The web is a graph, not a list
Every backlink tool presents your links as a list: a row per link, sortable by a metric, exportable to a spreadsheet. That presentation is convenient and it is the reason most people misunderstand their own link profile. The underlying object is not a list. It is a graph — a mathematical structure made of nodes (pages, or domains, depending on how you aggregate) and edges (the links between them).
The distinction is not academic. In a list, every row is independent and equal, so you count rows. In a graph, a node's significance comes from its position — what points at it, what it points at, and how those connections cluster. Two profiles with identical link counts can look completely different as graphs, and the graph view is the one that corresponds to how search engines actually process the web.
PageRank was a graph algorithm. It computed a value for each node by iterating over the edges, so a page's score depended on the scores of the pages linking to it, which depended in turn on their inbound links, and so on until the numbers settled. Nothing in that computation cared about your spreadsheet row count. It cared about structure.
What a tier is
A tier is a distance measurement in the graph, counted in hops from your site.
- Tier 1 is every site that links directly to you. This is what a backlink report shows.
- Tier 2 is every site that links to a tier 1 site. These are the sites feeding authority into the sites feeding you.
- Tier 3, 4 and 5 continue outward. By tier 4 you are usually looking at a substantial fraction of the indexed web, which is why the numbers stop being interesting as counts and start being interesting as shapes.
Majestic tiers the graph natively, which is unusual among link tools and is the practical reason most tier analysis in this industry is done with Majestic data. You can ask what is behind the sites that link to you without building the crawl yourself.
Note that the word tier also gets used for something else entirely — tiered link building, the practice of pointing manufactured links at your own manufactured links to prop them up. Same word, different concept. One is a description of the web's structure; the other is a high-risk tactic that exploits it. Do not let the vocabulary collision confuse you.
Why tier one only shows the surface
A tier 1 report answers one question: who links to me? That question is easy to game and easy to misread.
Consider two domains, both showing Trust Flow 25 in your backlink list. Identical rows in the spreadsheet. Behind the first is a genuine profile — a few hundred referring domains from unrelated organizations, accumulated over years, in several countries. Behind the second are forty domains, all registered within a six-month window, all linking to each other, all with similar templates. In the tier 1 view those two links look the same. One hop deeper they look nothing alike.
This is why a tier analysis is the fastest way to identify a link network from the outside. Networks have to be built, and building leaves a shape: unusually uniform tier 2 composition, mutual linking, a narrow age range, and links per domain ratios that do not occur naturally. A real site accumulates messy, uneven, largely unrelated inbound links. Manufactured authority is tidy.
There is a second reason tier 1 flatters. A link report is a list of domains, and a domain is a large object — the score attached to it describes the whole site, not the specific page your link sits on. Two links from the same referring domain can differ by an order of magnitude in what they are worth, depending on whether the page hosting them receives internal links, is indexed, and is read. Nothing in a tier 1 export tells you that. You have to go and look at the page, which is graph thinking applied downward instead of outward.
The same logic explains why buying a link from a site with impressive tier 1 metrics can return nothing. If the domain's authority was itself manufactured, you are buying a share of something that has already been discounted.
Links per domain, and what it reveals
One of the most informative ratios in graph analysis is the number of links divided by the number of referring domains. It sounds crude. It is remarkably diagnostic.
On the twenty-year profile I use as this site's reference dataset, the raw figures are 1,301,839 links from 22,260 referring domains — a ratio of roughly 58 links per domain. That sounds alarming until you look at the distribution. A single domain sent 219,159 links, 16.8% of the entire profile, from one sitewide placement in a footer or sidebar. The top hundred domains accounted for 84.3% of all links. Strip those out and the long tail is mostly one or two links per domain, which is exactly what a naturally accumulated profile looks like.
That shape — a few enormous sitewide sources, then a very long thin tail — is normal and boring. The shape that is not normal is a large number of domains each sending a similar mid-sized number of links, because that is what happens when someone builds placements to a template rather than earning them one at a time.
This is also the reason I tell people to count referring domains rather than links. The link total on that profile is dominated by a single decision somebody made once in a footer.
Fresh and Historic: two different graphs
A practical trap worth naming. Majestic maintains two indexes. The Fresh Index covers links seen recently — a rolling recent crawl. The Historic Index covers everything ever recorded, including links that have since disappeared. They are different graphs and they will give you different answers to the same question.
Comparing a Historic figure from last quarter with a Fresh figure from today is how people manufacture false alarms about link loss, and I have been handed reports built on exactly that error. Fix the index before you interpret the change. If you are tracking a trend, track it in one index consistently.
The gap between the two indexes is also information in itself. On the reference profile, 49.6% of all links ever recorded had been lost, with the median referring domain ceasing to link after 1,080 days. A large Historic-to-Fresh gap is not a problem to solve; it is the normal decay of a graph in which pages get rewritten, redesigned and retired.
What to do with this, and where the method lives
Three things follow from understanding the graph, and all three change how you work.
- Qualify links one hop deeper. Before valuing a prospect, look at what links to it. Five minutes at tier 2 disqualifies more bad prospects than any metric threshold.
- Read shape, not size. Distribution across domains tells you more than any total. Uniformity is the signature of construction.
- Expect decay and monitor for it. The graph is not static. Edges vanish constantly, and 97.4% of the time on the reference profile the source page was still perfectly alive when the link went.
Module 3 of the course, Reading the link graph, is where the actual method lives. It covers tiers two through five in operational detail, how to read Trust Flow and Topical Trust Flow against the graph, how to detect networks and footprints, co-citation and related-site analysis, and a step-by-step tiered audit you can run and hand to a client. This page is the concept; that module is the procedure. If you are here because you need to do the work rather than understand the idea, go there.
Questions
What is the difference between a link graph and a backlink profile?
Your backlink profile is the set of edges pointing directly at your site — tier 1 only. The link graph is the entire network those edges sit inside, extending outward through every site that links to your linkers and beyond. A profile is a slice; the graph is the object being sliced. Most analysis errors come from treating the slice as the whole.
How many tiers deep is it worth looking?
Two, in almost every practical case. Tier 2 tells you whether a prospect's authority is genuine or constructed, which is the question you actually need answered. Tier 3 is occasionally useful when investigating a suspected network, because the reciprocal structure shows up clearly. Beyond that you are looking at a large share of the web and the signal is gone.
Does tier 2 link building work?
Building links to the pages that link to you is a different thing from understanding tiers, and it is graded high-risk on this site. The mechanism is not imaginary — strengthening a linking page can in principle strengthen what it passes — but in practice it means building manufactured links, which carries the risks manufactured links always carry, at one remove where you control even less.
Which tools show the link graph rather than a list?
Majestic tiers natively and exposes tier 2 through 5 counts directly, which is why graph work in this industry tends to be done there. Other major tools expose tier 1 thoroughly and require you to run a second lookup on each referring domain to get one hop deeper. That works, it is just slower, and it makes systematic tier analysis impractical at scale.