MY WORK
INSTAGRAM REACH ANALYSIS
Enhance Your Social Media Reach and Results with Data-Driven Insights
I uncover the dynamics of your social media channels to help you communicate more effectively with your followers and consumers. Through in-depth analysis of interaction patterns and reach, I help you identify which content has the biggest impact and how to increase your reach!
Increase the efficiency of your social media posts through new insights
Improve your conversions and click-through rates
Get Strategies that facilitate long-term growth
HIGHLIGHTED PROJECT FRAGMENTS
Below are some highlights from one of my marketing performance analysis projects. Although you will only see a few snippets from a much larger project below, it can give you a better idea of how I work, and what I can do for you.
Note: names and other sensitive information regarding involved parties and the data and other information used will be masked, modified or anonymised.
PROJECT OBJECTIVES
-
- Mapping the performance and effectiveness of the client's Instagram content.
- Identifying the primary sources of traffic and engagement for the client's Instagram content.
- Discover which content, captions and hashtags are most effective in generating traffic and engagement
- Developing recommendations and a data-driven strategy for the client's future Instagram content.
DATA OPTIMIZATION
A first step in almost every project is collecting and cleaning the relevant data. In this specific project, the data was provided by the client in several separate data frames and Excel files. These were then combined into a single large dataset, allowing for a precise and comprehensive analysis that leverages all aspects of the available data.
Before conducting data analysis, it is crucial to check the quality of the data and understand the various available variables and observations. Most datasets, including the one for this project, contain errors or are incomplete. Therefore, each dataset is thoroughly verified first.
After any missing data is supplemented or removed, a further initial exploration of the data is conducted. This involves examining the distributions of the values across all variables. Histograms are often used to visualize the distributions, providing insights into the skewness and kurtosis of the data. Additionally, boxplots can reveal the spread of the data and the different variables, and they can identify outliers—values and observations that deviate significantly from the mean and can greatly impact models and analyses. These graphs often provide a very accurate picture of the distribution of variables, informing us on how to handle the dataset and which methods to use before and during the analysis. This initial exploration is crucial; all subsequent analyses are built on these initial findings.
The data for this Instagram Reach Analysis follows a distribution significantly different from a normal distribution. The histograms show that almost all variables have a positive skew. Furthermore, all variables contain outliers, all of which significantly exceed the means and never fall below them. While these findings do not directly impact our planned analysis, it is important to note them for any future analyses that might depend on a normal distribution or be sensitive to the effects of outliers.
DATA ANALYSIS
The power of data lies largely in a solid analysis of the right information. Most organizations have a wealth of data at their disposal, although they may not always realize it or make full use of it. This is a missed opportunity because data can be the key to better insights and improved results for almost any organization. Data analysis can take many forms, with each project and goal requiring a customized approach tailored to the specific circumstances and characteristics of the client and the assignment.
An initial step in the analysis often involves exploring the relationships between different variables, as was the case in this project. A correlation matrix analysis is conducted to identify strong linear relationships between pairs of variables. The correlation between variables can be crucial for the predictive power of models, but it also poses risks for many statistical analysis methods, as some methods become unreliable with too strong correlations. Therefore, the methods for further data analysis often depend on the correlation between variables.
In the dataset for this project, no problematic correlations were identified. While some correlations are on the higher side (> .8), these are exceptions that can mostly be explained by similarities between the variables in question. For example, we see this among several variables related to different sources of impressions, as well as between impressions and other engagement variables such as likes and follows.
In addition to a correlation analysis, this project also involves analyzing the non-numeric variables: captions and hashtags. These variables include all the words used per Instagram post in both the captions and hashtags. An initial overview of the frequently used words provides good insights into the client's strategy and the descriptions used to reach consumers on Instagram. A word cloud clearly shows which words or combinations of words are frequently used. The larger the words, the more often they are used. For example, we see that a few words clearly stand out in the captions, whereas there is more variation in the hashtags. We also see that the captions and hashtags used all cover the same topics, with little overall variation. These findings provide new insights on their own and are also important for further analysis.
Further analysis of the impressions or ‘reach’ of the client’s Instagram content provides a view of the long-term performance of the posts. Despite numerous outliers, both high and low, there is an upward trend, indicating that the reach of Instagram content is slowly but steadily increasing, which is a positive sign. Additionally, there has been a recent increase in the number of positive outliers, suggesting that recent posts should be analyzed to understand the reasons for this success and how it can be replicated. As expected, most impressions come from the home menu, but a significant portion can also be attributed to the used hashtags.
Follower Conversion Rate and Engagement Rate are not directly available in the dataset but are variables created during the analysis to provide a better view of the posts' performance.
We see that, similar to the impression trend mentioned above, the follower conversion rate trend is positive, with an increasing number of positive outliers in the last 20 posts. However, the Engagement Rate shows a declining trend, with mainly negative outliers in the last 20 posts.
Further analysis of this phenomenon points to an issue with recent content. Although recent content is successful in reaching a larger audience, it lacks an active call-to-action for further engagement; following, sharing, liking, etc.
The final part of the analysis focuses on the most common words or combinations of words in both the captions and hashtags used for the client's Instagram content. The analysis identifies a dozen words and word combinations for captions and fifteen hashtags. These are the most used, as confirmed by the earlier word clouds. Correlation and regression analysis of these words show that some have a strong positive correlation with impressions and reach, while others have a clear negative relationship with impressions.
RESULTS & RECOMMENDATIONS
A solid data analysis is crucial, but the conclusions drawn from it are perhaps even more important. After a careful analysis of the data, all information and results from the conducted tests and models are compiled. Based on these results, we can formulate answers to the research questions and objectives of the project. Below is a brief and anonymized summary of the data analysis results for this project.
-
- In the long term, there is a positive trend in the reach of the client's Instagram content. However, deeper analysis of the data reveals a potential issue: while a broader audience is being reached, engagement with the content is lacking. As a result, the client is missing out on further expanding their reach and converting impressions into potential followers, which could be detrimental in the long run. Data-driven strategies to address this issue are detailed in the full report.
- Most impressions come from the home menu. Interestingly, a disproportionately large share of impressions is generated by the used hashtags, indicating that using the right hashtags is crucial for the client in reaching a large audience.
- Hashtags prove to be a relatively significant source of traffic for the client. Further analysis shows that focusing on the hashtags "python" and "data" seems to have a counterproductive effect on reach. A more targeted approach focusing on data analysis and Python projects appears to be more efficient. Additionally, a focus on artificial intelligence is highly effective in reaching a large audience. Detailed strategies to optimize the use of captions and hashtags are available in the full report.
- Comprehensive reports of the analyses and in-depth recommendations and strategies are further elaborated in the private report. Due to privacy regulations, the protection of business information, and the confidentiality of our precise methodologies that ensure optimal results, we cannot display everything on this page.
CONCLUDING REPORT
At the end of each project, a comprehensive report is prepared for the client, detailing all analysis results, conclusions, and recommendations. This report is far more extensive and specific than what is presented on this webpage, as we cannot display all data and steps.
Based on this report, clients can take concrete and effective steps to achieve their personal goals, and improve their results.
Besides a final report for the relevant assignment, I can also help you implement the recommendations, or conduct follow-up research into new insights!
