Private search engines tested honestly: what actually changes when you switch

Every private search engine’s homepage says roughly the same thing: no tracking, no profiling, search with confidence. That’s marketing copy, not a specification. The actual differences between these tools come down to a handful of concrete mechanisms: what gets logged, where your query travels before it becomes a result, and whether the index is theirs or borrowed. Once you understand those mechanisms, you can evaluate any search engine’s privacy claims yourself instead of trusting a badge on a landing page.

What a search engine can see in the first place

When you type a query into any search engine, a few things travel with it whether you think about it or not: your IP address, your browser’s user agent and other fingerprintable details, any cookies the site has set on you, and the query text itself. If you click a result, the search engine often finds out which one, because many of them route clicks through a tracking redirect before sending you to the actual page.

Google’s business model depends on connecting that query to a persistent profile: your account if you’re logged in, or a long-lived identifier in a cookie if you’re not. That profile feeds ad targeting and gets tied to your other Google activity over time. This isn’t a hidden flaw, it’s the product. Search is free because the data has value elsewhere.

A private search engine is really just a search engine that has committed, in some verifiable way, to not building that profile. The word “private” is doing a lot of work there, and it means different things depending on which engine you’re looking at.

The index question: whose results are these, actually

This is the part most comparisons skip, and it matters more than people think.

Building a search index that covers a meaningful fraction of the web requires crawling billions of pages continuously. Very few organizations do this at scale: Google, Microsoft (Bing), and a small number of independent efforts like Brave Search’s index and Mojeek’s index. Everyone else licenses or blends results from one of those.

DuckDuckGo, for years, combined its own crawler (used for things like instant answers and DuckDuckGo bangs) with results sourced through a partnership with Microsoft’s Bing API, plus other sources. Startpage takes a different approach: it acts as a privacy-preserving proxy in front of Google’s own results, submitting your query to Google without your identifying information attached, then relaying the results back to you. Brave Search runs its own independent index, built to reduce reliance on Google or Bing entirely. Mojeek also runs its own independent crawler and index, making it one of the few genuinely independent options at any real scale. SearXNG (and public instances of it) is different again: it’s a metasearch aggregator that queries multiple engines on your behalf and strips identifying details before forwarding results to you.

Why this matters for privacy: if an engine is proxying or licensing results from Google or Bing, your query still reaches Google or Bing’s servers in some form, just (ideally) without your cookies or account tied to it. The privacy promise then rests entirely on how well that stripping is done and whether you trust the middle party’s claims about it. An independent index avoids that dependency, but independent indexes are harder to build and can lag behind on freshness or coverage for less common queries. Neither approach is automatically better. They’re different trade-offs, and “who actually sees my query” is the first honest question to ask about any tool that calls itself private.

What “no logging” actually claims to mean

“We don’t log your searches” is a specific technical claim, and it’s worth unpacking rather than accepting on faith. In practice it usually means: no IP address is stored alongside the query, no persistent tracking identifier is issued or read, and query history isn’t tied to an account (because there often isn’t one). Some engines back this up with published transparency reports or third-party audits of their infrastructure; others simply state it in a privacy policy.

The honest way to check this yourself, without needing to trust a claim outright, is to look at what’s actually happening in your browser. Open your browser’s network inspector (developer tools, Network tab) while using the search engine. Look for third-party requests firing off to ad networks or analytics domains, cookies being set that don’t seem necessary for the search to function, and whether clicking a result routes you through a tracking redirect domain or takes you straight to the destination URL. This takes a few minutes and tells you more than any headline claim does. It’s also the kind of check worth repeating occasionally, since a company’s practices can change even when its marketing page doesn’t.

Where the search engine choice stops mattering

This is the part that gets left out of most “best private search engine” roundups, and it’s the most important part for setting expectations correctly.

Switching your search engine changes who sees your search queries and what gets built from them. It does not change what your internet service provider can see about your traffic in general, unless you’re also using something like encrypted DNS or a VPN alongside it. It does not stop the websites you visit after clicking a result from tracking you once you’re there, that’s a separate problem handled by browser settings, tracker blockers, and cookie management. And it doesn’t affect anything else you do online: the accounts you’re logged into, the apps on your phone, or any other service that has its own relationship with your identity.

No single tool, search engine or otherwise, makes someone “private” as a blanket state. Privacy here is closer to a set of independent decisions layered against specific things you’re trying to keep from specific parties: your ISP, advertisers, a data broker, a nosy platform. A private search engine addresses one of those relationships, the one between you and whoever builds a profile from your queries. It’s a real and useful piece, but treating it as the whole picture leads to a false sense of security that a determined or well resourced observer wouldn’t actually run into.

A more useful question than “which one is best”

Instead of asking which private search engine is objectively the best, it helps to ask what you’re actually trying to avoid. If your concern is a company quietly building an advertising profile from years of your search history, an engine with no persistent identifiers and a genuine no-logging architecture addresses that directly. If your concern is result quality dropping off for niche or local queries, that’s more about index coverage, independent indexes sometimes lag the giants there, and it’s worth testing with your own real queries rather than trusting a review. If your concern is broader, covering your traffic patterns as a whole, a search engine change is one piece of a larger setup, not a replacement for it.

Testing these tools honestly means resisting the urge to declare a winner and move on. It means checking the network traffic yourself, reading what the privacy policy actually commits to rather than what the homepage implies, and being clear about which specific risk you’re addressing before you decide a switch has solved something. That’s slower than a ranked list, but it’s the only version of “tested” that actually holds up.

If you want more breakdowns like this, ones that explain the mechanism instead of just the verdict, you can find the rest of our coverage on privacy tools and threat modeling on the home page.