How to use descriptive analytics to drive company growth?

1.  What is descriptive analytics?

Descriptive analytics is a branch of analytics that answers the business question: “What happened?”

It looks at current and historical data to understand what is happening in the business. For example, the above question could be translated into:

1. How much revenue did we generate last quarter?
2. Are the revenue numbers growing or falling?
3. How many new customers did we acquire?
4. How many existing customers did we lose? 
5. Etc.

Usually, descriptive analytics is the foundation upon which all other analytic branches are built. 

But with the advent of big data and exponential growth in the amounts of data modern enterprises hold in their data warehouses, new more advanced analytics techniques are developed. 

With the use of machine learning and data mining algorithms, descriptive analytics is turning from a foundational technique to an advanced R & D approach to framing business decisions and business goals.

2.  How does descriptive analytics fit within the company’s data analytics work?

Data analytics is a wider area concerned with providing revenue-generating and de-risking insights. Data scientists or data analysts usually perform one of four types of analytics to answer different questions:

1. Descriptive analytics: What happened?
2. Diagnostic analytics: Why did it happen?
3. Prescriptive analytics: What should we do next?
4. Predictive analytics: What will happen next?

So if your data analysis is involved in determining what happened (descriptive analytics: How much revenue did we generate last quarter?), you will look at different data, then if you’d like to decide what future courses of action or optimizations you should take (prescriptive analytics: Should we sell more in the U.S.?), given the most likely future outcomes (predictive analytics: Revenue in the U.S. will grow faster than in other countries according to our regression analysis). 

Descriptive analytics serves as a foundation for all other data analyses. It is employed in data science and business analytics as the first step before further analysis is conducted. 

Let’s dive deeper into the practical applications of descriptive analytics.

3. What are the use cases of descriptive analytics?

Descriptive analytics reports on what happened, but not every report is part of descriptive analytics.

Let’s look at an example. Say your company made $1 million in revenue last year. Sounds like a lot, but what if I told you that that was the third year in a row where revenue dropped?

The role of descriptive analytics is to provide insights for decision-making and understanding of business performance.

This is why descriptive analytics is often used in business intelligence. By analyzing data, we can check how our performance is pacing against KPIs. For example, descriptive analytics can be used to report on how much revenue we made year-to-date and whether we hit our sales target.

Another use case of descriptive analytics is benchmarking. Setting up expectations of what is a good performance. 

Concretely, the examples of descriptive analytics will vary by stakeholders:

• Executive: Is revenue growing Year-over-Year?
• Product: How were sales impacted by increasing the pricing of our products?
• Marketing: How many marketing campaigns have had a ROAS over 5x? How many of those were run on social media vs search engines?
• Customer Success: What is the average NPS score according to customer survey results?
• Etc. 

How does descriptive analytics go about answering these questions?

4.  What techniques are used in descriptive analytics?

There are three general areas of techniques used by descriptive analytics:

1. Data aggregation. The majority of descriptive analytics can be done by aggregating data. Counting customers, summing revenue, and groping those numbers into buckets (geographical region, product, …), already provides great insights into a company’s performance. 
2. Descriptive statistics. Providing summary statistics with measures of centrality (average) or dispersion (range) often answers a lot of questions. For example, “on average, we acquired 100 more customers per week this quarter compared to last quarter” tells you a lot about growth. 
3. Data visualization. Irrespective of the type of data visualization used (bar charts, line graphs, pie charts, histograms, …), showing data with a picture tells a story of a million words. Creating dashboards with multiple visualizations is a common tool used in descriptive analytics to showcase growth trends. 

The success of descriptive analytics for driving insights is not only tied to the technique used but also to the overall process of performing analytics within your company.

5. What does a successful process deploying descriptive analytics look like?

For a descriptive analysis to be successful, you need to run through 4 necessary steps:

1. Decide on metrics to report. A successful analytic endeavor will start at the end - by asking yourself, what metrics we want to measure and report on. Picking the right metrics is the most important step. When you decide what metrics you want to showcase on your reports, you decide how decision-making will be done. For example, picking total revenue ($1M/year) vs Year-over-Year revenue change (-150% from last year) will tell two drastically different stories.
2. Prepare datasets. Once you decide on your metrics, you have to prepare the datasets to compute those metrics. The ETL process that collects raw data, transforms it and cleans it, and saves it to your data storage (database, data warehouse, or data lake) is either already done and you have to pick the right tables or you have to engineer the entire pipeline from scratch. Here engineering concerns arise. You have to foresee how the descriptive analysis will be used, to engineer the ETL data pipeline correctly. If you present data sets in real-time, your ETL pipeline needs to reflect those needs and stream data instead of batching it. 
3. Analyze datasets. Once the data has been collected and sanitized, you can analyze it using data aggregations, visualizations, and descriptive statistics. 
4. Present data sets. Deploy your insights with visualizations, metrics, and summary statistics to effectuate change. Usually, you will showcase your work in a dashboard within tools such as Excel, Power BI, or Tableau. 

Building the end-to-end descriptive analytics pipeline can be time-consuming and expensive with limited engineering resources. Unless you can automate it.

6. How to automate end-to-end descriptive analytics?

Keboola can help you automate your descriptive analytics pipelines end-to-end.

Keboola is a data stack as a service platform that helps you integrate all your data tools and automate them. 

Use it to build and deploy your analytic pipelines:

• Automate data extraction and loading with over 250 integrations between sources and targets (be it databases, data warehouses, data marts, or data lakes).
• Automate data cleaning and transformations with Transformations and Applications (bonus: it comes with data versioning).
• Automate your ETL jobs to run in real-time or in scheduled batches with a couple of clicks. 
• Connect Keboola to multiple data visualization tools (Power BI, Tableau, Excel, …) to automate your business analytics.

Try it out. Keboola offers a no-questions-asked, always-free tier, so you can play around and build your pipelines with a couple of clicks.