Methodology

Rankeez was created because Alexa Internet shut down its ranking service in 2022, leaving developers and site owners without a free, simple way to check a domain's global popularity. We rebuilt that with full transparency about how the numbers are produced — and unlike Alexa, we don't hide the formula.

Our data sources

No single ranking list is reliable on its own — each has blind spots and can be gamed. Link-based lists reward SEO manipulation, DNS-based lists pick up bot noise, and crawl-based lists lag behind real traffic shifts. So instead of trusting one signal, Rankeez aggregates independent, methodologically different data sources so that the weaknesses of one are corrected by the others:

The algorithm: weighted Reciprocal Rank Fusion

Rankeez's core ranking algorithm is a weighted variant of Reciprocal Rank Fusion (RRF), a rank-aggregation technique from information retrieval originally used to merge result sets from independent search engines into a single, more reliable ranking (it's the same family of technique used internally by modern hybrid search systems). The intuition: instead of trusting any single ranking, sum each source's inverse rank, weighted by how trustworthy/manipulation-resistant that source is:

score(domain) = 30 / Tranco_rank
              + 20 / Majestic_rank
              + 12 / DomCop_rank
              +  6 / Umbrella_rank
              +  1 / CommonCrawl_rank

A domain absent from a given source simply isn't scored on that term — the fused score only sums the sources that actually have data for that domain, so a missing signal contributes nothing rather than a placeholder value. Domains are then sorted by descending fused score and re-numbered from #1 to produce the public Rankeez rank.

Tranco carries the highest weight because it's purpose-built to resist gaming; CommonCrawl carries the lowest because its domain-level hyperlink graph is the broadest but least curated of our sources; Umbrella sits above it because raw DNS query volume includes automated and background traffic that doesn't reflect genuine human visits.

Before scoring, we also check each source against the domain's other sources: if one source's rank is wildly out of line with the rest (roughly, more than 50x off — e.g. #1 everywhere but #134 on a single list), that source is treated as unreliable for that domain specifically. It contributes nothing to that domain's fused score (same as a missing signal) and is hidden from the Source breakdown table on the domain's page, so one inconsistent data point can't distort either the rank or what's shown.

Finally, the published rank blends this month's freshly computed position with last month's published rank (70% weight on the fresh position, 30% on where it was last month), so a domain doesn't swing wildly from one update to the next purely from source noise — a real, sustained change in a domain's standing still shows up over a couple of monthly cycles, just without the single-month whiplash. Newly tracked domains get their fresh position immediately, with nothing to smooth against yet.

Update schedule

All sources are re-downloaded and the full ranking is recomputed every month, automatically. The homepage shows the date of the last successful update.

What we don't do

We don't track individual users, we don't inject ads into rank results, and we don't offer a paid tier to see "real" numbers hidden behind a paywall. The free API and website show the same data.

← Back to home · Read the FAQ →

Chat with Rankeez
Ask us anything about your ranking, the API, or the data sources. We usually reply by email.