Banja Luka, Bosnia and Herzegovina

Product Analytics Tools

Project Overview

.

Project time frame: September 2015 – April 2019

About the employer: Infomedia d.o.o. Banja Luka (infomedia.ba) develops custom software and internet marketing solutions with the purpose of web sales and package distribution.

Project challenge

Since the company was constantly growing, we needed to build reports related to new business activities. The information about the calls made using call panels was stored in our statistic dashboards and visually presented in the form of tables and charts.

The goal was to automate the reports which would result in avoiding human factor mistakes, saving time for manual data inputs and calculations, creating effective business analytics, and optimizing business processes in general.

Project solution

I built various product analytics and data visualization tools in statistic dashboards of the Phoneorder telemarketing web application that were used by executives and project leaders to analyze sales outcomes and create future sales strategies.

We also used these tools to control the work of call center agents since the reports based on them contained data about the success of each call.

Project activities:
– Looking at metrics in statistic dashboards for any weird numbers and fixing bugs if there are any.
– Using feedback from dashboard users (colleagues from other departments and business owners) to build new reports in the form of charts and tables.
– Based on my own experience with the dashboard and project needs, actively suggesting and pitching new reports and tools. 
– Creating formulas for automation of reports according to requirements.
– Creating requirements specification sheets and assigning tasks in Trello tickets.
– Looking regularly at metrics and deciding which of the existing reports need to be improved.

Project tools
– Google sheets for creating prototypes of new tables and charts.
– Trello for organizing tasks and priorities.

The Team: Medior and senior PHP, Python and MySQL developers.

.

Deliverables

.

SMS campaigns efficiency tracker

We needed to know how did a certain message performed during SMS marketing campaigns. To be able to do that, we built a tracker within the product order panel that collected and sorted data about message performances on the one hand, and message texts that yet need to be launched, on the other. That way, we were able to compare how the message of a specific product performed on a certain market. Also, since all the new message suggestions by writers went directly to the panel, it was easier to decide which message to use in a certain campaign during campaign planning. Here you can see a specification document I made for this project:

Message Performance Tracking – Strategy and Specifications (Google Doc)

.

Call agents performance statistics

Since call center agents were using our app to process the calls, we had all the data we need to analyze if they are doing it properly or not. After success factors were determined (such as average call duration, percentage of answered calls, percentage of ordered products, percentage of upsell orders, percentage of canceled orders, etc.) that data needed to be displayed in tables. This way I was tracking performances of agents on calls and acted accordingly. Also, the graph was introduced to compare performances between different agents.

Figure 1. Call agents performance statistics

.

Call center performance statistics

Similarly to the above, we built a page with a performance table for the entire call center, which enabled us to compare performances between different call center partner companies.

Figure 2. Call center performance statistics

.

Statistic of call center costs

After we hired outsourced call centers, we needed a way to track costs to see how much this investment pays off for us. I made separate formulas that calculate these costs for each of the call centers because their service fee methods were diverse (some charged costs per calls, some per talked minutes and some had a combined method). I made a specification for a table in Phoneorder app that automatically calculates cost per call and cost per order after total costs are entered. We also added a graph so we could track the tendency of costs throughout the time.

Figure 3. Call center costs tracker

.

Statistics of not called customers

Since outbound customers needed to be called regularly, there was a need for making the system that could control whether agents are doing that job properly. First, I created a formula that calculates those percentages of not called customers. Then I manually entered those figures and made the Google Sheet table. Since all that data from the formula can be found in the statistical part of Phoneorder, I made a specification for creating the table in the app that will automatically pull and calculate the mentioned figures.

Figure 4. Statistics of not called customers

.

User list table

We needed a place where all users of the app are listed and named accordingly, so we could assign them specific user roles in Phoneorder and make them active/inactive as needed. This opened a way to track the performances of call agents who used call panels.

Figure 5. User list table

.

Product order tables

Manual data input and calculations were replaced with inbound and outbound product order tables, so the costs for each product were calculated automatically by simply selecting corresponding values from the drop-down menu.

Figure 6. Product order tables