array(['sandshrew', 'blastoisemega blastoise', 'golem', 'nidoran♀',
'beedrclefable', 'victreebel', 'bellsprout', 'venonat', 'nidorino',
'value 7 corgi', 'machoke', 'german shebellsprouterd', 'raichu',
'golbat', 'vileplume', 'nidoking', 'rattata', 'primeape',
'nidoqueen', 'poliwag', 'value 6', 'arcanine', 'ekans', 'nidoran♂',
'metapodmencement', 'ninetales', 'husky', 'value 5', 'pikachu',
'fearow', 'parasect', 'clefable', 'graveler', 'psyduck',
'tentacruel', 'charmeleon', 'alakazam', 'blastoise', 'abra',
'mankey', 'pidgeotto', 'butterfree', 'corgi', 'ponyta', 'golduck',
'charizardmega charizard y', 'clefairy', 'diglett', 'squirtle',
'kakuna', 'weedle', 'value 4', 'beedrill', 'wartortle', 'persian',
'poliwhirl', 'core value 1', 'vulpix', 'venomoth', 'sandslash',
'venusaur', 'growlithe', 'meowth', 'ivysaur', 'core value 3',
'pidgeotmega pidgeot', 'venusaurmega venusaur', 'oddish',
'weepinbell', 'cleponytable', 'charmander', 'machamp',
'charizardmega charizard x', 'pidgeot', 'core value 2',
'alakazammega alakazam', 'value 7', 'metapod', 'machop', 'dugtrio',
'nidorina', 'spikachuing start', 'geodude', 'charizard', 'gloom',
'pidgey', 'poliwrath', 'spearow', 'paras', 'pikachue college',
'zubat', 'arbok', 'tentacool', 'raticate', 'caterpie'],
dtype=object)
Social Media Content Analysis#
I analyzed the performance of social media posts by content categories and social media platforms (Facebook, Instagram, Twitter, and LinkedIn) to identify the most optimal platform to communicate a company’s value propositions.
The analysis relived that not only the posts about the company’s core values were underperforming compared to other content categories but also the value of posts the core values was much smaller. This result led me to focus on reinforcing the brand messaging as one of the strategic plans.
Background#
My social media team’s goal was to communicate the company’s value propositions in engaging ways while maximizing the exposure (=impressions) of individual posts. In order to acquire high impressions (due to most of the social media’s algorithms), constantly posting highly engaging content is essential. However, the company’s core values are not necessarily “fun” topics to discuss, and it’s tempting to just post engagement-driven posts such as memes and pretty images.
The social media team used to post the exact same content across 4 major social media channels (Facebook, LinkedIn, Instagram, and Twitter). However, as each platform has a different audience, it is required to create content that caters to each platform.
Objective#
Identify the most optimal platform(s) for each core value.
Measure of Success#
Engagement rate (# of post engagement / # of impression) *– Engagement is user’ behaivors such as “Like”, “Comments”, and “Share”. Higher the engagement rate, the higher the quality of content is.
The number of posts
The number of impression – This indicate how much a single social media post is shown on user’s timeline. This is usually used as an indirect measure of brand awareness
Data Source#
The raw data of social media posts published between the past 12 months (January 1, 2020, December 21, 2020) were exported from a social media management platform, Sprout Social as
.csv.Each post has a few tags representing the content categories and subcategories. Content categories are consist of company’s core values, values, and social media-specific content categories (such as memes, or user generated content). Subcategories are used to label campaigns, experimental posts, and types of products.
For confidentiality, any information that can identify the company was scraped out for this case study. Content categories and subcategories are modified as below:
Company’s core values: core value 1, core value 2, core value 3
Company’s values: value, 4, value 5, value 6
Social media specific categories: corgi, husky
Subcategories: Pokemon names
Results#
All three core values (core value 1, core value 2, core value 3) were not performing well across the platforms. Also, the number of posts about these core values is fewer than the other categories.
The number of posts is heavily skewed to the “value 4” category on social media, and the category is performing well.
The “value 5” content generates the highest engagement on LinkedIn while the category does not perform as much on other platforms.
The “value 6” content performs well on all channels except Facebook.
The engagement of the “value 4” category on Facebook is higher than other content categories. However, the impression is not necessarily higher, which implies there might not be a strong correlation between post engagements and impression on Facebook.
Insights#
Core values are not necessarily easy topics to cover on social media and making the topic engaging requires creativity. My assumption was the team did not fully understand the company’s values and how to talk about them. That’s why they tried to avoid the topic. Instead, they posted value 4 content because it’s an “easy win”.
Action items#
As the low performance of content about the three core values is a critical issue, reinforce the brand messaging to communicate core values by providing the social media team with brand training.
Continue to produce the “value 4” content as it helps maintain the social media accounts’ engagement.
Post more of the “value 5” content on LinkedIn.
Reduce the amount of the “value 6” content on Facebook.
Explore new ways to improve Facebook’s organic impression since there is little correlation between post engagements and impressions. The basic strategy for social media is to improve engagement, which also helps organic impressions, but Facebook needs a different strategy.
Limitation and future project#
The social media team had to manually tag the content type, which was extremely time-consuming. Also, tagging can be subjective without a clear guideline. This process could be improved by using NLP.
Processing Tags#
DataFrame by platforms#
DATA TO USE#
twitter_cleanedfacebook_cleanedinstagram_cleanedlinkedin_cleanedCreating Subsettings for content types#
tagsarray(['sandshrew', 'blastoisemega blastoise', 'golem', 'nidoran♀', 'beedrclefable', 'victreebel', 'bellsprout', 'venonat', 'nidorino', 'value 7 corgi', 'machoke', 'german shebellsprouterd', 'raichu', 'golbat', 'vileplume', 'nidoking', 'rattata', 'primeape', 'nidoqueen', 'poliwag', 'value 6', 'arcanine', 'ekans', 'nidoran♂', 'metapodmencement', 'ninetales', 'husky', 'value 5', 'pikachu', 'fearow', 'parasect', 'clefable', 'graveler', 'psyduck', 'tentacruel', 'charmeleon', 'alakazam', 'blastoise', 'abra', 'mankey', 'pidgeotto', 'butterfree', 'corgi', 'ponyta', 'golduck', 'charizardmega charizard y', 'clefairy', 'diglett', 'squirtle', 'kakuna', 'weedle', 'value 4', 'beedrill', 'wartortle', 'persian', 'poliwhirl', 'core value 1', 'vulpix', 'venomoth', 'sandslash', 'venusaur', 'growlithe', 'meowth', 'ivysaur', 'core value 3', 'pidgeotmega pidgeot', 'venusaurmega venusaur', 'oddish', 'weepinbell', 'cleponytable', 'charmander', 'machamp', 'charizardmega charizard x', 'pidgeot', 'core value 2', 'alakazammega alakazam', 'value 7', 'metapod', 'machop', 'dugtrio', 'nidorina', 'spikachuing start', 'geodude', 'charizard', 'gloom', 'pidgey', 'poliwrath', 'spearow', 'paras', 'pikachue college', 'zubat', 'arbok', 'tentacool', 'raticate', 'caterpie'], dtype=object)Visualizayion: Primary Content Category#
Performance by Platform#
Twitter#
Facebook#
Instagram#
LinkedIn#