A valid comp is a game released in the last 3 years, in your exact subgenre, in your price band, made by a team roughly your size with roughly your marketing weight. Most pitch decks fail that test on slide one by listing Stardew Valley, Hades, and whatever sold 2 million copies last quarter, and every publisher reading the deck knows it.
Comps are the part of a publisher pitch where you either build credibility or spend it. Publishers see hundreds of decks a year, and the comp slide is where they check whether you understand your own market. The rest of this post covers the four selection criteria, the revenue math, and a worked example built on public data from three mining roguelites, ending with the table format that survives a publisher’s spot-check.
What makes a game a valid comp
Chris Zukowski’s working definition, from his pricing guide, is games “released in the last 3 years that are in the same genre and scale as yours” - and he specifically tells you to include the failures and the quiet mid-performers, not just the hits everyone talks about. I’d break “same genre and scale” into four checks:
| Criterion | Rule | Why it matters |
|---|---|---|
| Recency | Released within 3 years | Steam’s competitive density changes fast; a 2019 result is a different market |
| Subgenre | Shares your top 4-5 tags, not just the umbrella genre | “Roguelike” contains deckbuilders, shooters, and miners with wildly different revenue curves |
| Price band | Within roughly $5 of your planned price | Unit counts are not comparable across price tiers; a $6.99 game inflates units |
| Team and marketing weight | Similar team size, budget, and publisher status | A comp with a publisher’s ad spend behind it is measuring the publisher, not the game |
The recency rule is the one devs resist most, because the games that inspired their project are usually older than 3 years. GameDiscoverCo’s Simon Carless makes the case bluntly in Sales comps for your game: how not to do it: evergreen comps like Dead Cells (2018) launched into far less genre competition and then compounded years of updates and discounts, so their lifetime numbers tell you almost nothing about what a new entrant earns today. Your inspiration and your comps are different lists. Keep them separate in the deck.
The marketing-weight criterion needs the most honesty. If a comp launched with 189,000 wishlists built by a publisher’s creator outreach and paid social (as Dome Keeper did - more on that below), and you are self-funded with 8,000 wishlists, the comp still belongs in your table. It just belongs there labeled as the ceiling case, not the median.
Where to find comp candidates
You need about 10 candidates before filtering; expect to keep 3-5. The fastest sources:
- SteamDB tag pages. Open the tag view for your subgenre, sort by release date, and walk the last 3 years. This surfaces the mid-list and the failures that never got press, which are exactly the games Zukowski says to include.
- “More Like This” on your closest comp’s Steam page. Steam populates it from shared tags and purchase overlap, so it reflects how the algorithm itself clusters your market. If your game’s planned tags do not produce a sensible More Like This neighborhood, fix the tags first with the tag optimizer.
- Genre revenue data. Our Steam revenue by genre breakdown tells you what the median and top-quartile outcomes look like in your category before you pick individual games, which stops you from anchoring on an outlier.
For each candidate, record: release date, launch price, review count, review score, team size (check the credits or the studio’s site), and whether it had a publisher. A spreadsheet with those seven columns is the whole tool. Zukowski suggests adding a blunt quality column - “better than mine, worse than mine, equal to mine” - and it works because it forces a judgment you will otherwise avoid.
Include at least one comp that underperformed. A table with three winners reads as a sales document. A table with two winners and one honest failure reads as market research, and it sets up the “here is why we avoid that outcome” slide that publishers actually want to see.
Estimating what each comp actually earned
Almost no comp publishes its revenue, so you estimate from reviews. The Boxleiter method multiplies a game’s Steam review count by a sales multiplier: the working range is 20-60x, with ~30x the common middle for 2023+ releases. Multiply estimated units by price, then haircut for discounts, regional pricing, refunds, and Steam’s 30% cut. Our revenue calculator runs the whole chain from a review count in one pass, and the multiplier lookup table has genre-by-genre adjustments if your category skews high or low.
Two provenance rules when these numbers go in a pitch deck:
- Label every estimate as an estimate. Write “est. via review multiplier” in the table cell. Publishers use the same method internally (or subscribe to VG Insights and GameDiscoverCo, which build on similar signals); pretending an estimate is a fact tells them you don’t know the difference.
- Calibrate against announced figures when they exist. If a comp’s developer ever announced a sales milestone, check your estimate against it and say so. Agreement within 30-40% is normal and worth showing.
For pulling the raw inputs - review counts over time, price history, follower counts - the workflow in how to estimate Steam sales from SteamDB covers it step by step.
Worked example: comps for a mining base-defense roguelite
Say you are pitching a mining roguelite with base defense, planned at $12.99, made by a three-person team. Here is a real comp set, with review data I pulled from Steam’s public reviews API on July 2, 2026:
| Game | Released | Launch team/pub | Price | Reviews | Score |
|---|---|---|---|---|---|
| Dome Keeper | Sep 27, 2022 | 2-person dev + Raw Fury | $17.99 | 21,543 | 92% Very Positive |
| Wall World | Apr 5, 2023 | Small team (Alawar) | $6.99 | 12,384 | 91% Very Positive |
| Wall World 2 | Nov 11, 2025 | Small team (Alawar) | $11.99 | 1,854 | 74% Mostly Positive |
Dome Keeper slightly breaks the 3-year recency rule by launch date and clearly breaks the marketing-weight rule - it launched with 189,000 wishlists, a publisher running paid social and creator outreach, and a demo whose median playtime hit 1.5 hours against a ~25-minute indie average, per Zukowski’s launch breakdown. It stays in the table anyway, labeled as the genre-defining ceiling. That wishlist curve is worth showing because it is the “marketing weight” criterion in one picture:
Now the revenue math. At the ~30x multiplier, the review counts convert to estimated lifetime units:
Calibration check: Dome Keeper grossed $1 million at launch, per Game World Observer’s report, and bippinbits and Raw Fury announced it had passed 500,000 copies in their one-year anniversary press release in October 2023. An estimate of ~646,000 units almost three years later is consistent with that trajectory, which is exactly the kind of sanity check you note in the deck’s appendix.
Reading the set honestly:
- Dome Keeper is the ceiling: publisher, huge demo, genre-defining timing. Roughly $11.6M gross at full price, realistically well under that after heavy discounting (its price history shows regular 60-70% cuts).
- Wall World is the encouraging middle, but it sits $6 below your price band. Its ~372,000 units came at $6.99 with deep discounts to about a dollar; do not read its unit count as achievable at $12.99.
- Wall World 2 is the sobering recent data point: same studio, same subgenre, November 2025 release, ~56,000 estimated units at $11.99 with a 74% score. That is your realistic base case, worth around $670K gross at full price and meaningfully less after discounts and Steam’s cut.
So the pitch range this table supports is roughly “$400K-$1M gross over two years, with publisher-backed upside demonstrated by Dome Keeper.” That is a fundable claim. “The genre supports $10M, see Dome Keeper” is not.
Presenting comps in the pitch deck
The comp slide answers a question the publisher is already asking: what does this category pay, and where in the range does this team land? Bryan Finck’s pitching guide at SUPERJUMP puts the core rule well: don’t forecast your own sales; show what comparable recent releases did and let the publisher run their own model. Publishers trust their model more than yours, so your job is to hand them clean inputs.
Format that works:
- One table, 3-5 comps, with release date, price, review count, estimated units (labeled as estimates), and a one-line note per game on why it over- or under-performed.
- A stated range, not a point estimate, with your base case anchored to the least flattering comp.
- A sentence on what you will do differently from the underperformer. This turns the weak comp from a liability into evidence you have studied the market.
Carless offers a useful realism test for the set itself: three comps that all sold 1M+ units “is just plain unrealistic. But two and a 50,000-seller? Sure!” If every game in your table is a hit, you have curated inspiration, not comps. And if the math says you need an unlikely outcome to recoup, better to discover that in the spreadsheet than in a publisher meeting - the publisher vs self-publishing breakdown covers what recoup terms typically look like against these numbers.
The flattering-comp trap
Your game is not the next Stardew Valley. Neither is anyone else’s, statistically: more than 19,000 games shipped on Steam in 2025 alone, and citing one of the handful of outlier hits as a comp signals that you have anchored your plan on a lottery ticket.
The trap has three common forms, all fixable:
| Trap | What it signals | Fix |
|---|---|---|
| The outlier comp (Stardew, Vampire Survivors, Lethal Company) | You confused inspiration with market data | Move it to an “influences” slide; comps get recent mid-list games |
| The stale comp (a 2018-2020 hit) | You haven’t looked at the current market | Replace with the best performer from the last 3 years, even if smaller |
| The all-winners table | You filtered out every uncomfortable data point | Add the most recent underperformer in your subgenre and address it directly |
The deeper problem with flattering comps is that they poison your own planning before they ever reach a publisher. If your budget assumes Wall World’s units at Dome Keeper’s price, you have already made the decision the comp table exists to prevent.
Publishers check your comps. A scout can pull review counts and run the same multiplier in five minutes, and the tools they pay for do it automatically. A comp table that survives their spot-check builds more trust than any single slide in the deck.
Comp selection checklist
Run every candidate through this before it enters the deck:
- Released within the last 3 years? (Older: cut, or explicitly label as ceiling/legacy.)
- Shares your top 4-5 Steam tags, not just the umbrella genre?
- Launch price within about $5 of yours? (Outside the band: keep only with a written caveat.)
- Team size and publisher status noted next to the game, not hidden?
- At least one comp in the table underperformed?
- Every revenue figure labeled as an estimate, with the multiplier stated?
- Estimates cross-checked against any announced milestones?
- Your base case anchored to the weakest recent comp, not the strongest old one?
If all eight pass, run each comp’s review count through the revenue calculator for the deck’s numbers, then slot the table into the structure from our publisher pitch guide. And before any publisher sees the deck, make sure the store page they will immediately look up matches the market position your comps claim - the free Steam Page Analyzer will tell you whether your tags, capsule, and description actually read as the subgenre you are comping against.