Introduction
One of the major challenges faced by technology product marketing teams is placing a dollar value on their earned media. Earned media value (EMV) is essentially the unpaid visibility a brand gets via third-party entities such as media publications, customers, influencers, bloggers, product reviewers, etc., and it would be extremely beneficial if the marketing teams can measure this. They can understand which influencers are creating value for them, re-calibrate spend on paid marketing efforts, identify the best channels for earned media, etc.
A recent trend that has emerged is that of tech influencers on YouTube who review products (often mobile phones) and post their reviews on YouTube. This often plays a key role in buyers’ decision making. We’ve therefore decided to specifically focus on measuring Earned Media Value from YouTube in this article. A technical approach to implementing this has been provided as well to help you deep-dive.
How Do You Calculate Earned Media Value?
The calculation of Earned Media Value differs from platform to platform due to multiple factors such as the number of people who have an account on a platform, the number of active users, number of followers, etc. The typical metrics considered include likes, shares, comments, subscribers, views, etc.
In our scenario, we are primarily looking at calculating EMV from YouTube, focusing on three major metrics – likes, comments, and views. The formula for this (derived from a formula proposed by Ayzenberg, an advertising agency) is:
Earned Media Value Index = Total Views x DV_PV + Total Likes x DV_PL + Total Comments x DV_PCO
DV_PV = Value Per View DV_PCO = Value Per Comment DV_PL = Value Per Like
The above 3 indexes differ for each social media platform. For YouTube, the values are:
DV_PL: $0.72 DV_PCO: $8.20 DV_PV: $0.14
Implementation Approach
To implement this, we recommend using Python for data processing and manipulation. Firstly, we need to create a link between Python and YouTube to extract these required metrics. Google has an API called the YouTube API which acts as the interface between Python and YouTube.
Step 1: Setting up Google API
Please follow these steps:
1. Go to https://console.developers.google.com/
2. Create a Project.
3. Go to Library and search for the API Libraries, “YouTube Analytics API, YouTube Data API v3”. Click on the respective APIs to enable them.
4. Go to Credentials. Click on “Create credentials” and choose API Key.
5. This will generate a unique API key for your project which will be then used in the Python program.
Once you setup a Google API, in order to extract the review videos related to the product, we need to use a keyword/search phrase that yields the desired result.
Before starting the next step, please refer to this codebase here
Step 2:
To start with, import all the required packages. The packages being imported are:
1. Google Cloud – Invokes the YouTube API
2. YouTubeTranscript API – Obtains the video transcripts
3. Langdetect – Detects the language the video is in
4. mTranslate – Translates all the other language content to English
5. Pandas – For data processing
Step 3:
Declare the YouTube developer key which we got from the initial Google API setup mentioned above.
Step 4:
Define a function that makes use of the ‘build’ function in the YouTubeTranscriptAPI package. This function will fetch all the required video information for the input keyword.
Step 5:
Call the function by passing the following inputs:
1. Keyword: Let’s say we want to extract the transcript of the reviews for ‘Redmi Note 8 Pro’, then, the keyword should be something like ‘Redmi Note 8 Pro Reviews’.
2. Number of Videos: This defines the maximum number of videos to be extracted at a single instance using the above keyword. In our case, we are giving the number 49. Note that the higher the value, the greater the chances for the content to be irrelevant.
Step 6:
When we search ‘Redmi Note 8 Pro Reviews’, there is a possibility to get tutorial videos as well in the search list. Since tutorial videos can’t be used for calculating EMV, we need to eliminate them.
Step 7:
Now that we have extracted the information required. We use the EMV calculation formula to calculate the EMV for each video.
Example
We tested this on 4 mobile phones launched in Q3 2019. As seen, it’s clear that there are millions of dollars worth EMV that gets generated, with varying numbers for different phones.
Conclusion
Through such methods of EMV calculation, companies will be able to identify better how they can plan their marketing spends. It will also give them an estimate of the value that technology influencers bring to the table and potentially look at using them in their marketing efforts in a more direct manner.
