How our metrics work

Authority Score

A 0–100 score for how strong a domain’s (or a page’s) backlink profile is. Links from sites that are themselves well linked count for more than links from weak ones, the same idea as Google’s original PageRank.

How it is calculated

DataForSEO scores every domain and page in its backlink index from 0 to 1,000. We divide that by ten and round it:

Nothing else is mixed in, so anyone with the raw data can reproduce the number. The scale is compressed: going from 60 to 70 takes far more links than going from 20 to 30.

Why it does not match Ahrefs DR, Semrush Authority Score or Moz DA

• Each tool crawls its own copy of the web. They find different links, at different times, and drop dead links on different schedules.
• Each tool uses its own formula. Semrush, for example, adds traffic and spam signals. Ours is links only.
• Our score is not Ahrefs Domain Rating. We never call it DR, and a site with DR 50 can score well above or below 50 here.

How to use it

Compare a site with its competitors and with its own history, all inside this tool. A gap of a few points between two sites matters less than which way each one is heading. When you import an Ahrefs or Semrush export, we keep their links but not their DR or Authority Score, because those are a different metric.

Keyword difficulty (KD)

A 0–100 estimate of how hard it is to reach Google’s top 10 for a keyword, provided by DataForSEO Labs. It is based mainly on how strong the backlink profiles of the pages already ranking there are. Higher means harder. Like Authority Score, it is built from a different link index than Ahrefs or Semrush KD, so the numbers differ between tools. Use it to rank keywords against each other, not as an exact threshold. Read the full guide to keyword difficulty.

Search volume

The average number of Google searches a month over the last 12 months, for the market (country and language) you picked. It comes from Google Ads data via DataForSEO. Google rounds volumes, groups close variants of a keyword, and reports with a delay of about a month. So small or new keywords can show 0 or “—” even when people search for them, and two tools can show different figures for the same keyword.

Traffic estimates (Site explorer, Traffic analytics)

Estimated monthly organic visits to a domain or page. They are calculated, not measured: for every keyword the site ranks for, we multiply the keyword’s search volume by the share of clicks its position usually gets, then add up the results. That means they miss branded and long-tail searches nobody tracks, and they lag when rankings change. We do not use clickstream data, so there are no direct, referral or social estimates for other sites. Treat them as a way to compare competitors. For your own site, connect Google Search Console and our analytics script to see real clicks and visits.

Rankings, visibility and traffic forecast

Positions are checked live in Google, Bing or Yahoo for the location, language and device you chose for each search engine, on your schedule. They are the organic position of your site’s best-ranking page. Positions beyond the checked depth count as not ranking.

The rank tracker’s summary numbers use one click-through curve, the share of clicks a position usually gets:

• Traffic forecast = the sum, across your tracked keywords, of search volume × the share of clicks for your position.
• Search visibility = that forecast as a percentage of what you would get ranking first for every keyword. 100% means position 1 everywhere.
• Share of voice = the same visibility calculation for you and each tracked competitor, over the same keywords, so the lines are directly comparable.
• Average position counts ranking keywords only. Keywords outside the checked depth are left out, not counted as 101.
• Clicks and impressions in the rank tracker are real Google Search Console data, when it is connected.

AI visibility

Each week, every tracked prompt is sent to four models through their APIs, with web search enabled: OpenAI GPT-5 mini, Google Gemini 2.5 Flash, Anthropic Claude Haiku 4.5 and Perplexity Sonar. For each answer we record whether your brand is mentioned, its position among the brands named, the sources cited and whether your domain is one of them.

• Mention rate = answers that mention your brand ÷ answers checked.
• Share of voice = your mentions as a share of all brand mentions in the answers. Each answer counts a brand once, however often it repeats it.
• Consumer apps such as ChatGPT can use other models and personalise answers, so what one person sees may differ.