your launch — press attention, streamer reach, festival placements, viral demo moments. It compounds with both wishlist conversion and organic discovery (people who heard about you but didn't wishlist). Multipliers: None = 1.0×, Some = 1.15×, Building = 1.4×, Strong = 1.8×, Phenomenon = 2.5×. The top tier is the once-per-year breakout — Hades, Vampire Survivors, Balatro — and is rare for a reason: part craft, part timing, part luck.

Post-launch wishlists. Wishlists that arrive after launch — driven by reviews, content creator coverage, sale event visibility, and Steam's discoverability ripples — convert at roughly 70% of pre-launch wishlists. A healthy launch typically adds 50–200% of the launch-day count over the rest of Year 1, sometimes much more for sleeper hits.

Wishlist CSV import. The wishlist report from your Steamworks Sales & Activation Reports portal can be dropped straight onto the inputs panel. It is parsed entirely in your browser — the file is never uploaded, to this tool or anywhere else — which is the only appropriate handling for pre-launch wishlist data. From the daily rows the parser derives the three numbers the cohort model needs and that are easiest to get wrong by hand: outstanding wishlists (cumulative adds minus deletes and post-release activations), adds in the last six months, and store-page age from the earliest dated row. Valve has changed these exports' column names over the years and publishes several variants, so columns are detected by fuzzy header match rather than a fixed layout: DateLocal, Date or MonthCohort for dates; Adds or Wishlists for additions; deletes, gifts and activations where present. Comma and tab delimited files, quoted fields, thousands separators, BOM and Windows line endings all work. If the export has no deletes column the total is flagged as potentially high, and a cohort-style export is flagged because its dates are add-months rather than activity days.

Import from Steam. The import field at the top of the inputs panel pulls public store data for any Steam page: the game's user tags, current price, release status, whether a demo is published, and — for released titles — the review count and score. Tags are mapped to one of the 42 sub-genres by the same rank-weighted scorer the comparables pipeline uses, so the sub-genre you get from an import is the same one a comparable title would be filed under. Review scores are translated to the model's tiers on Steam's own published bands (95%+ Overwhelmingly Positive, 80%+ Very Positive, 70%+ Mostly Positive, 40%+ Mixed). Anything the import can't establish confidently is left alone and listed as skipped rather than guessed. Wishlist counts are deliberately not imported. Valve exposes them only inside the Steamworks partner dashboard — there is no API and no legitimate scrape — so they remain a manual input. Nothing about the import touches your Steamworks account, and no credentials are involved.

Live benchmarks. The calibration data behind this tool — the 42 sub-genre conversion benchmarks, the 52-week seasonality curve, the regional revenue splits, the review/buzz/demo multipliers, price elasticity, scenario bands, and the sale-uplift percentages — is read on page load from ZR's shared Supabase backend, the same database that powers the Industry Calendar. Steam sale dates come directly from the Calendar's event table, so when Valve publishes next year's schedule it is entered once and both tools pick it up. If the backend is unreachable the tool falls back to built-in defaults and says so in the stamp above; the model never blocks on the network. The stamp shows the data version so a forecast can be tied to the benchmark set it was computed against.

Comparable launches. The Reality Check panel is the one part of this tool that is observation rather than model. Every released title in ZR's Industry Calendar (1,400+ and growing) is matched to Steam by name in a nightly job, which pulls the public review count, review score, current price and Steam genres, and — where a Gamalytic key is configured — Gamalytic's copies and revenue estimates. Without Gamalytic, copies are estimated as review count × ~30, the review-to-sales multiplier Chris Zukowski and others have converged on for 2024+ titles (down from the older "Boxleiter number" of 50 as review rates rose); it is rough per title (±50%) and good in aggregate. Titles are mapped to the same 42 sub-genres the forecast uses via an editable tag-rule table, then filtered to your sub-genre and a ±40% price band, widening to the parent genre if fewer than five match. P25 / median / P75 copies and your base-case Y1 percentile are shown so the forecast can be sanity-checked against what actually happened to similar games. Each row links to Steam so a mismatch can be spotted. Sources: Steam store and review APIs (public), Gamalytic (optional), ZR Industry Calendar.

Launch-window seasonality. Calendar week of release matters more than most forecasts admit. The model carries a 52-entry seasonality curve calibrated from monthly launch-cohort data: post-holiday cold start (W1–4, ~-15%), February quiet zone (W6–10, neutral), April–May sweet spot (W14–22, +5%), Summer Game Fest crush + pre-Summer-Sale FOMO (W23–25, -10%), Summer Sale collision (W26–27, -15%), August trough (W31–34, -30% — the single softest window of the year, vacations and weather collapse PC engagement), September back-to-school recovery (W37–39, neutral-to-good), October sweet spot (W40–42, +10% — pre-AAA crush, Halloween horror tailwind), AAA crush window (W43–45, -15%), Black Friday / Autumn Sale (W47, -20%), and Winter Sale collision (W51–52, -30% — worst of the year for full-price launches). The multiplier applies at full strength to Week-1 revenue and at ~38% strength to Year-1 (since later months in the game's first year land in different calendar weeks and average out). This is separate from — and stacks on top of — your manual Year-1 blended seasonal discount, which models price discounting across the year, not launch timing. Sources: GameDiscoverCo monthly launch-cohort analyses, SteamDB concurrent-player aggregates, Chris Zukowski release-timing guidance, ZR client launch outcomes 2022–2026.

Launch context. A playable demo at launch lifts conversion by roughly 8%. Next Fest adds another smaller bump (~4%) and shifts more revenue into the launch month through concentrated visibility. Both are independent of the wishlist count: they affect how those wishlists convert, not how many you have.

Organic, non-wishlist sales. Wishlists are not the whole funnel. Steam discoverability, store browse, and sale events drive additional sales not attributable to wishlists. We add this as a percentage of wishlist sales (20% default, range typically 5–50% depending on visibility). Buzz lifts this further at runtime.

Effective ARPU. List price × (1 − regional haircut) × (1 − blended discount). Steam's regional pricing pulls blended revenue per copy down by 15–25% vs straight USD math. For the year-1 number, the blended discount is a weighted average of your launch discount (applied to the launch month, which carries ~38% of Year-1 revenue) and your seasonal blend (applied to the remaining ~62%). This is why both discounts matter — and why a launch discount that's too aggressive can quietly chew into the year.

Regional pricing in detail. Steam recommends prices per region based on local purchasing power. A $19.99 game might be ~$14 in Brazil, ~$12 in Türkiye, ~$10 in India, ~$8 in some SEA markets. You can override the recommendations, but doing so usually costs more in lost conversion than it recovers in price. The default "haircut" assumes you follow Steam's recommendations — globally-appealing games sit at 20–30% haircut, US/EU-focused titles at 5–15%.

Refunds. Steam's 2-hour refund policy yields 5–10% refund rates for most indies, higher for short games. Refunds are deducted from gross copies before revenue is computed.

Steam's cut. 30% on all revenue up to $10M, 25% to $50M, 20% above. Only the 30% tier is relevant for nearly all indies, so we apply a flat 30%.

Price elasticity. Conversion is not price-independent — raising the sticker price lowers the share of wishlisters who buy, lowering it raises that share. Sub-genre benchmarks are calibrated at a $19.99 reference; conversion is then scaled by (19.99 / price)0.40. In practice: $24.99 costs ~8.5% of conversion, $29.99 costs ~15%, $14.99 gains ~12%. That tracks the 10–20%-per-$5-step range Chris Zukowski reports for indie titles in the $10–30 band, and the sub-linear exponent keeps a $4.99 price from implying absurd conversion. The effect is what makes the "Price" driver in the tornado an honest trade-off rather than free money: revenue per copy rises, copies fall, and the net depends on where you sit on the curve. Anyone keeping the default price sees no change from this term. Sources: HowToMarketAGame pricing analyses; GameDiscoverCo price-tier conversion work; ZR client launch data 2023–2026.

Sensitivity bands. The downside scenario assumes wishlists at 55% of input, review tier one notch lower, and buzz one notch lower. Upside assumes wishlists at 180% and review and buzz tiers one notch higher. The band is deliberately asymmetric: indie outcome distributions are right-skewed, with a fat left tail of under-performers and a thin right tail of breakouts, so the realistic 80% range runs roughly ×0.55 to ×1.8 around the base — the same band shown in the Wishlist Conversion Check panel, so the two readouts never disagree. Treat it as a planning range, not a confidence interval.

Tornado chart. Each driver is moved through a defensible range (wishlists ×0.55 / ×1.8, price ±$5, review and buzz tiers ±1, refund/organic/seasonal/regional ±5–10pp, launch discount ±5pp, launch context toggled) while every other input is held at base. Sorted by total swing. Anything sitting at the top is what to focus on first — both in production and in deal negotiation.

Time distribution. Year-1 revenue concentrates heavily in launch month (~38% cumulative) with major lifts at the first Steam-wide sale events (typically months 3 and 6). Year-2 adds roughly 35% on top of year-1 total in a long tail.

52-week trajectory. The weekly view splits Year-1 into 52 buckets calibrated so each block matches the monthly model exactly: M1-M8 span 4 weeks each (W1-32, where ~84% of Y1 revenue lives), M9-M12 span 5 weeks each (W33-52, reflecting the real 4.33 weeks/month). Within each block the curve front-loads — Week 1 alone captures ~20% of Year-1 revenue. Sale events add incremental revenue as a percentage of total Y1 (Lunar +1.5%, Spring +2%, Summer +6% over 2 weeks, Autumn +2.5%, Winter +6% over 2 weeks) — calibrated from GameDiscoverCo sale-event tracking and HowToMarketAGame case studies, then scaled by remaining wishlist tail (a Summer Sale 4 weeks after launch hits at ~98% effectiveness; the same sale 40 weeks in only ~55%). DLC drops follow the same additive model: ~4.5% of Y1 spread across 4 weeks (45% / 28% / 17% / 10% of the budget), scaled by tail effectiveness. Stacking multiple events on the same week is additive. The shift to percent-of-Y1 (rather than multiplier-on-baseline) avoids the trap where early-game sale uplifts would over-inflate because baseline weeks 1-4 carry most of the year's revenue.

Deal mechanics. Revenue share splits gross post-Steam revenue. Recoup-first sends 100% to the publisher until a defined amount is recouped, then splits. MG against royalty pays the developer an upfront minimum guarantee, which the publisher then recoups before paying further; the developer keeps the MG regardless. Two contract conventions exist and the MG card lets you pick: recoup from dev royalties (the default, and the standard "recoupable advance against royalties" you'll see in most term sheets — the dev's royalty share accrues against the MG, so dev cash resumes only once cumulative royalties exceed it, i.e. at post-Steam revenue = MG ÷ royalty%), or recoup from gross (publisher takes the MG off 100% of post-Steam before any split, then the dev gets their royalty of the remainder — more dev-favourable and less common). The difference is material: on a $75k MG at 60% royalty and $200k post-Steam Y1, the dev nets $120k under the standard convention versus $150k under gross. Always confirm which one the paper says before comparing offers. Marketing fee takes a fixed fee off the top before rev share kicks in.

These are planning estimates. Real launches vary on press, virality, genre saturation, timing, and luck. A good consultant won't pretend otherwise.