Most advice about getting found online is asserted rather than measured. These are the measurements we run for our own work, published with the method, the limits and the raw counts, so you can check them or repeat them rather than take our word for it.
Whether “go long-tail” actually moves a small retailer away from the marketplaces — 30 product categories × 3 levels of term specificity, with seller type and page type crossed as independent axes.
19.3% marketplace share on a broad category term
48.3% on a long-tail term — the opposite of the prediction
Whether a local business ranks on a page it owns or a profile it rents on somebody else's platform — 30 trades × 9 towns across three city-size tiers, with subject scope and domain control crossed as independent axes.
38.6% of large-metro results are on the business’s own site
21.3% in the smallest towns — the opposite of the prediction
Who actually publishes the round-up articles that rank for "best X software" — and how rarely a product page appears at all. 30 categories × 3 query forms, with publisher and page type crossed as independent axes.
59% of ranking round-ups are published by software vendors
What KIND of page ranks: a homepage, or a page about the specific job searched for. 30 trades × 3 query forms, with the structural axis derived from the URL rather than judged.
126 ranking business URLs
25.6% are homepages when the query names the trade
What the top three results actually contain when somebody searches the price of a service, across 80 of them — and how far that differs between what a business buys and what a household buys.
What goes here, and what does not. A study on this page is backed by a measurement
someone else could repeat, and its figures are generated from the recorded data rather than typed — a test
fails the build if any number on a study page cannot be traced back to its dataset. Written pieces and
opinion live on the blog, which says plainly that it is written.
Take the numbers. Every figure behind every study is published as JSON under
CC BY 4.0, with the method and the
limits attached to it — so a figure cannot travel without the caveat that qualifies it. Chart it, re-cut it,
or check our arithmetic: the data index. New studies arrive by
RSS or
JSON Feed.