Most Steam store page traffic comes from inside Steam itself: the home page, the Discovery Queue, search, tag pages, and “More Like This” widgets. In every developer postmortem I have seen publish real numbers, external websites send under 15% of visits unless the game is mid-press-cycle, and for quiet games the report is dominated by low-intent rows like Direct Navigation and Bot Traffic. Valve shows you all of this in the Store & Steam Platform Traffic Breakdown inside Steamworks, but it never explains what the categories mean or what a healthy split looks like.
That second part is what this post covers: every category decoded, the few numbers Valve has actually published, and real splits from developer postmortems with years attached, because the mix has shifted hard. Plus UTM links, so your external pushes stop showing up as anonymous “External Website” rows.
Where the traffic report lives and what it counts
In Steamworks, open your app’s landing page, scroll to Marketing Tools & Data, click Marketing & Visibility, and open the “Store & Steam Platform Traffic Breakdown” tab. The official traffic reporting documentation defines three metrics:
- Impressions: every time your capsule or brand is displayed somewhere on Steam, whether or not anyone noticed it. A capsule scrolling past in a sale grid counts.
- Visits: unique loads of your store page, tagged with where the customer came from.
- Click-through rate: visits divided by impressions, per placement. Some placements (like friend activity notifications) show impressions but have no measurable click-through.
Two things the report does not do. It does not tell you which visits converted to wishlists or purchases per source - that data only exists for links you tag yourself with UTM parameters, covered below. And the official doc does not enumerate the category names; the taxonomy below is community-documented from what the report actually displays, which Valve last overhauled in a 2020 “Traffic Reports Update” that recategorized confusing buckets and split out bot traffic.
I checked both Steamworks marketing docs on July 2 and neither publishes a category glossary. Treat the next section as field notes, not official definitions.
Every Steamworks traffic category, decoded
| Category (as shown in the report) | What it actually is | Intent level |
|---|---|---|
| Home Page | Front page: featured capsules, clusters, recommendation shelves | Medium |
| Discovery Queue | The 12-game personalized queue players step through | Medium-high |
| Search Results / Search Suggestions | Typed searches in the store search bar, plus autocomplete clicks | High |
| Browse Search Results by Tag | Tag and category browse pages | Medium |
| Other Product Pages / More Like This | Widgets on other games’ pages pointing at yours | Medium |
| Steam Client - Library / Wishlist | Players clicking through from their own wishlist or library | Very high |
| Community Hub / Friend Activity | Community content, friends’ recent activity | Medium |
| External Website | Any site that passed a referrer: Reddit, YouTube, press, your site | Varies wildly |
| Direct Navigation | No referrer at all: bookmarks, pasted links, app-to-browser jumps, many Steam client views | Unknown |
| Other Pages | Everything Steam could not classify more specifically | Unknown |
| Bot Traffic | Scrapers: SteamDB, SteamSpy, VG Insights, and friends | Zero |
Three traps in that table:
- Direct Navigation is not “people who typed your URL.” Any visit that arrives without a referrer lands here, including links opened from Discord, many mobile apps, and chunks of Steam client navigation. It is a junk drawer, not a fan club.
- Bot Traffic is normal and means nothing. Every game with a public page gets scraped daily. Before Valve’s 2020 recategorization this traffic inflated Direct Navigation, Browse Search Results, and More Like This; now it gets its own row. Ignore it.
- External Website undercounts your marketing. Links with
rel="noreferrer", most chat apps, and browser privacy settings all strip the referrer and dump the visit into Direct Navigation. This is exactly why UTM tagging exists.
For how these placements are earned in the first place, the mechanics live in our Steam algorithm explainer: the store feeds visibility to games that generate wishlists, purchases, and playtime, not to games that merely collect page views.
What Valve has published about discovery queue and search shares
Valve publishes platform-wide traffic numbers rarely, so the honest list is short.
The biggest disclosure came after the 2014 Discovery Update. Valve reported that product page views across Steam rose 30% after the update shipped, that 75% of that increase came from the Discovery Queue, and that the queue was generating 16% of all product page views platform-wide, per Valve’s figures reported by PC Gamer. In a 2022 Steam Labs experiment announcement, Valve added that 115 million players had used the queue to view a combined 18 billion store pages since launch.
That is essentially the whole official record. Valve has never published a current platform-wide share for search, home page, or external traffic, and the 2014 queue figure is over a decade old. Everything more recent comes from developers publishing their own dashboards, which is the next section.
For the per-surface mechanics - how the queue picks its 12 games, and what search actually ranks on - see the discovery queue deep dive and the search ranking breakdown.
What real postmortems show: internal traffic dominates, but the mix moved
The cleanest early dataset is Lars Doucet’s Defender’s Quest traffic deep dive, which published the full source split for a quiet “vanilla” month in fall 2014, with no sale and no external push:
Read the bottom row first: with zero marketing running, 94% of visits came from inside Steam. Doucet also found that an external push (a Humble Bundle feature) lifted External Website to 14%. It lifted search too: autocomplete clicks jumped from 38% to 64% of search traffic, because people who hear about a game go type its name.
The mix has not stayed still. Patrick Seibert ran a controlled experiment for GameDiscoverCo, launching two games with zero marketing, one in 2018 and one in 2021, and published both traffic reports:
Same developer, same no-marketing approach, and the Discovery Queue’s share of referral traffic fell from 63% to 8%. The 2021 game’s traffic was spread across many smaller sources, with tag-based browse pages taking the largest share of pre-launch impressions. As release volume grew (Steam passed more than 19,000 releases in 2025 alone), the queue stopped being an automatic firehose for unmarketed games.
A third data point: Chris Zukowski surveyed dozens of developers’ traffic reports in a 2022 traffic study and found a sharp split by tier. Top-earning games showed diverse source mixes led by search and algorithmic widgets. Games below roughly 12,500 weekly impressions showed the same three rows over and over: Direct Navigation, Other Pages, Bot Traffic. He also noticed high search traffic tracked with streamer coverage - viewers watch a video, then go search the name.
All three datasets are single-developer or small-sample publications, not Valve statistics. The direction is consistent across them (internal traffic dominates; the queue’s free ride shrank; search follows external attention), but do not treat any specific percentage as a benchmark for your game.
The default traffic floor, and what “real” traffic looks like
Put the postmortems together and you get a useful diagnostic. Every Steam page, marketed or not, has a traffic floor: bots scraping it, unclassifiable direct hits, and stray “Other Pages” visits. That floor is what Zukowski’s low-tier games showed, and it can easily run to hundreds of “visits” a week that represent nobody.
Signs your page has escaped the floor:
- Discovery Queue appears as a top-five source. The queue only feeds games the recommender has reasons to serve, so its presence means wishlists and engagement are registering. Our discovery queue guide covers what triggers it.
- Search Results grows without a sale running. People searching your name is the strongest organic signal in the report.
- More Like This shows up. Other games’ pages are now advertising you, which the algorithm does only for games with tag-relevant traction.
If your report is still Direct Navigation / Other Pages / Bot Traffic in that order, the fix is not traffic analysis. It is getting wishlists moving - see how to get Steam wishlists - so Steam’s surfaces have a reason to pick you up.
Which traffic sources convert best
Steam’s breakdown report gives you visits per source but not wishlists or purchases per source, so conversion-by-source is only directly measurable for UTM-tagged links. What follows is community-documented pattern, consistent across the postmortems above and our own conversion benchmarks, not a Valve table:
| Source | Typical conversion quality | Why |
|---|---|---|
| Wishlist / Library clicks | Highest | Already sold; they are checking price or updates |
| Search by name | Very high | Arrived with intent from outside attention |
| External: press / video coverage | High | Pre-warmed by context about the game |
| Discovery Queue | Medium | Browsing mood, personalized, but cold |
| Tag / browse pages | Medium | Genre intent, zero knowledge of you |
| Home page clusters | Low-medium | Broadest, coldest audience |
| Direct Navigation / Other | Unmeasurable | Mixed bag with bot residue |
Two practical consequences. First, a spike of home-page or queue impressions with flat wishlists is normal, not a page failure; cold-source click-through runs far below warm sources, which is why our CTR benchmarks segment by placement. Second, the highest-converting sources (name search, wishlist returns) are downstream of external attention, so the traffic report understates marketing impact: your Reddit post shows up partly as “Search Results” a day later.
The median game converts store visits to wishlists in the low single digits, and roughly 12% of day-one wishlists convert to launch purchases - run your own numbers through the wishlist calculator to see what a given traffic mix is worth.
UTM tracking: make your external traffic legible
The External Website row tells you a referrer domain sent people, but not which post, video, or campaign. Valve’s fix is UTM analytics: append standard UTM parameters to any link pointing at your store page, and Steamworks reports visits, wishlists, purchases, and activations attributed to each parameter combination.
The rules that matter, straight from the doc:
- The link must include at least
utm_sourceorutm_campaignto be tracked. All five standard parameters work: source, medium, campaign, content, term. - Attribution window is 72 hours from click to conversion.
- Three visit counts appear: total, “trusted” (bot-filtered), and “tracked” (logged-in users whose cookie preferences allow it). Only tracked visits can attribute conversions.
- Cross-device drift is real and acknowledged by Valve: someone clicks your link on their phone, then buys in the desktop client later, and the conversion is lost. Treat UTM numbers as a floor, not a total.
A workable convention: utm_source for the platform (reddit, youtube, newsletter), utm_campaign for the beat (demo-launch, nextfest), utm_content for the specific post or video. Tag every link you control - your website, press kit, video descriptions, newsletter. Do not bother tagging links you expect people to copy around; stripped parameters just fall back to normal referrer counting.
UTM data is also the only clean way to measure whether store page changes move conversion for a specific audience, which is why it pairs with the methodology in our store page A/B testing guide: stable traffic source in, conversion delta out. Without source control, a capsule change and a traffic mix shift are indistinguishable.
Reading your own traffic report: a five-minute routine
Do this monthly, and any week where something spiked:
- Filter out Bot Traffic first. Then look at what is left. If the remainder is mostly Direct Navigation and Other Pages, you are at the traffic floor; work on wishlist drivers, not page tweaks.
- Check the Discovery Queue row against last month. Growing queue visits mean the recommender likes your trajectory. A sudden drop after launch week is the normal decay of launch visibility, not a penalty.
- Compare Search Results to your external calendar. Search spikes 24-72 hours after coverage confirm the coverage worked, even if External Website barely moved.
- Divide impressions by visits per placement. A placement with big impressions and terrible click-through is a capsule problem for that context - run your page through the free Steam Page Analyzer to find which asset is underperforming.
- Reconcile UTM conversions against the External Website row. Big gap means untagged links are circulating; find where and tag the origin.
- Log the split each month (a spreadsheet with one row per category is enough). The absolute numbers matter less than the shape changing over time, and Steamworks only makes trends obvious if you keep your own history.
The report will not tell you what to do next, but it reliably tells you which half of the problem you have: not enough people arriving, or the wrong people arriving. Everything else in your marketing plan hangs off that distinction.