03 - What Google Ranks On - SEO Fundamentals

1The signal families

The short version

Ranking is the third thing Google does to a page, after crawling and indexing it. To rank, Google takes every indexed page that could answer a query and orders them using many measurable facts about each page and each site. This module groups those facts into eight families and reads each family against our own site: what it already shows Google, what it lacks, and what the agency's 57 tourism pages and 13 city pages look like when held up against Google's spam policies. At the end you write a one-page inventory of the evidence the site offers for experience, expertise, authoritativeness and trust, and the gaps, which module 04 uses in the site triage.

A ranking signal is one measurable fact about a page or a site that Google uses when it orders results: whether the words of the query appear in the page title, how many other sites link to the page, when the page was last changed, how far the searcher is from the business. A ranking system is a program inside Google that combines signals for one purpose. Google publishes a guide that names its systems: among them a link analysis system built on PageRank, freshness systems, a site diversity system, deduplication systems and spam detection systems. The guide says what each system is for and gives no weights, and no published source does, so any claim that "links are 30 percent of the algorithm" is a guess. This module gives no weights either.

The eight families below organize what Google has said so that each piece of work on the site can be tied to the signal it serves. They overlap: a link from the Richmond chamber of commerce is a link signal, a local signal and evidence of authoritativeness at once.

The eight families of signal this module covers, around one page from the planned page map. Google combines them; it publishes which systems exist and what they are for, and publishes no weights. Illustration.

2Relevance

Relevance is how well a page matches what the searcher meant. Google's own description of ranking says the most basic relevance signal is that the page contains the same words as the query. A Richmond office manager types "managed IT services Richmond VA". A page titled "Managed IT Services in Richmond, VA" whose first paragraph names the service and the city matches every word, and that is the oldest and plainest way a page becomes a candidate. Module 02 covered where those words count most: the title, the headings, the first paragraph and the URL.

Word matching is the first layer; meaning is the second. Google's ranking systems guide names systems built to match concepts rather than strings, among them neural matching and BERT, a language model that reads a word in the context of the words around it. Two consequences follow for us.

The first is synonyms. A searcher who has never heard the phrase "managed services" types "outsource our IT department Richmond" or "company to run our IT". To Google, those can be the same request as "managed IT services Richmond", because the pages that rank for one tend to rank for the other, and because its language systems have learned that an outsourced IT department and a managed services agreement describe the same arrangement: a monthly fee, a helpdesk, someone else owning the servers and the patching. Our page qualifies for all of those searches when it describes the arrangement in plain terms a buyer would use, including a sentence saying the service replaces or supplements an in-house IT department. The obvious response is to paste every synonym into the page. Google's spam policies name keyword stuffing, words repeated without regard to how the page reads, as a violation, and a paragraph that reads as a list of search terms reads worse to the person who arrived. Module 05 collects the synonyms Richmond buyers use; module 06 decides which page carries each one. On the page, a synonym appears where a writer would naturally use it, once or twice.

The second is entities. Module 02 introduced the Knowledge Graph as Google's store of things it knows about and how they relate: Richmond, Virginia is an entity; Henrico County is an entity next to it; Microsoft 365 is an entity that belongs to the Microsoft entity. A page that mentions Microsoft 365, Intune and Entra ID while explaining what we manage tells Google what topic the page sits in without any of those words being in the query. A page that mentions Maymont and the Metro Richmond Zoo tells Google the page is about things to do in Richmond, which is the topic 57 of our 139 pages sit in today. Those 57 pages are relevant to searches we have no reason to want.

3Experience, expertise, authoritativeness and trust

E-E-A-T is Google's shorthand for the four qualities it asks its quality raters to look for: experience, expertise, authoritativeness and trustworthiness. A quality rater is a person Google pays to score search results against a long written guideline document. Google says the raters' scores do not change the ranking of any page directly; it uses the scores to test whether a change to its systems made results better or worse. Until December 2022 the term was E-A-T. Google added the first E, experience, to the rater guidelines that month, so a page is now also judged on whether whoever made it has first-hand experience of the subject. Google's helpful-content guidance says that E-E-A-T itself is not a ranking factor, that its systems use a mix of signals that tend to line up with it, and that of the four, trust matters most. Google is measuring these four things by proxy, and a site that shows evidence for them gives the proxies something to find.

Each letter has a plain meaning and a concrete form on a ten-person MSP's website.

Experience. The person behind the content has done the thing. A page about ransomware recovery written by or with the technician who restored a Richmond client's file server from backup after an attack, and saying so, shows it. A case study naming the industry, the problem, what was done and who did it is the clearest form; a post written from a real Syncro ticket, with the client's name removed, is another.
Expertise. The content shows knowledge of the subject. A Microsoft 365 page that explains when a 15-seat law firm needs Business Premium rather than Business Standard shows expertise; a page that says "we offer Microsoft 365 solutions" shows nothing. Named authors with titles and certifications carry expertise onto the page. Today the site's three author archives are Jeff, the info@ mailbox and the agency, so a reader of most posts sees a mailbox or a marketing company as the author.
Authoritativeness. Other people treat the site as a source. This is the letter the site cannot supply on its own: links and mentions from other sites, reviews, and being named in places a Richmond business reads. Section 4 covers links.
Trustworthiness. The site is accurate, safe, honest about who runs it, and easy to contact. A street address, a phone number, a named owner, HTTPS, a privacy policy, and details that match the Google Business Profile (GBP), our listing at business.google.com, are the evidence. The site has HTTPS behind Cloudflare and one canonical hostname. Whether the address appears on the pages, and what the phone number and hours are, is not known yet.

E-E-A-T read against www.hermeticnetworks.com. The middle row is what the crawl of 2026-09-29 could see; the bottom row is what the Do this now steps confirm or fill in. Google says trust is the most important of the four.

What money cannot buy

Experience and expertise can be put on the page by the people who have them. Trust is mostly accuracy and completeness, also within reach. Authority comes from other people, over time: the chamber listing a member, a client naming a vendor, a reporter quoting a technician, a customer leaving a review. Reviews bought or written in-house violate Google's business profile policies and are removed when found; links bought for ranking violate the spam policies in section 9. The Google Business Profile Help page on local ranking names prominence, how well known a business is on the wider web, as one of its three local factors, and prominence has no settings page.

4Links and anchor text

A link is the clickable connection from one page to another. When the two pages are on different sites, the receiving site calls it a backlink or an inbound link. The anchor text is the visible, clickable words of the link. Google's ranking systems guide names a link analysis system based on PageRank, the algorithm that made Google different from the search engines before it: it treats a link from page A to page B as A vouching for B, and counts how many pages vouch for B and how much standing those pages have. The usual phrase is that links are votes, and the votes are weighted. A link from a page with many links of its own, on a site about a related subject, counts for more than a link from a page nobody links to, and a link Google judges to have been bought or traded counts for nothing or counts against the site.

One legitimate local link. The other member names are placeholders. The anchor text, "managed IT services in Richmond", tells Google what the destination is about in words we did not write. Illustration.

Anchor text matters because it is a description of our page written by someone else. A link whose anchor text is "managed IT services in Richmond" tells Google more than one whose anchor text is "click here" or "www.hermeticnetworks.com". The obvious move is to ask every linking site to use the exact phrase we want to rank for. Google's link spam policy treats a pattern of inbound links with identical, keyword-matching anchor text as a sign of manipulation, because real links vary: a client writes "our IT company", the chamber writes the company name, a reporter writes "Hermetic Networks, a Richmond managed services provider". The variety is the evidence that the links are real.

No plugin creates these links. Rank Math and Yoast count links and suggest internal ones, and neither can place a link on a site we do not control; the SEO score in either plugin says nothing about backlinks. Internal links, the ones between our own pages, are a separate signal the plugins do help with: the pages linked from the navigation and from many other pages are the ones Google reads as the site's main pages. Three managed IT URLs split those internal links three ways.

A legitimate local link for a Richmond MSP looks like one of these.

A chamber or association listing

ChamberRVA's member directory, or the vendor directory of a trade association our clients belong to, such as the Virginia Dental Association. Membership is the price; the link is a side effect.

A client's vendor or partner page

Some clients publish a page naming the firms they work with: accountant, lawyer, IT provider. A Richmond law firm listing Hermetic as its IT partner gives a link from a real business, with anchor text the client wrote.

A local business publication

Richmond BizSense or the Richmond Times-Dispatch business section covers hires, moves and local incidents. A quote from a Hermetic technician about a ransomware wave hitting Richmond practices, with a link, is a link from a site about Richmond business.

Links that look wrong to Google: a link in exchange for money without a rel="sponsored" attribute, a swap of links between two unrelated sites, a listing in a directory of thousands of unrelated businesses that exists to sell listings, and a blog comment with our URL in it. Google's guidance on qualifying outbound links defines three link attributes: rel="sponsored" for paid links, rel="ugc" for links in comments and other user content, and rel="nofollow" for any link the publisher does not vouch for. A paid link marked sponsored is permitted and passes no vote; the same link unmarked is a violation on both sites.

5Local signals and personalization

A local signal is anything that tells Google the searcher wants results near a place: a place word in the query ("IT support Henrico"), the location of the device, or a kind of query that Google has learned is usually local ("IT company near me", or even "IT support" typed with no place at all). Module 01 covered the local kind of search; this section is about what it does to ranking. When Google decides a query is local, it filters by place before it ranks by anything else, so a well written page from an MSP in Charlotte will not appear for "managed IT services Richmond VA" however many links it has. For the local pack, the map with three businesses under it, the Google Business Profile Help page on local ranking names three factors: relevance (how well the profile matches the query), distance (from the searcher or the place in the query) and prominence (how well known the business is, including reviews and mentions across the web).

Personalization is Google adjusting results to the person: their location, search history, device and settings. Someone who searched "Hermetic Networks" last week and then types "IT support Richmond" is more likely to see our page than a stranger is. Someone on a phone sees the local pack higher and fewer organic results on the first screen than someone on a laptop. Someone signed in to a Google account that has visited competitor sites for a month may see those competitors more often.

The consequence is that two people searching the same words from two desks in Richmond see different results, and a person in Lynchburg sees a third set. "We are number three for managed IT Richmond" describes one person, one device, one location and one moment. There are two honest ways to look at rank. A private browser window with location set to Richmond removes the history and keeps the location, and is good enough to see whether a page appears at all. Search Console's Performance report gives the average position across every searcher who saw the page, which is the figure to track month to month. Whether a Search Console property exists for the domain is not known yet. The Google Business Profile, which feeds the local pack, is managed at business.google.com and is set up in Rank Math module 08; whether the profile is claimed today is also not known.

6Freshness and tenure

Freshness is how recently a page was published or materially changed. Google's ranking systems guide lists freshness systems whose job is to prefer newer pages for queries where the searcher would expect something current, and to leave other queries alone. "Windows 10 end of support" is a query where currency matters: Microsoft ended support for Windows 10 on 14 October 2025, and the useful page is one that says what to do now. "Managed IT services Richmond VA" is a query where currency matters little; the searcher wants a provider, and an accurate page written three years ago serves them as well as one written yesterday. Freshness is a signal per query, and the work is to know which of our searches are the current-events kind. Most service searches are evergreen; most problem searches about a product version or a price change are time-sensitive.

Stale content hurts in a specific way. Suppose the blog holds a post from 2019 telling Richmond businesses to prepare for the end of Windows 7 support, which arrived on 14 January 2020. Three things happen to that post in 2026. A searcher who lands on it from "Windows end of support small business" finds advice about a product six years past its end date and leaves, having learned that the site's advice may be out of date. Google, which shows a date beside many results, is offering a 2019 page for a query it treats as current, and will prefer a newer page from another site. And the post occupies the search that a current post about Windows 10 and 11 should own, so two posts from one site compete. Whether the blog holds such a post is not known; module 04 inventories the 148 posts. The choices for any post like it are to update it with a dated note and current advice, to redirect it with a 301 to a current post on the same subject, or to remove it with a 410 if nothing replaces it.

Tenure is how long the domain has been online and behaving consistently. Google's staff have said in public that the age of a domain on its own is not something they rank by. What age brings is accumulated evidence: years of links from other sites, years of mentions, the same business name at the same address, and a record of never having hosted spam or malware. A domain registered last month has none of that, however good its pages. The registration date of hermeticnetworks.com is not known to this course; the Do this now steps look it up. Two rules follow. The agency's redesign did not reset the domain's history, because the domain stayed the same. Moving to a new domain would reset it, because every backlink would point at the old address until redirected, and the history belongs to the hostname.

7Legitimacy and source diversity

Google reads the context around a link to decide how legitimate it is. A link inside the body of an article about Richmond IT firms, surrounded by text about IT firms, under a heading about IT firms, on a page other pages link to, is a strong vote. The same link in a footer block of forty links to unrelated sites, on a page nothing links to, is a weak vote or none. The context Google is believed to read is the link's placement on the page, the words around it, the nearest heading, what else the page links to, and whether the linking page has inbound links of its own. Google does not publish this list; it follows from the link spam policy, which describes the patterns it penalizes, and from PageRank, which passes a page's standing on to the pages it links to.

Source diversity is the variety of kinds of site that link to us. Thirty links that all come from blog comments look like one person's work; thirty links spread across a chamber directory, two client vendor pages, a Richmond BizSense article, a vendor's partner directory, a nonprofit's sponsor page and a few blogs look like a business that exists in the world. Google's spam policies describe large-scale link exchanges and automated link building as link spam, and a link profile from one kind of source is the shape those schemes leave behind.

Diversity also applies to the results page itself, and this is the point with the biggest consequence for our site. Google's ranking systems guide describes a site diversity system that generally shows no more than two listings from one site in the top results for a query, so that a searcher sees different sources. Our site has three pages about managed IT in Richmond: /managed-services/, /managed-it-support-richmond-va/ and /managed-it-services-in-richmond-expertise-to-elevate-your-business/. At most two of them could appear together, and in practice Google's deduplication systems pick one, while the others hold internal links, backlinks and content that could have strengthened the winner. Backup and disaster recovery has three URLs, VoIP two, cybersecurity three. The repair is the page map, one page per search, which Rank Math module 02 build.

8User behavior data

User behavior data is what Google records about what searchers do: which results they click, whether they come back to the results page, how they refine the query, what they do on other Google products. The claim that Google ranks pages by how often they are clicked is the most repeated one in SEO, so this section sticks to what Google has said and marks the rest as unknown.

Google's public statement is narrow. On its How Search Works pages it says it uses aggregated and anonymized interaction data to assess whether search results are relevant to queries, and it describes that data, alongside the quality raters' scores, as an input to the testing of changes to its ranking systems. Google's representatives have also said in public talks that click-through rate on a result is not used directly as a ranking signal for that page. Beyond those statements, Google publishes nothing about how interaction data affects a given page's position. In 2023, testimony in the United States antitrust case against Google described internal systems that use click data; Google has not documented them. The honest position: Google says it uses behavior in aggregate to evaluate its systems, denies using clicks directly to rank a page, and the full picture is not public.

The work is the same under every reading. A page built to be clicked, with a title that promises more than the page delivers, loses either way: if clicks count, the searcher who leaves in five seconds cancels the gain; if they do not count, the title bought nothing. A page built so the visitor finds what the title promised, reads it and calls, wins either way. The measurable version is in Google Analytics 4 (GA4), Google's current analytics product, which reports whether a visitor engaged with the page or left at once. Whether GA4 is installed is not known yet; Rank Math module 06 connects it.

9Negative factors

A negative factor is anything that lowers a page or a site rather than raising it. Google publishes them as its spam policies, a list of practices that lead to a page or site ranking lower or being removed from the index. The policies that a small business site is most likely to brush against are these.

Spam policies most relevant to a small business site, and the state of www.hermeticnetworks.com against each on 2026-09-29
Policy	What it describes	On our site today
Doorway pages	Pages made to rank for many similar searches that all lead the visitor to the same place. Google's own example is multiple pages targeted at specific regions or cities that funnel users to one page.	13 city landing pages and 21 service pages with a place in the URL. Whether they differ beyond the city name is not known; the crawl could read only the homepage in full.
Thin and scaled content	Many pages with little original value, produced at volume for search traffic. Google's helpful-content self-assessment asks whether a site produces content on many topics in the hope some of it ranks, and whether it entered a topic without expertise because the traffic looked attractive.	57 tourism and city-guide pages about parks, museums and a racetrack on an IT company's site. Off-topic by any reading; whether each is thin is not known until it is read.
Keyword stuffing	Words or place names repeated without regard to how the text reads, including lists of cities.	The homepage description repeats the title's phrase "IT Support & Cybersecurity Services" and lists four places. One repetition is ordinary; the city pages need reading.
Hidden text and links	Text colored to match the background, placed off-screen or sized to zero, or links hidden in a single character.	Not known. Checked by viewing the source of a city page and a tourism page.
Link spam	Buying or selling links that pass ranking value, large-scale link exchanges, automated link building, and paid links without `rel="sponsored"`.	Whether the agency built or bought links is not known. Module 04 looks for it.
Cloaking	Showing Google different content from what visitors see.	No evidence. Cloudflare blocked our automated reads of most pages; Googlebot is normally allowed through, and Search Console's URL Inspection tool shows what Google fetched.
Hacked content and malware	Pages or code injected by a third party, often invisible to the owner.	Not known. Search Console's Security Issues report is the place it would show.

The honest reading of the 57 tourism pages and the 13 city pages against that list has two parts. The city pages match the doorway description if, opened side by side, they turn out to be the same page with the city name swapped; that test is in the Do this now steps, and the answer is not known yet. The tourism pages miss the doorway description, because a page about the Metro Richmond Zoo funnels nobody toward IT services. They match the helpful-content description of content made for search traffic on a topic the site has no standing in, and at 57 of 139 pages they pull Google's reading of what the site is about toward Richmond tourism by that proportion.

A penalty, in Google's vocabulary, means one of two things. A manual action is a decision by a person at Google that a site violates a spam policy; it is reported in Search Console under Manual actions, and the owner can fix the problem and request a review. An algorithmic demotion is a ranking system lowering a page or site for what it detected, with no notice sent. Whether either applies to our site is not known. A manual action is visible only in a Search Console property for the domain, and whether one exists is not known. An algorithmic demotion shows as a traffic drop against a baseline, and there is no baseline yet. The 57 and the 13 are a risk to document in module 04, and nothing in this course should call them a penalty until Search Console can be read. Module 08 covers what to do if one is found.

10Do this now

Open a spreadsheet and name a sheet "E-E-A-T inventory": four rows (Experience, Expertise, Authoritativeness, Trust) and three columns ("Site shows today", "Where (URL)", "Missing"). Open https://www.hermeticnetworks.com/ in a private browser window and fill the Trust row from the homepage, footer, contact page and About page: street address, phone number, named owner, HTTPS padlock, privacy policy link. Write "absent" for anything you cannot find in two minutes.
Open any three blog posts and write down the author shown on each. If it is "Hermetic Networks", the info@ mailbox or Tech Pro Marketing, record "bylines show a mailbox or the agency" in the Expertise row under Missing, and note which technician could own each post.
Ask each technician which certifications they hold. Write the list in the Expertise row, then search site:hermeticnetworks.com followed by each certification name. Record which appear on the site.
Open each case-study page you can find from the navigation or page-sitemap.xml. Record whether it names the client's industry, the problem, what was done and a named technician. The count with all four goes in the Experience row under "Site shows today"; the rest under Missing.
Search "Hermetic Networks" in the same private window. Write down whether a Business Profile panel appears, its review count, and every site other than ours on the first page: those are the mentions we already have. Record them in the Authoritativeness row. Then look up hermeticnetworks.com at https://lookup.icann.org/ and write the registration date in the Trust row.
Open two of the 13 city landing pages side by side, for example /midlothian-va/ and /sandston-va/, and one of the 57 tourism pages. Write one paragraph: what differs between the city pages besides the city name, and what the tourism page offers a visitor who wants IT help. Module 04 starts from that paragraph.

12Glossary

Ranking signal: One measurable fact about a page or site that Google uses when ordering results, such as whether the query words appear in the title.
Ranking system: A program inside Google that combines signals for one purpose: link analysis, freshness, site diversity, deduplication, spam detection and others named in Google's guide.
Relevance: How well a page matches what the searcher meant, judged by the words and then by the meaning, including synonyms and entities.
Entity: A thing Google knows about in its Knowledge Graph, such as Richmond, Virginia or Microsoft 365, recognized on a page whatever words name it.
E-E-A-T: Experience, expertise, authoritativeness and trustworthiness: the four qualities Google's quality raters look for. Experience was added in December 2022; trust matters most.
Quality rater: A person Google pays to score search results against its written guidelines. The scores test changes to Google's systems and change no page's ranking directly.
Backlink: A link to one of our pages from another site, counted by Google's link analysis system as a weighted vote.
Anchor text: The visible, clickable words of a link; a description of the destination page written by the linking site.
Link attributes: rel="sponsored" for paid links, rel="ugc" for links in user content, rel="nofollow" for any link the publisher does not vouch for. A paid link without one is link spam.
Personalization: Google adjusting results to the searcher's location, history, device and settings, so two people see different results for the same words.
Freshness: How recently a page was published or materially changed. Google prefers newer pages only for queries where currency matters.
Site diversity system: The Google system that generally shows no more than two listings from one site in the top results, so pages from one site aimed at one search compete with each other.
Doorway page: A page made to rank for one of many similar searches, such as a copy per city, that sends the visitor to the same place. Named in Google's spam policies.
Manual action: A decision by a person at Google that a site violates a spam policy, reported in Search Console and reviewable after a fix. An algorithmic demotion is a ranking system lowering a site without notice.

13Sources

Google Search Central, A guide to Google Search ranking systems: link analysis and PageRank, freshness, site diversity, deduplication and spam detection systems.
Google, How results are automatically generated and Rigorous testing: relevance beginning with matching words; aggregated, anonymized interaction data and quality raters used to evaluate results.
Google Search Central, Creating helpful, reliable, people-first content: E-E-A-T, quality raters, trust as the most important of the four, the self-assessment questions.
Google Search Central Blog, E-A-T gets an extra E for Experience, December 2022.
Google Search Central, Spam policies for Google web search: doorway pages, hidden text and links, keyword stuffing, link spam, cloaking, scaled content abuse, hacked content, malware.
Google Search Central, Google Search Essentials.
Google Search Central, Qualify your outbound links to Google: the three link attributes.
Google Business Profile Help, How to improve your local ranking on Google: relevance, distance and prominence.
Google Search Console Help, Manual Actions report.
Microsoft Lifecycle, Windows 7 (end of support 14 January 2020) and Windows 10 Home and Pro (end of support 14 October 2025).
Stephan Spencer, Eric Enge and Jessie Stricchiola, The Art of SEO, 4th edition (O'Reilly, early release 2021), chapter 2.
Site facts from a crawl of www.hermeticnetworks.com on 2026-09-29: homepage source, sitemap_index.xml and its child sitemaps, the navigation. Cloudflare blocked automated reads of most other pages.

Documentation read 2026-10-01. Google changes often; where this page and the documentation disagree, the documentation is current.

What Google Ranks On