Updated 26 August 2026 · Oernoe Editorial Team
This URL used to be another “privacy-first versus the field” essay. It overlapped Oernoe vs Google vs DuckDuckGo. Two documents with one job is thin content. This replacement is a method: twenty queries, three engines, notes on what came back. It is not a claim that Oernoe has the bigger index. It does not.
The test was run in a current desktop browser, signed out, in August 2026. Results move. If you repeat it next month you should get a different first page in places. That is expected. The point is the pattern, not a screenshot we can frame.
How the twenty were chosen
Navigational: oernoe, duckduckgo settings, google adsense help. Local-ish: coffee near union square, dmv appointment. News-ish: supreme court decision today, hurricane track. Reference: python list append, http 304 meaning. People: a public person listing we already know is in Oernoe, plus a name that is only famous on the wider web. Medical: strep throat symptoms (we are looking at result quality, not asking Search to diagnose). Product: replace macbook charger. Long tail: how to set firefox default search engine.
That mix is deliberate. A privacy engine that only looks good on navigational queries for its own brand is not a search engine. A tiny index that loses on breaking news is not a scandal. It is physics. We wrote down both.
What we recorded
For each query we looked at the first five blue links, whether the engine offered its own instant answer, and whether the page we wanted was in that set. We did not score “privacy” on the result page. Privacy is a product rule described in Search queries and Google ads, not a ranking bonus.
Google won most news-ish and long-tail reference queries. That is not subtle. The index is larger and fresher. DuckDuckGo was closer to Google on many of those, which is what you would expect from a service that can draw on a large partner index. Oernoe was weakest there. If your job is “what happened in the last six hours,” start somewhere else. We would.
Oernoe held up on navigational queries for our own hosts, on some reference queries with stable documentation, and on people and company listings that exist because someone submitted them and we approved them. That last category is not a crawl miracle. It is a separate pile of documents. Treating it as proof that the web index is complete would be cheating.
Where Oernoe was the wrong first stop
Breaking news. Rare error messages. Local results that need a map layer we do not ship as a homepage product. Medical queries where the useful page is a hospital system that our crawler has not fetched recently. In those cases the honest move is a second engine. The companion article is When Oernoe Search is the wrong tool.
We also lost some product queries to retailer pages that Google surfaces with extra shopping structure. Oernoe returned ordinary links. That is acceptable. Inventing a shopping graph to “keep up” would be a different product, and likely a tracking one.
Where the listing pile helped
Queries for a person or company that had an approved Oernoe listing put that listing in a place a stranger could find. Google often had a knowledge panel from a larger graph. DuckDuckGo often had a Wikipedia-led cluster. Ours was smaller and slower to appear, because approval is slow on purpose. Spam names do not belong in an index just because someone submitted them.
If your name is in that pile and should not be, that is a takedown of an index object. It is not a privacy export. The process is in How a privacy request is handled.
What we will not conclude
We will not say Oernoe “beat” Google. We will not say DuckDuckGo is the same as Oernoe because both talk about privacy. We will not turn twenty queries into a marketing table with trophies. The useful conclusion is narrower: use Oernoe Search when you want links without a query file for ads, and keep another engine for freshness and coverage. That is how we use it ourselves.
Ranking here is supposed to follow the query and the page, not a dossier of previous clicks. The crawler write-up is How Oernoe Search works. If a later crawl pass changes these twenty, this page should be re-run or dated as stale. A test log that never ages is fiction.
Repeat the test yourself if you like. Write down the twenty queries first so you cannot change the set after you see the results. Use a signed-out window. Ignore ads and widgets on Google. Count only the first five ordinary links. If your notes disagree with ours, send them to hello@oernoe.com with the date. A correction that comes with a method is worth more than a complaint that we lost.