One day my CEO called me and said, “Sobe, please I need you to analyse the data from XY company by the end of the day” and I was like “okay sure”. Then I set out to work. The moment I opened the excel data, I was overwhelmed. What information should I analyse? I realised that no matter the number of analytic courses taken, knowing the right information is not something that you can pick from that, it comes from experience, working in various areas of life and understanding what the business wants to know.

So here is the approach I took.

• Understand what’s really in the data and their definitions.
• Understand what my company is solving
• Understanding the client
• Understand how my company is solving our client's problem
• Understand if the client’s problem could be a similar problem in the industry.

Now when these have been done, you can take a second look at the data and figure out which features represent the understanding acquired. Do you need to create new features from existing ones? Next, we think of compositions and relationships. How much something appears in the population and if one feature affects the other and vice versa.

Let’s take a look at how this can be done.

Here is sales data from a big retail store. This data contains 3 years of sales records and was sent to your company for business strategy but first, you have to analyse the data. Your company is a Sales consulting company and you need to study their records to advise them on how to sell strategically in the next business year.

First, we would investigate the data.

The sales data contains 2823 records(rows) with 25 features (columns). The features are either categorical or numerical. Features are; ORDERNUMBER, QUANTITYORDERED, PRICEEACH, ORDERLINENUMBER, SALES, ORDERDATE, STATUS, QTR_ID, MONTH_ID, YEAR_ID, PRODUCTLINE, MSRP, PRODUCTCODE, CUSTOMERNAME, PHONE, ADDRESSLINE1, ADDRESSLINE2, CITY, STATE, POSTALCODE, COUNTRY, TERRITORY, CONTACTLASTNAME, CONTACTFIRSTNAME, DEALSIZE.

The features needed will be dependent on the subject matter asked to solve.

A question of composition would be.

• Amount of sales within a certain period of time
• How many orders with the status Shipped, Cancelled, Resolved, On Hold, In process within a period of time?
• How much of the various productlines was sold within a certain period of time?
• How much was sold in various cities, states, countries or territory?
• What productline makes the most sales?
• How much profit was made from selling various products cumulatively, yearly and across various regions?
• Most frequent or valuable customer in a year, or region

A question of relationship would be.

• How many sales were made in a period of time compared to another?
• How much was sold in a city, state, territory compared to another?
• Are there productlines that are more sellable in some cities, states, countries or territory compared to others?
• Does the status of an order have anything to do with the cities, states, countries or territory?
• Profit difference for various years across various regions and productlines

These are variants of how some questions would appear. When you are done, you should be able to report the following.

• If there are products that shouldn’t be sold in some areas or considering the number of sales that have occurred in those areas, the quantity that would be sent there would be considered.
• If a certain period has a case of high sales, then actions would be put in place to ensure that those products are made available to meet the need at those times.
• Considering the order status and areas, there would be a need to look at troubled areas and what remedy can be put in place to avoid issues.
• Yearly rise and drop in profit across productlines and regions. Possible projections can be made for the next business year as to how much sales would be made as well as profit.
• Investigate some customers further to know if they are valuable enough to strike a partnership deal with. And lots more.

If you understand the Business KPIs and yearly objectives then it makes it easier to narrow down what we are searching for in the data. Else, one can just get stuck reviewing a lot of things and end up making a heavy report that is not fast to grasp. As we know, business people want reports they can easily view and pick what’s needed at a glance without so much stress.