Stephanie Camello has been a data analyst in the marketing field for over 8 years. She has worked with marketing data across multiple industries and for various advertising agencies. And now she's here to help you with tips on how to be a great data analyst!
You may be considering a new career after the being negatively impacted by the pandemic. Or maybe you want to change careers for personal career growth. Whatever your reasons, here are Stephanie Camello's top 10 tips on how to create an analytics report as a data analyst!
1) If you are brand new to analytics, start with securing an internship in the field in any industry you prefer. Note though that analytics in finance is very different from analytics in marketing or medicine. Since Stephanie's specialty is in marketing analytics, you will get more of that side of the analytics coin in this post.
2) The key to knowing what to do with millions of rows of data is to first organize and "clean" the data. If you're given an Excel spreadsheet with raw data in columns and rows, you first want to become familiar with what's in the data. Apply filters to see what each column contains. The headers should give you an idea of what the data will contain. If a header is called "State," you should expect to see fields like "Arizona" or "Oklahoma." But if you find "France" or "Sacramento" then you know this is something you need to "clean." Change the field to the correct state by looking at the row the field is in. If you're in marketing, it probably is a placement name (placement of the online ad which gives you details of the ad itself and where it is being served, the ad size, etc). This should have the info you need to correct that wrong field. This is one example of how to clean data. Keep doing this across all columns until no more errors exist. Also, don't be afraid to ask your larger marketing team on what things mean in the placement name or how things should be labeled.
3) Once the data is cleaned, you can start actually analyzing it. Typically, Stephanie likes to meet with her larger marketing team to see what goals they are looking to achieve and what key performance indicators (KPIs) are important to both the end-user (typically the client) and the marketing campaign itself. For example, if you have a lot of video ad placements, you may want to look at how long the video is and how long someone viewed the video. Did the viewer (potential customer) watch 1 second of a 30 second ad? Or maybe they watched 29 of the 30 seconds. This calculation is called VCR (video completion rate) and it will tell you how interesting your content is and if viewers saw the whole ad or not. VCR (%) = video completions/video plays (plays can also be called impressions). Once you know the relevant KPIs, you can start analyzing the cleaned data.
4) Analysis of data includes looking at the data from many different view points. This is where the Excel pivot table comes in very handy. There you can look at your metrics and dimensions you gather from the raw data's columns and rows. A pivot will allow you to aggregate and group metrics and dimensions so that if your raw data has 10 rows of the state Arizona all with the impression (ad) count of 5, you can see Arizona has 50 impressions. It summarizes the data into fewer lines so that it's easier to analyze.
5) Now that the pivot is created, you can start to observe or find trends, similarities and anomalies within the data. Anomalies are typically outliers that don't align with the other data. If every day you serve 10 impressions but one day shows 600 impressions, then that is something you'd want to look into. It could be an error in the counting system being used (like an ad platform such as Google) or it could be fraudulent (computer robot) impressions. Either way, consult with the marketing team to find the issue. You may either need to throw that 600 out as fraud and replace it with what it should have been or note it as a caveat to explain this anomaly in your final report.
6) If your data contains a date dimension, you can use this to look at trends. See if particular months show increases in customers taking actions. See if a particular day of the week sells more product than other days. See if certain times of the day have more social media engagement than other times. All these observations can be used to derive insights you can include in your report. These insights can be turned into optimizations that the team can use to improve the marketing campaign. Making a trending graph chart will help to detect patterns and trends over a period of time you choose across the metrics you choose.
7) Gather up your top insights you've observed. These will be the baseline for the story you'll tell in your report.
8) Put all the KPIs, data tables, screenshots, graphs, maps, or whatever else is in the data that supports the story you're including into your final report (could be in Excel, PowerPoint or Tableau for example).
9) Double check everything (numerically, grammatically, format-wise) before sending a final report to the client. And have your story verbally prepared in case you need to explain the data story to the client.
10) If you know the data is correct and that you've made the data clean, the charts your own with different colors and labels and the insights story makes sense, you can feel confident in your finished product!
Stephanie Camello just walked you through a common task of a data analyst. Hopefully this gives you a good look at what it takes to be a great data analyst. Be sure to check out more of her posts on in-depth analytics topics!
Comments
Post a Comment