Why does SEO take 3 months? All about Transition Rank

You’ll often hear an SEO consultant or expert tell you that “SEO takes 3 months”. But based on the speed of the Babbar crawler (3 to 6 billion pages per day) and its position in the cloudflare index of verified bots (between 18^thand 26^thdepending on the average crawl speed observed), a search engine like Google could take just one second to crawl all the pages of a 69,000 url site once (if it went at the speed of the world’s 18^th, which it does). So it’s not the time it takes to crawl the urls that’s the issue, but rather the processing that follows.

Table of Contents

What is Transition Rank?

Simplified definition:

Described in patent US8924380B1 by Ross Koningstein, Transition Rank aims to maintain the “integrity” of search results by controlling and limiting the impact of artificial actions designed to improve a page’s ranking.

Take the patent summary:

What it means:

Instead of moving a document from an initial classification to a new classification, the system applies a transition period during which a transition rank is assigned to the document, a rank which changes progressively by following a rank transition function instead of using the usual factors impacting position.

Assumed objective

The stated aim is to combat practices that Google will define as spam, and which are frequently used as part of SEO actions to modify the ranking of a site’s urls.

Here, Ross Koningstein specifically names actions that are considered spam:

Over-optimizing keywords:
- This is a sign that semantics is essential for the engine. Over-optimization is a broad term that essentially covers keyword stuffing (repeating a term several times in the hope of increasing the document’s relevance to that or a related term).
Invisible text & tiny text:
- If text is made invisible to users by its size or color, the rendering does not allow the user to read it. Let’s also include the practice of putting a white element or photo on top of text so that it never appears to the reader. The idea is like keyword stuffing in that the HTML rendering displays this text, but the actual rendering doesn’t allow it to be displayed to users. It’s text specifically designed to fool the engine.
Page redirects:
- Mr. Koningstein is no doubt referring to the practice of redirecting pages from an expired site (a site that no longer exists and whose domain name has been reserved by a third party) to other pages on a site that should thus benefit from the popularity of the old pages. This practice aside, page redirects are generally used to move a page from its old canonical address to its new canonical version. Here he clarifies that the abuse concerns page redirects that have nothing to do with the source and target.
Over-optimizing META tags:
- A little abuse of language to refer to the attributes of the meta tag, which are used directly or indirectly for SEO purposes. Let’s specifically mention here the description, which should serve to reassure and encourage users to click, and which is not used directly, but is frequently reworked. Here, the point is made about a technique of stuffing keywords that are irrelevant to the actual content of the page.
Link-based manipulation:
- Here it’s clear, we’re talking about link acquisition practices, either through the purchase or exchange of links or participation in a third party’s content with a link to their content in return. It specifically mentions link farms, but this can be extended to link sellers in general.

All SEO practices are potentially concerned by what this patent aims to achieve. Indeed, since the aim of SEO is to improve the positioning of a page according to a targeted query, the notion of page relevance may differ between the point of view of the search engine and its administrators, and that of the site publisher.

Subjects concerned

The presentation of this patent specifically mentions “Transition” and rank modification.

In fact, the only pages that are affected are those that are modified to change the positions they had before the update.

It could be argued that a page that doesn’t exist yet has an indefinite rank that is modified when the page is published, but that wouldn’t really make sense, since Google doesn’t know about the page and therefore doesn’t have a modification history for it. We know that for pages it does know, this history goes back over the last 20 modifications.

How and why does it work?

In concrete terms, this algorithm is based on two pillars: detection and transition phase.

Detection: identifying document modifications

The patent defines that the system detects document modifications by comparing

(Please note that 10% is an example).

Here, the patent holder explains that, broadly speaking, any change in ranking factor is an attempt to alter a document’s rank and therefore makes a page a candidate for the transition phase, and indeed it can be. He specifically explains that everything is based on a comparison between the old rank and the target rank.

The transition phase to identify a reaction

The process is as follows:

Find the old rank
Find target rank
Select the rank transition function (there are several)
Use this function to obtain a transition rank
Use this rank and observe the site during a detection phase

During this detection phase, the document’s rank will change and get worse. The objective is simple: to observe the site editor’s behavior in the face of this rank change. It’s really a behavioral study (reaction to frustration) to identify the culprits of abusive modifications. According to the patent, if a modification is observed during this period, the document is indeed subject to abusive modifications. (Claims 6 and 19).

In the event of multiple negative signals, the system may designate a document, a page, a site or even a domain as spam (one domain may host several sites).

How long does this transition phase last?

The patent examples speak of 20 days of decline to the pre-transition position and 70 days of transition to the final position. In total, the example speaks of 90 days of transition, but that’s just an example. We can’t be sure that the probationary period lasts 90 days, but it’s the best information currently available. He even explains that the transition can be observed according to “a time-based factor or a random factor”.

The patent also suggests that the algorithm depends on more than just 2 functions and can inject random variations to better prompt the editor to react.

Defining Transition Rank and getting out of it

It’s particularly difficult to detect that you’ve been hit by Transition Rank: after all, the algorithm defines that there are several transition functions (it speaks of “damped response transition function” and “initially inverse response transition function” (here illustrated in the graphics)) but leaves the door open for modifications. Consider the following:

The transition rank only concerns pages that have been reworked after having obtained a position.
The transition phase is divided into 2 parts: the first is a movement towards a transition rank for a given duration, and the second follows a modification of the rank according to the transition function selected.

Tracking positions can be done at regular intervals to see the evolution of pages undergoing modifications: this is essential when reworking content with an SEO objective.

We’ve just talked about detecting the transition rank, but what’s the way out?

It’s not possible to get out of the transition rank on your own, the only method is to wait 3 months by monitoring positions in search results to see they stabilize. Why 3 months? Because the 90 days in the example correspond to 3 months, and it’s reasonable to assume that Google doesn’t need that much time to identify whether a position change is legitimate or not. As a precaution, you can wait longer than 3 months, but above all you should wait for positions to stabilize.

Avoiding Transition Rank

So how do you avoid Transition Rank?

It seems that the patent only talks about the transition to follow up a suspicious rank modification. It’s the suspicion part that needs to be avoided when modifying content or a page’s link profile.

However, it’s natural for a page to be modified: the web changes, URLs are sometimes updated and modified to better correspond to a user’s need, sites migrate to change a design or adopt a new strategic approach. A text will improve if the editor decides that his or her opinion has changed, or if he or she has found clarifications to improve the provision of information.

We can therefore take 2 approaches:

A methodological approach which consists of ensuring that a modification can activate an internal validation process before publication, to avoid further modifications in the future for corrections or adjustments: this is a preventive measure, because a modified page is a page concerned.
An approach using more precise elements after a specific analysis of SERP requirements to identify an acceptable zone for optimizations. Based on the examples provided by the patent, a 10% improvement in the ranking criterion is sufficient to justify the transition, so we’d need to keep the threshold of acceptability for these criteria below 10% of what is proposed by the average content that manages to position itself correctly on this SERP.

In conclusion, this patent above all demonstrates Google’s desire to avoid spam that too obviously modifies its search results. We could translate this approach as a desire on Google’s part not to blindly trust every modification that takes place on the web, and to evaluate the site editor’s reaction before giving him credit for his modifications.

From the SEO point of view, this will mean difficulty in obtaining fast, reliable results in the short term, which can be a point of contention with decision-makers who need to see results to make the most of a sometimes-considerable investment in practices they don’t necessarily master.

In short, SEO can take up to 3 months for modifications to existing sites to optimize positions.

Sources: