BTS: How We Built the Relevancy Tool

Building the Relevancy Tool in-house was an extensive project designed to address the challenge of accurately measuring backlink relevance. The primary objective was to develop an automated system capable of determining the relevance of backlinks for our clients. For example, a link from a home publication would be deemed relevant for a client in the home sector, whereas a link from a sports site would not. 

Here, we explain more about how we built the Relevancy Tool and the ins and outs of how it generates a reliable Relevancy Score using AI and machine learning. 

Integration of Multiple SEO Data Sources

To ensure the tool's effectiveness, we incorporated a wide range of SEO data into the system. This approach enables us to estimate how Google might assess a backlink's relevance to our clients' domains. We selected multiple data sources, including organic keyword metrics and the HTML structure of various aspects of the linked site.

Applying AI & Machine Learning

After selecting the data sources, we turned our attention to the use of AI and machine learning. We evaluated several cloud-based AI services, including those offered by Google Cloud and IBM, but ultimately found that traditional machine learning algorithms better suited our needs. These algorithms formed the foundation of the relevancy tool. This approach allowed us to tailor the analysis to our specific requirements rather than selecting an off-the-shelf solution.

The Quantitative Relevance Scoring Model

Once our data team finalised the relevancy algorithm, we presented it to our internal Digital PR team. Together, we determined that a simple binary classification of backlinks as either relevant or not was insufficient. We recognised that backlink relevance exists on a spectrum—from completely irrelevant to highly relevant at the domain or category level. This insight led us to develop a quantitative scoring model to more accurately reflect this range of relevance.

Testing & Refinement

After six months of development, our internal data team produced a quantitative scoring method that rates backlink relevance on a scale from 1 to 100. We began testing the tool with current digital PR clients, gathering feedback to refine the tool further. After incorporating feedback from both clients and our team, we’re confident that the tool is ready for broader use but still has room for further developments. 

Deployment & User Accessibility

To maximise the tool’s utility, we collaborated with a web developer to create a user-friendly platform. We decided to create a platform rather than offering it as an internal service because we wanted to ensure it could be used for consistent reporting month by month rather than as a one-time service. This platform enables users to assess the relevance of their backlinks and use the data to inform their future strategies. Our goal was to design a tool that is accessible to users of all technical skill levels, whether they are part of an in-house team or an agency managing multiple clients.

Over the past 12 months, we’ve worked to finalise the finer details of the Relevancy Tool to support digital PR’s and marketing teams to overcome common challenges when quantifying the relevance of their coverage and its value to their SEO and digital PR performance. 

Sign up and see how it can level up your strategy, today!