Prospecting into net new accounts is one thing. Selling to and keeping your current customers are some things else altogether. On average it costs 5 times more to accumulate new customers than it does to stay current ones. The power to predict when high value customers are likely to defect, are often the difference between business success and failure.
People like to be treated the way they deserve even before they mention it, Leveraging the powers of predictive analysis it is possible to understand your customer’s behavioral pattern over time and act before anything unexpected happens like drop in customer satisfaction levels and churn.
Key factors which contribute to success of a business and customer satisfaction levels are identifying the KPIs, picking suitable attributes, applying the gathered knowledge to a statistical model. Few examples are LTV(customer lifetime value prediction), churn prediction along with customer retention strategies, Next day purchase prediction. The stronger your insight into your customer preferences and their journeys, the more accurate your retention strategy. Predictive analytics utilized to watch the buying intent of your current customers, allows you to know which customers to focus on next for brand spanking new products. Deploying predictive analytics also makes it easier to spot your “at risk” customers who could also be researching a competitor’s solution.
For a better understanding let’s look at what a KPI is and how it helps with our analytics.
One of the common KPI (Key performance metric) used is Retention Rate. It is defined as the number of customers who have returned for business divided by the total customer base over a time period. This helps us understand which customers came back for business after a month or 2 months and answers the question of how likely they are to return next month or month after that?. Retention Rate combined with user data collected (Ex. User purchase history) will provide a complete view on their purchasing pattern, what brand of product attract a specific cluster of customers?, Which offer/discount strategies worked and which didn’t ?. This improves the sales and strategies every cycle.
Tackling competitive situations.
AI-powered predictive analytics also can provide you with a warning to those customers in danger of attrition. Monitoring which keywords your customers are searching, what sites they’re visiting and what content they’re viewing, can provide you with a warning to when a customer is indicative of a high intent of defection. Monitoring at-risk customers with ongoing behavior and buying patterns, MRP Prelytix gives you the info you would like to alert your sales & marketing team when accounts are showing activity around competitive offers. Now you can come up with immediate action to a competitive situation and get a hold on your customer churn rate.
The goal for any sales team is to spend longer selling and fewer time trying to seek out subsequent prospects. AI-assisted technologies like predictive analytics and machine learning will enhance your sales department performance by providing insights they have to spot a total addressable market, prioritize the accounts presumably to shop for , and ultimately retain more clients.
Now after understanding the background lets get started with some actual analysis.
Above is the image of the dataset used for analysis, this dataset contains sales information of an online retail store in different countries. Key features needed from this data are
- CustomerID
- Quantity
- UnitPrice
- InvoiceDate
To find a direction to proceed with our analysis there is a need to define a KPI which will enlighten our further journey. Defining a KPI at start can be tricky as there are multiple important features which provide a lot of info, in that case best if to perform feature engineering to develop a new feature which can summarize multiple features at once and provide a much in depth insight. Here the REVENUE generated by each customer by date can be the required KPI. The equation to define our KPI is :
Revenue = No. of active consumers * mean order quantity * mean order price
This No. of active consumers refers to no. of people who made a purchase within a decided month date range, thus averaging the order quantity and order price.
Here’s how the new feature looks like when grouped by Months -
Here’s the visualization of monthly revenue:
Consistent revenue growth can be observed in the above figure from August and a huge fall in December(data is missing for December, can be ignored for now).
Let’s look at the Growth rate by month:
Growth rate provides more specific details of how much the revenue went up relative to the previous month, here Growth in sales can be observed in month of April and August.
Everything appears good, we tend to see 36.5% growth previous month (December is excluded within the code since it hasn’t been completed yet). however we’d like to spot what precisely happened in April. Was it because of less active customers or our customers did less orders?
perhaps they only began to obtain cheaper products? we tend to can’t say something while not doing a deep-dive analysis.
Active consumers by month:
To see the small print active consumers by month, we’ll follow the steps we tend to specifically did for Monthly Revenue. ranging from this part, we will be that specialize in kingdom information solely (which has the foremost records). we will get the monthly active customers by investigating unique CustomerIDs
In April, monthly active consumers dropped to 817 from 923 (-11.5%). We will observe the same trend for a no. of orders as well.
As expected, decline is no. of order can be observed in the month of April by 8%. We recognize that Active client Count directly affected Order Count decrease. At the end, we must always positively check our Average Revenue per Order as well.
Mean revenue per order
Even the monthly order average observed for April (16.7 to 15.8). we tend to ascertain slow-down in each metric moving our KPI.
We have checked out our major metrics. In fact there are more and it varies across industries. Let’s continue investigation another vital metrics:
New consumer Ratio: a decent indicator of if we are losing our existing customers or unable to draw in new ones
Retention Rate: King of the metrics. Indicates what percentage customers we retain over specific time windows. we are going to be showing examples for monthly retention rate and cohort primarily based retention rate.
This metric can be defined by considering customers’ purchase dates. Considering consumers earliest purchase date can checking for their latest purchase provide a given time window.
Existing consumers are showing a positive trend and tell the US that our customer base is growing however new customers have a small negative trend.
Consumer Retention Rate by month:
Retention rate ought to be monitored terribly closely as a result of it indicates how sticky is your service and the way well your product fits the market. for creating Monthly Retention Rate visualized, we want to calculate what percentage customers preserved from the previous month. Equation for this metric is :
Retention Rate = Retained Consumers from prev. Month/Total active consumers
Retention Rate considerably jumped from June to August and went back to previous levels afterwards.
From all the above metrics, analytics and visualizations it becomes easy to understand the story lying underneath the data. This is how defining KPI can help in analyzing patterns in data, track consumers and develop new strategies to retain customers, and come up with new methods to increase your sales.
