This marketing analytics example is illustrated through the Tableau Superstore.

Marketing analytics is the practice of collecting and analyzing data on marketing initiatives to optimize these efforts and improve ROI. A good marketing analytics program uses quantitative methods to inform decision-making and bridges data science, subject matter expertise, and business knowledge. When it’s working well, a strong analytics program can help you increase revenue, mitigate risk, and uncover customer insights. So how exactly can you turn data into dollars?

Here are three steps you can take to elevate your analytics beyond basic descriptives. The goal of these steps is to answer questions about how your metrics performed over time, how they performed relative to each other, and where there is room for optimization.  

It is important to consider the business use case – what is the goal of this analysis? Is it exploratory, or are there specific metrics that the use case calls for? Sometimes you have to make some wise guesses about the use case based on the type of data available and the target audience.  

Let’s assume, for this use case, that we are a manager on the Tableau Superstore sales team and our goal is to increase sales by 20% this year.

1. Look at the data

The first step in your analysis is to understand what the data looks like through an exploratory data analysis. What dimensions are available? What metrics are available? Does the data look trustworthy & reliable? Are there missing fields or too many nulls? What are some basic insights?

Start with a look at the fields in your dataset so that you can understand what type of data is being captured. For example, using the Tableau Superstore data source, a quick look at the tables tells us:  

• Which fields are dimensions (e.g., Order ID, Order Date, Ship Date).
• Which fields are metrics (e.g., Sales, Quantity, Discount, Profit).
• What the data types are (e.g., text, float).
• What the values look like (for instance, by clicking on "Describe..." in the dropdown menu next to the "Ship Mode" field, you can see that there are four values in the table for Ship Mode: First Class, Same Day, Second Class, and Standard Class.

Next, you can investigate what the shape of the data looks like. In this step, you’ll answer questions like:

• What is the distribution of the data?
• How does the data trend over time?
• Are there any anomalies in the data? (e.g., missing data points, outliers, duplicates)

You can plot the data with a histogram to see if your data points follow a bell-shaped curve, determine the average, and check if the data is skewed left or right.

Here’s a simple histogram of the quantity of orders, which is clearly right-skewed and indicates that the majority of data points are on the low end of the distribution.

Using any data visualization tool, you can also plot the distributions with box-and-whisker plots to see the spread of data by segment or other dimensions.

A third simple type of analysis is plotting the data along a time component, like order date. Here, I plotted the sum of sales by month to see the general trend of data over time. As a bonus, I also added a linear trend line that more clearly shows that sales have trended upwards since January 2020.

You can get fancier with these charts by further slicing and dicing the data, for example, to see if there is any correlation between your variables. Each of these charts take one to two minutes to build but can provide a wealth of information about your data scope, spread, trends, and patterns.

2. Quantifying the data

Now that you’ve got a basic grasp about what your data looks like, it’s time to put some numbers into your analysis.

With the sample Superstore data available to us, it is safe to assume that the business questions will include:

• What were our total sales last year?
• How are we pacing to achieve a target of a 20% YoY increase in total sales?
• How is our store performing by region?

Next, determine which KPI’s will answer these questions and which data slices will provide the most meaningful insights. Remember, the first few times you do an analysis, it’s okay to play around with different visuals and data slices until you find something interesting. You typically want to pick a few key visuals to focus on rather than overwhelming yourself and your audience with dozens of pieces of information without clear takeaways. Once you find something that works, operationalize it into a business intelligence dashboard that takes away the temptation to cherry-pick the data points you like best. Data is most useful when it’s telling the full story with as little bias as possible.

For the questions above, let’s list our KPIs and possible cross-sections of data to examine:

• What were our total sales last year?
  • Total sales last year
• How are we pacing to achieve a target of a 20% YoY increase in total sales?
  • Total sales this year
  • Sum(total sales last year) * 1.2 = target sales this year
  • Sum(total sales this year) - sum(target sales this year) = total gap
  • Target sales this year/12 = target sales per month
  • Total sales per month – target sales per month = gap per month
• How is our store performing by region?
  • Profit or sales by country, region, state
  • Profit or sales by segment, product
  • Trending profit and sales for the above slices

Along with KPI cards, visuals can be used to show the dynamic nature and direction of the data points. A visual is worth not only a thousand words but can also be worth a thousand numbers!

Along with a single number, which might be valuable for someone who just wants a quick answer, these visuals can provide a much more detailed and nuanced look into the overall pacing of the company in reaching its sales goals. Here are just a few of those insights:

• Sales started off pacing lower at the beginning of the year, but the company exceeded the monthly sales target by September of this year.
• Since September, the company has been consistently exceeding monthly sales targets.
• February had the lowest sales, while November had the highest sales.

There are numerous other possibilities of slicing the data and analyzing different metrics to better understand the larger picture and identify key patterns and opportunities for this business.

3. Add more data

Just like building a puzzle, adding more pieces can bring a clearer picture to the surface about what is truly causing the various observed patterns in our data. For this use case, a few possible sources of new data points can include marketing channels, more comprehensive cost metrics, and possibly even data about customer sentiment and journey through the purchasing funnels (e.g., in-store versus online, keyword analysis, etc.).

When analyzing cost, consider including not just monetary spend, but time, resources, and risk, to capture the most comprehensive estimates for marketing channels.

Once you master these basic steps, you can explore more complex analytics tools like attribution analysis, regression/correlation analysis, testing and experimentation, and automation and AI/Machine Learning. However, if you need help along the way or need a hand from data experts in the end-to-end process of the data lifecycle, we can help!

Our eBook Data to Dollars: Driving Retail Revenue with AI and Analytics, provides real-world examples and insights into successfully leveraging AI and analytics for revenue growth, increasing your retail profits.