I Read Google’s Head of Search Testifying Under Oath. Here’s What He Admitted.
In October 2023, the US Department of Justice put Google’s SVP of Search on the stand and made him explain the algorithm. The transcript is public. Almost nobody in Nigeria has read it. I did.
The case that changed how I think about SEO happened in October 2023.
The US Department of Justice had sued Google for monopolising the search market. To build their case, they deposed Pandu Nayak, Google’s Senior Vice President of Search. He sat in a Washington DC courtroom and explained, under oath, how the ranking system actually works. The transcript runs to hundreds of pages. It is publicly available.
I read all of it.
I also found the patent Google filed in 2012 for what they call a Site Quality Score. And a 2020 internal email from Hal Varian, Google’s Chief Economist, that was entered into evidence as a trial exhibit.
What these three documents describe is significantly more complex than anything on most SEO blogs. They also explain things that had been confusing me about client results for years. Why some pages with genuinely good content don’t rank. Why new sites struggle even when they do everything technically right. Why some clients see movement in two months and others take eight.
This article is my attempt to translate what those primary sources actually say into something actionable for Nigerian businesses. Not theories. Not correlation studies. What the people who built the algorithm said about it when they were legally required to be honest.
The Belief That Keeps Most Nigerian Businesses Stuck
Most businesses approaching SEO in Nigeria still think of it as a keywords-and-content game. Put the right words on the page, get some links, Google rewards you.
That model is about 15 years out of date. And the agencies still selling “10 keywords” or “30 blog posts a month” packages are built on it. Here is what to watch out for when hiring.
The modern Google ranking system is a multi-stage pipeline with over 100 signals, three machine learning models trained on actual user behaviour, a click-signal system that has been running since at least 2005, and a site-level quality score tied directly to how many people search for your brand by name. None of that is adequately captured by “put the right keywords on the page.”
Understanding the actual system is the difference between doing SEO that works and doing SEO that keeps looking busy.
What Nayak Actually Described: The Four Stages
When you type a query into Google, a four-stage filtering process begins. Each stage reduces the number of candidate pages dramatically.
Nayak described four distinct stages of filtering, each handled by different systems. Most SEO advice focuses on Stage 1. The systems that actually decide your final ranking live in Stages 2 and 3.
The system is called Navboost. It stores 13 months of click data. Every search your target audience performs, every result they click or ignore, is actively shaping your rankings right now.
Navboost. This Is the Part That Changes Everything.
Navboost is a ranking system that memorises user clicks on search results and uses that data as a ranking signal. Every time a user searches for something and clicks a result, that interaction is recorded. Navboost has been running since at least 2005. Under cross-examination, Nayak confirmed it directly.
“The navboost system memorizes past clicks that have been issued for past queries… Navboost memorizes this information for all queries received in the past 13 months.”
Pandu Nayak, SVP of Search, Google. Under cross-examination, US v. Google antitrust trial, October 18, 2023.Thirteen months. Every search, every click, from every Google user in your target market, for the past 13 months. That data is actively shaping which pages rank right now.
Navboost does not just count raw clicks. It slices data by locale, so what users in Lagos click carries different weight than what users in London click for the same query. It also slices by device type, because mobile and desktop intent differ. The testimony gave a specific example: a search for “Bank of America” on desktop often means online banking. The same search on mobile often means finding a nearby branch. Different results, same query. Navboost handles that distinction automatically, because it learned it from user behaviour.
How far back Navboost remembers every click
Confirmed under oath. Before 2017, Navboost stored 18 months of click history. Google ran extensive experiments before reducing it to 13. The reduction was not arbitrary: they tested it against quality metrics first. This click memory is actively shaping your rankings today.
What does this mean for a new Nigerian business website? Your site has almost no Navboost history. Google has no click data on your pages because users have not clicked them. This is one real reason new sites struggle regardless of content quality. It is not just about backlinks or technical setup. It is about the absence of 13 months of positive user signals.
I had a client in the professional services space in Lagos who had been publishing well-written content for six months with almost no ranking movement. Technically sound site, good content. What they lacked was click history. We shifted focus to building their Google Business Profile, getting them listed on the major directories, and directing their LinkedIn audience back to the site. Rankings started moving within eight weeks. Not because the content changed. Because Navboost finally had data to work with.
This is why Google Business Profile and directory listings are not optional extras for a new Nigerian business. They are the mechanism through which you build click history on platforms Google already trusts, while your own domain builds signals over time.
The Site Quality Score. Google Patented This in 2012.
Navboost operates at the query-result level. There is a separate, higher-level system that evaluates your entire domain. Google filed a patent for it on January 5, 2012. It was granted May 12, 2015. Patent number: US 9,031,929 B1. Title: “Site Quality Score.”
The mechanism is not complicated once you see it. Google tracks two categories of queries related to your site.
First: navigational or brand queries. These are searches where the user is specifically looking for you. “UGO Anums SEO.” “ugoanums.com.” Your name. Your business name. Your domain. Second: all queries that lead to a click on your site, regardless of what the user was originally searching for.
“A site quality score for a particular site can be determined by computing a ratio of a numerator that represents user interest in the site as reflected in user queries directed to the site, and a denominator that represents user interest in the resources found in the site as responses to queries of all kinds.“
Brand queries as a ratio of total queries that bring people to your site. That is the Site Quality Score.
Not a vanity metric. A ranking input that affects every page on your domain.
Every time someone hears your name at an event and Googles you: quality signal. Every LinkedIn post that makes someone search your name: quality signal. Every training session, podcast, or press mention where someone later types your name into Google: quality signal. A new Nigerian business starts with a ratio close to zero. Every piece of brand-building activity raises it. Slowly, yes. But the signals compound over time, and they feed into the ranking system at the domain level.
This is why I keep telling clients that brand building is not separate from SEO. It is not even adjacent to SEO. For a new domain in this market, it is SEO.
The Three Machine Learning Models
RankBrain, DeepRank, and RankEmbed evaluate the top 20-30 results and can adjust their scores. All three are trained on click and query data. RankBrain specifically retrains on a rolling schedule because, as Nayak explained, without fresh data it would be “blind to new events.” It needs current user behaviour to stay accurate.
What these models are actually doing is evaluating whether your page satisfies the user’s real intent, not just whether it contains the right words. They have been trained on the behaviour of hundreds of millions of searchers. If users click your result and immediately return to Google to try something else, that pattern is observed and weighted. If they stay and engage, that pattern is observed too.
Writing for genuine user intent is not a soft content recommendation. It is a technical requirement of how the ranking system is built.
What Hal Varian’s Email Actually Said
In May 2020, Hal Varian, Google’s Chief Economist, sent an internal email to Daniel Russell at Google’s Search Quality team. It was later entered into evidence as Trial Exhibit UPX0243.
Varian was pushing back against the argument that Google’s dominance comes purely from having more data than any competitor. His point was sharper than that.
“Much of Google’s improvement over the years has been due to thousands of people… identifying tweaks that have added up to Google as it is today.”
Hal Varian, Chief Economist, Google. Internal email, May 29, 2020. Trial Exhibit UPX0243, US v. Google.He was saying the algorithmic improvements, made by skilled analysts looking at quality data and making small, deliberate adjustments, matter more than raw data volume. The data network effect is real, but it is not the whole story.
For us, the practical application is this. Google’s algorithm is not a fixed target you crack once and then hold. It is being continuously adjusted by human beings asking: does this ranking serve the user better or worse? Every major update is an expression of someone at Google deciding the previous system was getting something wrong for a class of queries.
Your strategy has to work with that reality. You are not trying to outsmart a machine. You are trying to build a site that a thoughtful person at Google would look at and say: this is clearly the most useful, trustworthy result for this query in this market. That standard does not change with algorithm updates. The path to meeting it might, but the standard itself is stable.
What to Do With This in Nigeria
Three things, in order of urgency if you are starting from zero. (If you are also evaluating SEO agencies right now, read this first — most of what they sell does not account for any of the systems above.) And before you agree to any pricing, understand what the work should actually cost. If you want the full strategic framework for how these signals translate into a growth system for a Nigerian business, that is here. And if you want to see how the four pipeline stages map to a step-by-step ranking strategy, the practical guide to getting on the first page of Google in Nigeria is there.
Build click signals before you need them. Before your site has Navboost history, build it through platforms Google already trusts. Google Business Profile first. Then the major directories: Sortlist, TechBehemoths, Clutch. These give Google evidence that real users are finding you useful before your own domain has any history. A new client I worked with in the professional services space in Lagos went from zero to first-page Local Pack results in three weeks, not because of anything on their website, but because we built and fully optimised their GBP profile and got consistent clicks going to it. The site rankings followed about six weeks later.
Make your brand deliberately searchable. People Googling your name is not an ego metric. It feeds directly into your Site Quality Score ratio. Every speaking engagement, LinkedIn post, or press mention that causes someone to type your name into Google is a direct SEO input. Build your brand with the same intention you build your backlinks.
Write for the person, not the machine. I took over a client account in 2024: a Samsung partner running an ecommerce store in Nigeria. The previous approach had produced traffic. None of it was built around what a buyer searches when they are actually ready to purchase. After rebuilding the content strategy around transactional intent, that store saw a 1,002.55% increase in users reaching the checkout page in eight months. Same products. Same site. Different understanding of what the searcher needed at that exact moment in the buying process. If your traffic is growing but revenue isn’t following, that gap has a specific diagnosis. And if you want to understand why treating SEO and conversion as separate functions is what causes it, that argument is made in full here.
The Pandu Nayak testimony, the Site Quality Score patent, and the Hal Varian email are all publicly available documents. None of this is secret. It is primary source material that most people in the Nigerian SEO space have not made the time to read.
Now you have. If you want to talk through what any of it means for your specific business, book a free 30-minute call. I’ll tell you honestly what your situation looks like and where to start.
New insights straight to your inbox.
When I publish something new — a case study, an analysis, a breakdown — I send it out. No filler, no weekly digest of industry links. Just the writing, when it’s ready.
No spam. Unsubscribe anytime.
Want this thinking applied to your business?
Free 30-minute strategy call. You leave with a clear picture of what’s holding your organic growth back and what to do about it.
