Categories
seo theory & Babbar

About babbar’s metrics

Beginning with our first article on this blog, we detailed how Babbar is made from a technical point of view. We are really proud of this technical side since we built a truly efficient software stack.

The main objective of Babbar remains to offer the webmarketing and SEO communities a tool that provides accurate and useful information about the structure of the web graph together with the high-level semantics that govern this graph. In this article we will discuss further Babbar’s metrics.

In this article, we will focus on describing babbar’s metrics and how they may be an asset in a SEO workflow rather than how they can be integrated in such workflow.

Before we start describing the metrics, let’s discuss the scope. The web graph represents interactions between pages (here graph’s nodes) through the links (here graph’s edges). Therefore, linking metrics calculated by a modern search engine are computed at page level (i.e. URL-level). To facilitate the decision making process, these metrics can be aggregated at the host-level. This level is the closest one to what the general public calls a website. Both URL and host levels are thus the most important for linking metrics

Most tools (Babbar included) offer domain-level metrics. Are they really for SEO? It is a subject open for discussion. Babbar also provides domain-level metrics because some SEOs built processes are based on this granularity, but the soundness of SEO choices which are based on domain-level metrics could be discussed (and should be in fact).

Now let’s talk about the metrics.

Page/Host/Domain Value

Babbar offers at all three different levels a metric that accounts for the popularity, resp. Page Value (PV), Host Value (HV) and Domain Value (DV).

The Page Value computation is based on a streaming method that approximates the behavior of the reasonable surfer without using semantics information carried by the links.

The Page Value is then aggregated at the host level to give the Host Value. Host Values are then aggregated at the domain level to give the Domain Value.

Value type metrics are designed to provide values ranging between 0 and 100 and on the complete web graph we expect the average to be around 50. With 77, the “frères Peyronnet” website is rather popular.

Page/Host/Domain Trust

Babbar also offers a metric that accounts for the Trust at the three different levels, resp. Page Trust (PT), Host Trust (HT) and Domain Trust (DT).

The computation is done using a transmission algorithm “a la Trustrank”. the base element is the webpage and the values are then aggregated at the host and domain levels.

The Trust ranges between 0 and 100, these values should not be directly compared to the Values’ ones (popularity). If you are familiar with other trust metrics you will quickly notice that Babbar is stricter than other tools and that many small websites have low Trust values.

The “frères Peyronnet” website has a HT of 20 and therefore has a low level of trust.

For the sake of comparison www.abondance.com has a HT of 41 (and a HV of 85, they are not in the same league).

Semantic Value for  page, host and domain

A popularity and a trust metric, up until now Babbar is not revolutionizing anything except maybe in the quality of the computation. And yes, we chose to introduce first the metrics that most of you are familiar with. But now we will talk about a new metric (Babbar is probably the only actor that can provide such a metric): the Semantic Value (SV).

Babbar provides this metric at the three levels (URL, host and domain). The Semantic Value is a popularity metric, as are the PV, HV and DV, but take into account the semantic compatibility between the source of the backlinks pointing to the analyzed page, host or domain and the target of these links. Thus it is a metric representing the behavior of a topical random surfer (topical pagerank).

Let’s consider auditing a website, the Semantic Value is the perfect metric to determine if a linking profile is topically sound.

www.abondance.com has a HV of 85 and a host-level SV of 84, therefore its linking profile is topically sound.

A non specialized website like www.spoonylife.org has a HV of 76 and host-level SV of 65, indicating that not all links are coherent with what the search engine understands about the “local” semantics.

If you are buying links, the Semantic Value of a website in your topic is the best indicator of its true popularity.

Internal Page Value

It represents the internal popularity of a page within the host it belongs to. It is the equivalent of an internal pagerank that you would compute with screaming frog and gephi for example, but using a reasonable surfer model.

Induced strength

Last but not least, Babbar allows you to simulate the induced strength of an hypothetical link between two pages. The three characteristics of a good link are known to all: 

  1. A good link comes from a page transmitting popularity
  2. A good link is topically coherent
  3. A good link comes from a trusted source

The induced strength is a metric that uses the PV, the PT and semantic distance between the two considered pages content to provide a unique value that include all the desired characteristics. As it may be difficult to provide a perfect answer every time, Babbar will approximate some intermediate results and therefore this value comes with a confidence indicator regarding the computation. This indicator can take two separate values “Low” or “High” and is written next to the induced strength.

It’s a wrap for this article, soon you will discover how you can use these metrics in order to enhance your SEO.