Conversation Share of Voice, the percentage of any specific conversation you own, is an adaptation of the global Share of Voice (SOV) metric. Conversation Share of Voice is more precise becasue it looks at specific conversations within a market versus focusing only on global SOV compared to competitors. While it’s interesting to know how your brand or product is doing in the world of all products, you can make the greatest impact by going local with specific topic areas.
Fixate’s Re:each platform has algorithms which derive conversation share of voice across traditional and social media. The phases of calculation are data collection, normalization, and interpretation. We can’t give you the secret sauce, but this post will give you an idea of how we do it.
1. Identify your place: Re:each will identify specific keywords and concepts associated with your brand and product based on those concepts that appear the most in all conversations you participate in.
2. Determine your conversations: From there, the concepts are applied across a body of sources in order to identify the three conversations which are most relevant to you. For each vendor, there are three types of conversations identified:
3. Find your competition: Competition is derived by identifying the top 4-9 vendors in each conversation based on their SoV in those conversations. Competition very often will not translate to your direct product competition. In addition to your direct competition, our approach looks at all brands or products with which you are competing for attention.
4. Determine your share: No matter how small or large your share of a conversation is, your brand will always show up in the calculation.
5.Determining relevant topics: Topic suggestions are derived from entity/concept extraction of content that was most prevalent in each conversation selected over the set period of time. Those concepts that had the greatest reach in that conversation are weighted and end up as the core elements of a suggestion.
Data is collected from traditional social media sources as well as trusted media sources for each broad market. Weight is put on content based on the source it came from using a proprietary algorithm.
Trusted media sources are selected by a combination of calculating that media source’s SoV in a specific market, and vetted domain individuals.
Vendors can evaluate the data based on several key use cases:
Currently, calculations are done at the end of each month for the entire month’s worth of data.
The machine-learning used in SOV is human supervised (Human-In-The-Loop). SoV calculations can be fully automated; however, topic suggestions are subject to language challenges, and domain expertise based on raw data collection. Therefore, the final phase of any report is reviewed by domain experts. Domain experts validate SoV calculations, and reformulate raw entity extraction on top-performing content in each conversation to build coherent topic suggestions. Human domain expertise is part of the process for validating accurate conversation-based SoV and creating topic suggestions for every report. Domain experts are selected based on hands-on experience with the products and services in each conversation.