How RevScope estimates revenue
No private data, no tracking pixels. Everything comes from what any visitor's browser can see, plus published averages. That makes the numbers ranges, not facts.
1. Traffic from rank
We look the domain up in the Tranco top-1M list, a research ranking that blends several popularity sources. Rank is converted to monthly visits on a log-log curve. The top of the curve is set from public figures for giant sites; the rest is fitted to 33 publishers that publish their own traffic in media kits:
Rank is a noisy signal. Sites with lots of links (tech, developer blogs) rank higher than their traffic deserves, and recipe or home blogs that live on Google and Pinterest rank lower. So we apply a ×/÷ 2.8 band. Sites outside the top million aren't necessarily tiny: we use their ad network as a size clue, because Mediavine and Raptive only accept sites above a traffic minimum. A premium-network site is assumed to get ~55K visits a month; a site with no ads ~5K.
Chrome popularity data comes first. Google's Chrome UX Report places every site in a popularity bucket (top 1K, 5K, 10K, 50K, 100K, 500K, 1M) based on real Chrome page loads. That bucket sets the traffic range; Tranco only places the site inside it. A site that isn't in Chrome's top million is capped at ~250K visits a month, however well it ranks in Tranco. This corrects Tranco's biggest error: link-heavy sites that look far bigger than they are.
2. What the site sells
We read the homepage, one article page (found through a homepage link, the RSS feed or the sitemap; ads often load only on articles) and /ads.txt, including its comments and MANAGERDOMAIN line, and look for fingerprints:
- Ad networks — Mediavine, Raptive, Ezoic, AdSense, Adsterra and others. The network sets how much of the niche's top rate the site earns.
- Affiliate links — Amazon Associates, Skimlinks, Impact, CJ, Awin, Rakuten.
- Stores — Shopify, WooCommerce, BigCommerce, Magento.
- Subscriptions — Stripe, Paddle, Piano paywalls, pricing pages.
The niche (finance, food, tech…) comes from keywords in the title, description and page text.
3. Rates
Ad setup multiplier on the premium rate:
Stores: 1–2.5% conversion × $45–110 order value (gross sales, not profit). Subscriptions: 0.05–0.4% of visits paying $12–60 a month (new MRR only). Native widgets: $0.50–1.50 per 1,000 pageviews.
Audience country matters as much as niche. Ads shown to US readers earn several times more than the same ads in India or Brazil. We set the audience from the site's country domain (.de, .in…) or page language. For English sites on generic endings (.com, .net, .io) we measure it from Chrome's per-country popularity lists for 49 countries. Rates relative to the US: UK/Canada/Australia 100%, English-mixed 90%, Western Europe 45%, Japan/Korea 50%, Eastern Europe/Latin America/Turkey/Middle East 18%, South & Southeast Asia/Africa 8%. You can change the audience under Assumptions.
4. The range
Traffic and rate uncertainty are combined (log-normal), so the low–high band is wide on purpose. "Medium" confidence is the best we ever claim. Large brands (roughly the top 700) earn mostly from things a webpage can't show — apps, enterprise deals, offline sales — so check their public filings instead.
5. Value
Small online businesses usually sell for a multiple of monthly profit: content sites 30–45×, stores 25–40×, SaaS 36–60×. We assume 80%, 15% and 50% profit margins respectively.
6. How accurate is it?
We test the model against sites that publish their own numbers (bench/ in the source). On 33 publishers' self-reported traffic, the median estimate is off by about 1.5×; 67% land within 2× and 82% within 3× (Tranco alone: 2×, 48% and 76%). Few benchmark sites are in Chrome's top 10K, so estimates for the very largest sites rest mostly on published figures for giant sites rather than this test. Given the real pageviews, our ad-earnings rates matched one Raptive travel blog almost exactly and came in 40% low for a top-earning Raptive food blog. Self-reported numbers can be inflated, and sites with media kits tend to be bigger than average, so treat this as a rough check rather than proof.