The launch report is $299. The first 25 founder slots are at $79 so I can stress-test the product with early users and respond directly to every cone-miss. Same report, same guarantee.
Simple pricing.
Free single-game forecast on every release. The full launch report is a one-time $299 per launch — not a subscription. Right now, the first 25 founder customers get access at $79 (24 slots remaining).
- Calibrated P10–P90 revenue cone
- Boxleiter cross-check from your wishlist count
- Top 3 reasons for the forecast (SHAP feature attributions)
- 5 named comp-set launches with revenue grades
- Out-of-distribution flag when comp similarity is low
- Open methodology + per-genre coverage stats
- Everything in Free, plus:
- Marketing-lever causal estimates with 95% CIs (demos, festivals, price changes, trailers)
- Explicit “not significant” tagging when the CI overlaps zero
- Total Lift Attribution: upload your Steamworks “Wishlists by day” CSV; recover paid-campaign wishlists Steam’s UTM dashboard under-reports by ≈75%
- Comp-set rationale (full retrieval explainer, not just the names)
- Re-runnable through launch. Wishlists grow → cone narrows → you re-run, no extra charge
- Authenticated dashboard with your launch’s history
P10–P90 Performance Guarantee: if your 30-day net revenue lands outside our forecasted cone, full refund. Details
Why $299 one-time and not $19/mo
Indie launches are punctuated events, not ongoing subscriptions. You need rigorous calibration during the 90 days leading up to launch. After launch, the question changes — you don’t need a forecaster anymore, you need actuals.
We could charge $19–$50/mo and call it a day. Most analytics SaaS does. But it would mean buying you a tool you’ll forget to cancel after launch — or rent the tool twice if you’re launching one game per year.
$299 once per launch matches the actual usage shape. Re-run the forecast as your wishlists grow; the dashboard stays available through your launch window without recurring charges. Ship a second game later, buy another report. Or don’t. We won’t auto-bill you.
If our cone misses, you don’t pay.
If your game’s first 30-day net revenue lands outside our forecasted P10–P90 cone, we refund the launch report in full.
Our model targets 80% empirical coverage on a held-out launch corpus. Held-out coverage on n=1,560 historical launches is 82% — meaning ~82 of 100 launches land inside their predicted cone. If yours doesn’t, the model was wrong about your launch and you shouldn’t pay for it.
How to claim
30+ days post-launch, email [email protected] with a screenshot of your Steamworks 30-day net revenue (or your seller portal’s equivalent). We compare against the P10 and P90 figures in your delivered report. If actual revenue is outside [P10, P90], full refund within 14 days. No questions asked, no hard feelings — the model owes you the money back when it’s wrong.
Time-based refund policy
In addition to the performance guarantee above — whichever applies in your favor.
14-day full refund if you haven’t uploaded Steamworks data to Total Lift Attribution yet.
50% refund within 30 days if you’ve used Total Lift Attribution but the cone hasn’t shipped your launch yet.
After 30 days or after launch, refunds are case-by-case — email [email protected] and explain.
See the full terms for the legal version.
Frequently asked questions
Is the free forecast the same model as the paid report?
Yes. Same calibrated cone, same comp-set, same coverage gates. The paid report adds the causal lever effects (marketing what-if estimates with confidence intervals) + Total Lift Attribution + the dashboard so you can re-run and track your launch over time.
Can I buy the report before my game has a Steam page?
You need a published Steam app ID. The free forecast lets you sanity-check whether we have your game in our corpus before purchase. If your game is unlisted or in early development, wait until your Steam page is live.
How accurate is the model?
Active model passes the size-stratified coverage gate at 82% empirical coverage on a 1,560-launch held-out corpus. Full methodology + per-genre coverage table is open. We also publish a wrong-predictions log for the cases where any pre-launch model breaks down. As new games launch post-training-cutoff, our daily cohort-verification pipeline tracks ongoing performance independently of who calls the endpoint.
What if my game is unlike anything in your data?
The forecast page shows an out-of-distribution flag when comp-set similarity falls below threshold. In that case, treat the cone as illustrative rather than budget-grade, and email me before purchase if you want a sanity check.
Do you offer team or studio licenses?
The single-launch report is per-launch, not per-seat. Team members can access the report through the same authenticated dashboard. If you’re a publisher with multiple titles, buy one report per launch — volume discount available, email for details.
Can I get a sample report before purchase?
The free forecast IS the sample — same model, same outputs, just without the causal lever effects + attribution. Run it on your own appid. If the cone makes sense, the paid report adds the planning tools on top of the same foundation.
Try a forecast first: free single-game forecast · or read the methodology.