Online Reputation and De-indexing

De-indexing is not deletion, and the difference decides the case

An index is a search engine's stored representation of the web: the pages it has crawled, parsed, and made eligible to be returned for a query. To de-index a page is to make it ineligible. The page is untouched - it still resolves, and it can still be copied.

Google states the distinction directly: "When you remove a result that shows your personal info from Google Search, it doesn't mean that the info is gone from the internet." And it describes the ordinary path in the other direction: "If the website's owner has removed the information, it'll eventually be removed from Google Search as part of our regular updating process."

That second sentence is the whole architecture. Search removal follows publisher removal, not the reverse. Most of what goes wrong here - the removal that came back, the judgment that produced no visible change, the vendor invoice describing work nobody can verify - traces to treating the index as the thing being changed when the publisher governs.

So the disputed question is one of reconstruction: what existed, at which addresses, in which countries, on which dates, and who could change it.

The removal menu, and what each entry actually does

Several processes exist, with different owners, scopes, and durations. Conflating them is how a case acquires an unenforceable expectation.

• The Search Console Removals tool, available only to a verified property owner. It hides a URL temporarily. Google states that "a successful request lasts only about six months" and that "blocking a URL does not prevent Google from crawling your page, only from showing it in Search results."
• The Remove Outdated Content tool, available to non-owners, for content the publisher has already removed or changed. It expedites what the crawl would do anyway, and the property owner sees these requests marked Approved, Denied, or Duplicate.
• Personal content removal, a policy process covering private, sensitive, or sexual content: personally identifiable information, doxxing, non-consensual intimate imagery, and content associated with exploitative removal practices.
• Results about you, a monitoring product rather than a removal one, reporting when a person's home address, phone number, or email address appears in results.
• Legal removal requests, where Google says it "will consider blocking, limiting, or removing access to" content that violates the law or someone's rights. Requesters supply specific URLs, explain what is violative and why, and receive a reference number.

What Google does not state in retrievable terms is what it requires when a court order is submitted - a certified copy, identification of the author, independent checking against court records. I would not describe those intake requirements with specificity, and neither should a report.

The six-month expiry, and why removal claims are testable

The most misunderstood fact in reputation work is that the owner-side removal is temporary. A vendor who "removed" a URL through the Removals tool bought six months. If the underlying page still returns a 200 status and is still crawlable, the result comes back, on a schedule nobody calendared.

Permanent removal requires a change at the publisher. Google is explicit about what qualifies: a 404 (Not Found) or 410 (Gone) response, password protection, or a noindex directive. It is equally explicit about what does not: "do not use robots.txt as a blocking mechanism." That matters, because robots.txt (a file at the root of a site telling crawlers which paths not to fetch) prevents crawling, and a page that cannot be crawled cannot be seen to carry a noindex directive. Blocking the crawler can preserve the listing it was meant to remove. Google's Removals documentation states all of this.

Every one of those conditions is checkable from outside, which makes "we had it removed" a testable claim rather than an assertion.

Global removal versus country-restricted delisting

Google's troubleshooter draws a distinction that decides what a client actually received. It separates legal grounds - intellectual property, court orders, defamation - from policy grounds such as malware and phishing, and states that "legal standards vary greatly by country/region." The consequence is that Google typically removes content globally when it violates Google's own policies, whereas content removed because it is unlawful is often restricted only in the relevant jurisdiction.

So a successful legal removal frequently produces a country-restricted delisting, not a global one. A client told "it's gone" who searches from abroad will find it.

This is verifiable, and it is one of the few places in reputation work where a documented protocol produces a genuinely probative exhibit. Repeat the query across country and language parameters, from clean profiles, recording the parameters with the results. Absence in one country and presence in another demonstrates scope rather than asserting it. Repeat at intervals, because scope changes.

Orders aimed at the search engine, and the European contrast

In the United States there is no general right to be forgotten, and the structural reasons are the First Amendment and the text of 47 U.S.C. section 230(c)(1): "No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider."

The California Supreme Court addressed enforcement in Hassell v. Bird (Cal. 2018), decided 2 July 2018. A plurality held that section 230 bars enforcement of a removal order against a non-party platform where the order's substance treats the platform as a publisher of third-party content: "where, as here, an Internet intermediary's relevant conduct in a defamation case goes no further than the mere act of publication ... section 230 prohibits this kind of directive." A plaintiff cannot, the plurality reasoned, "accomplish indirectly what Congress has clearly forbidden them to achieve directly" by leaving the platform out of the underlying suit. It is a plurality opinion and carries the weight a plurality carries.

Europe went the other way. The right to be forgotten derives from the Court of Justice's judgment in Case C-131/12, delivered 13 May 2014, and is codified in Article 17 of the GDPR. On territorial scope, the Grand Chamber held in Case C-507/17, decided 24 September 2019, that a search engine operator need not de-reference across all versions of its search engine worldwide, only across those corresponding to the EU Member States, with measures discouraging access from within the EU to non-EU versions. In the United Kingdom, NT1 and NT2 v Google LLC [2018] EWHC 799 (QB), decided 13 April 2018, dismissed one claimant's delisting claim and allowed the other's in part.

For a US litigator the takeaway is that de-indexing is achieved, if at all, through the publisher removing the content, a judgment the publisher honors, the search engine's own policy categories, or time and displacement. Not through an order directed at the search engine.

Court orders that turn out not to be court orders

This is not a hypothetical risk. Eugene Volokh's study Shenanigans (Internet Takedown Edition), 2021 Utah L. Rev. 237, examined roughly 700 apparently legitimate US libel case orders submitted to Google for takedown between 2012 and mid-October 2016. Over 200 showed obvious fraud or were highly suspicious: more than 90 apparent forged orders, roughly 30 with fabricated defendants, likely more than 30 fake-claim-of-authorship cases, and about 60 showing no genuine attempt to locate and serve the defendant.

The consequences are real. One individual produced eleven forged orders by modifying authentic documents in an image editor and received nine months in federal prison; a reputation-management operator paid $71,000 in sanctions over a fake-defendant scheme; a state attorney general obtained a $300,000 default judgment against another. Google typically checks submissions against available online court records and catches many forgeries - but some forged orders did produce actual de-indexing. The paper is published by UCLA School of Law.

Operationally, "there is a court order" is the beginning of the inquiry. The verification tasks are discrete: confirm the order exists on the court's own docket, confirm the named defendant is a real person reachable at the address of record, confirm service was attempted by means other than publication, and confirm the notarization against the state's notary registry.

What an expert can actually verify

Every step below is repeatable by an opposing expert, which is what makes it worth putting in a report.

1. HTTP status of the URL - 200, 301 or 308 and to where, 404, 410, 403, or 503. This alone distinguishes a real removal from a hidden one.
2. Robots directives - rules in robots.txt matching the path, any <meta name="robots"> tag, and the X-Robots-Tag response header, which carries the same directives at server level and is routinely missed.
3. Index status - whether a site: query returns the URL, and, where the property is controlled, what URL Inspection reports for coverageState, indexingState, googleCanonical, and lastCrawlTime.
4. Country variation - the same query across country and language parameters, to detect jurisdiction-limited delisting.
5. Mirrors and copies - exact-phrase searching on distinctive strings from the original text, to locate scraped republications that no removal touched.
6. Archive persistence - whether the Internet Archive and others still hold captures.
7. Expiry tracking - if a temporary removal was used, calendar the six-month point and re-check.

Two failure modes. Checking index status without HTTP status produces an opinion that something was removed when it was suppressed. And checking once: a removal that held in March and lapsed in September is a different fact than one that held.

Litigation itself changes the result set

The mechanism behind the Streisand effect is concrete and measurable in search terms. Litigation generates coverage. Coverage generates links and fresh content on high-authority domains. Search systems favor fresh, well-linked material for the queries at issue. The new coverage frequently outranks the item that prompted the suit, and the query consisting of a name plus "lawsuit" acquires results that did not exist before the filing.

The origin story supports the point with numbers. Suit was filed on 20 May 2003 against a photographer over an aerial image of a coastal property, seeking $50 million. Before the suit the download interface had been used for that image six times, two of them by the plaintiff's own attorneys. Afterward it drew roughly 500,000 views. The photographer prevailed on an anti-SLAPP motion and recovered $155,567.04 in fees.

The consequence for counsel is that this is measurable rather than anecdotal. Capture the results for the relevant queries before filing and at intervals afterward, and the change is documented rather than argued - and it cuts for whichever side the data favors, which is why it should be collected early.

Frequently Asked Questions

Published 2026-08-11