Gamalytic vs VG Insights: Which Steam Estimates to Trust in 2026

Buy Gamalytic if you want a Steam number you can defend, because it is the only one of the two that publishes its methodology, its error margin and its price in plain text. Buy Video Game Insights if you need PlayStation and Xbox alongside Steam, because Gamalytic does not cover console at all. Those two sentences decide most cases.

The rest of this post is the working behind them: where each product’s numbers come from, what each claims about its own accuracy, what each costs and how you buy it, and the specific reasons the two will hand you different figures for the same game without either being wrong. We checked everything against the vendors’ own pages on 26 July 2026 and flag every point where a page would not load.

Two caveats on that table before you use it. Gamalytic’s live pages were behind a bot check returning HTTP 429 to plain requests on 26 July 2026, so its prices come from the most recent archived copy of its own pricing page, dated 16 December 2025. VG Insights runs as a client-rendered application whose pricing and methodology routes return nothing but “Please enable JavaScript to continue using this application” to a plain request, and the Internet Archive’s 2026 captures contain the same empty shell.

Gamalytic states its method on its about page in four bullets, quoted here in full because the specificity is the point:

1. 01“Tracking the Steam top seller rank to estimate revenue and sales.”
2. 02“Using the number of concurrent players to estimate game’s playerbase”
3. 03“An advanced version of the boxleiter method. Regression analysis based on year of release, review score, average playtimes and similar.”
4. 04“Using any publicly disclosed data of actual units sold to further adjust the estimates”

That is a blend, not a single signal. Top seller rank captures revenue rate at a moment in time. Concurrent players capture the active playerbase. The reviews multiplier captures lifetime scale. Publicly disclosed sales figures act as the calibration set. Each of the three signals fails in a different situation, which is the whole argument for combining them: a game with an unusual review rate is still ranked on the seller charts, and a game with a strange launch pattern still has a concurrent player curve.

Video Game Insights links a “Methodology & Definitions” page from its help centre, which is exactly the right thing for a vendor to publish. We could not read it. The application renders entirely client-side, and neither a plain request nor an archived capture from 2026 returned any text from it. We are not going to restate VGI’s methodology from secondhand summaries, because that is precisely the kind of claim that goes stale and gets repeated forever. If you take a demo call, ask for the methodology document and read it before you sign.

What we can verify about VGI on 26 July 2026: Sensor Tower’s product page says it “covers over 140,000 PC and Console games” across “Steam, Xbox and Playstation”, reporting “game purchases and units sold, revenue”, plus DAU, price, cumulative revenue, wishlists and reviews. The application’s own metadata claims “50+ datapoints per game for over 150,000 games”.

Gamalytic publishes a number:

For individual games, our tests have shown that 77% of estimates were within 30% margin of error, while 98% were within 50% error margin. At an aggregate level, estimates were 99% accurate

It is a self-graded exam, and we would say the same about any vendor that published one. But it is a specific, falsifiable, self-graded exam, which is more than the rest of this market offers. It also comes with a list of admitted failure modes that most vendors would leave out: smaller games are less accurate, free-to-play games are less accurate, revenue estimates ignore external sources and may mishandle Steam bundles, and there may simply be bugs. Their words: “Always sanity check any estimates.”

We found no readable accuracy claim from VG Insights and are not going to invent one.

Now the part that matters more than either number. Sit with what “77% within 30%” means in currency. Take a game the tool estimates at $500,000 lifetime gross. A 30% margin puts the true figure somewhere between $350,000 and $650,000, and that is the good case that covers roughly three quarters of games. The 50% band, covering 98% of games, spans $250,000 to $750,000. There is also a 2% tail where the estimate is worse than that and you have no way to tell from the outside which games are in it.

That is not a criticism of Gamalytic. It is the honest resolution of estimating private sales from public signals, and any vendor quoting you tighter numbers without publishing a study should be asked how they measured it. The practical consequence: these tools are for ranking, sizing and sanity-checking. They are not for building a cash flow model to two decimal places.

This is the cleanest difference between the two and the one most likely to make the decision for you.

Gamalytic is a Steam product. Its data model, its glossary, its publisher classes and its API are all built around the Steam catalogue. Its footer states plainly: “We are not affiliated with Valve or Steam in any way.”

VGI covers Steam plus Xbox and PlayStation. If you are mapping a genre that lives partly on console, modelling a multi-platform launch, or answering a publisher who wants to know what the same game does on PS5, no independent Steam tracker can answer that question. This is the strongest single argument for VGI and we would not try to argue around it.

If you are a Steam-only indie developer, console coverage is a feature you will pay for and never open.

Gamalytic, per the archived copy of its pricing page dated 16 December 2025:

• —Starter, $25/month. All filters on the analytics tools and game list, historical data within the last year, players by country for the top 10 countries, player overlaps for the top 50 items, wishlist estimates, visual asset change history from June 2024, sales and revenue estimates for pre-order games, and 2,500 API requests per day.
• —Professional, $75/month, or $810/year. Historical data back to 2015, country split up to 50 countries, player overlaps up to 500 games, detailed playtime distribution, regional sales and revenue history from June 2024, DAU and MAU from 2023, review sentiment, experimental DLC and IAP revenue estimates, full wishlist data including overlaps and library size distributions, and full API access.
• —A free tier underneath both, a 30-day money-back guarantee, and billing through Paddle.

You buy it by clicking a button.

Video Game Insights publishes no price. Sensor Tower’s product page routes to Request Demo and Contact Sales. For historical reference, the archived VG Insights pricing page from 1 April 2025, which already carried a “We’ve been acquired by Sensor Tower” banner, listed an indie plan at $20 per month billed annually or $30 per month billed monthly, and a business tier at “Custom pricing based on # of licenses and extra options”. We could not verify whether either survives.

You buy it by booking a call. If you are a small studio, ask about the indie tier by name, because it existed recently enough to be worth asking about.

Often neither. Most gaps between Steam estimation tools are definitional rather than numerical, and Gamalytic’s glossary is the clearest published statement of why:

• —Owners is the estimated number of Steam accounts holding the game. It is inflated by giveaways, bundles and free weekends.
• —Players counts owners who launched the game at least once, including key-activated copies.
• —Copies sold counts only purchases made directly on Steam, and excludes key-activated units entirely.

Consider a game that sold 200,000 copies on Steam, distributed 60,000 keys through a bundle, and gave 40,000 more away in a promotion. Owners is 300,000. Players might be 250,000. Copies sold is 200,000. Three tools reporting those three numbers would look like they disagree by 50%, and all three would be correct. Before you conclude a vendor is wrong, check which of the three it is reporting.

The genuine accuracy gaps show up in four places, and they hit every tool in this market:

1. 01Free-to-play. Revenue comes from in-game purchases that leave no public trace. Both vendors flag this.
2. 02Heavy key distribution. Retail keys, Humble bundles and storefront resellers move units that Steam’s public signals barely register.
3. 03Regional pricing. A game with most of its sales in low-price regions earns far less than a units count multiplied by the US price suggests. Gamalytic states its gross revenue figure accounts for discounts and regional prices.
4. 04Very small games. Under a few dozen reviews, every statistical method loses resolution at once.

Before you pay either vendor, work out whether a free estimate is good enough for the decision you are making. Ours is one of several.

Our Steam revenue calculator takes a Steam App ID or store URL, pulls the game’s live review counts and price, applies a genre-adjusted reviews-to-sales multiplier from the 20x to 60x range, subtracts Steam’s tiered revenue share and an average regional discount, and returns units, gross and net revenue with a confidence rating and a genre percentile. Free, no account, one game at a time.

We will state its limits as plainly as we stated theirs. It uses one signal, review count, where Gamalytic blends four. It has no API, no historical timelines, no wishlist estimates, no country splits and no console data. We publish no error study on it, which means Gamalytic beats us on exactly the dimension we praised it for. Use ours for the first pass and to sanity-check a paid tool. Use a paid tool when the answer has to survive scrutiny.

SteamDB is the other free tool worth having open, for the opposite reason: it publishes exact concurrent player counts and price history and refuses to publish estimates at all, pointing users to SteamSpy, PlayTracker, VG Insights and Gamalytic instead. Facts on one screen, estimates on another, is a healthy way to work.

• —Steam-only research, published price, need to cite an error margin: Gamalytic. The self-published study is the differentiator.
• —Console in scope, or bulk datasets for custom analysis: VG Insights. Nothing else on this list covers PlayStation and Xbox.
• —One competitor lookup this month: neither. Use our free revenue calculator and SteamDB.
• —Pitching a publisher or investor: buy one, cite the vendor and the date, and state the margin out loud. A number with a source and an error bar reads as competence.
• —You cannot decide: take Gamalytic’s 30-day money-back guarantee for a month of intensive research, and book the VGI demo in the same week so you can compare like for like.

For the wider tool set, see our Steam page analysis tools roundup, the four-way comparison in SteamDB vs SteamSpy vs VG Insights vs Gamalytic, and the maths behind every reviews-based estimate in the Boxleiter method explainer.

Is Gamalytic more accurate than VG Insights?

Unknown, and anyone claiming otherwise has not measured it. Gamalytic publishes a self-graded error study reporting 77% of individual-game estimates within a 30% margin and 98% within 50%. We found no readable accuracy claim from VG Insights on 26 July 2026. Publishing a study is evidence of transparency, not proof of superiority.

Does Gamalytic cover PlayStation or Xbox?

No. Gamalytic is a Steam-only product. Sensor Tower states VG Insights covers over 140,000 PC and console games across Steam, Xbox and PlayStation. If console is in scope, this alone decides it.

Why do Gamalytic and VG Insights show different numbers for the same game?

Usually because they are reporting different quantities. Owners includes bundles, gifts and free weekends. Players counts anyone who launched the game, including key activations. Copies sold counts only direct Steam purchases. A 50% gap can be entirely definitional. Check which metric each tool is showing before treating the difference as an error.

Do I need a paid tool, or is the free reviews multiplier enough?

For genre sizing, competitor shortlists and sanity checks, the multiplier is usually enough, and our free revenue calculator applies it in one step. Pay when you need revenue over time, wishlist estimates, country splits, review sentiment or console data, none of which can be derived from a public review count.