Robotic Process Automation

OCR-Enhanced RPA Solution for a Retail Network

Industry: Retail
pa solution for an upper scale retailer


Retail is, arguably, one of the most flexible industries, always ready to adopt new methods and tools. There’s a good chance that the days of traditional commerce are approaching its end. Standard techniques of retail, as we knew them, are no longer relevant. The industries grow more and more complex every day, bringing the challenge of ever-evolving consumer demands and ever-increasing customer service expectations.

For many of these challenges, robotic process automation or RPA can become the most effective medium in achieving better growth figures while addressing the complex matters with a simple and easy-to-implement solution.

Our customer is one of the country’s largest electronics and home appliance stores since the early 2000s. These days their infrastructure is truly immense. With 650+ brick and mortar shops paired with one of the country’s largest e-stores, the number of customer applications, orders, and reports grew progressively. To address this, they contacted OTR to develop a solution that would cut operational costs, daily risks, and the probability of errors and speed up processing the orders’ registration and execution.

Solution: Description and Key Capabilities

RPA for a Retail Company
Automation Anywhere RPA Enterprise platform, IQ Bot v6.5
Waterfall, Scrum
Project Duration:
14 weeks
Project Team:
1 Business Analyst/RPA developer, 2 IQ Bot experts

The main target we were aiming at was to replace the traditionally manual operations with an automated solution capable of allotting purchase orders, invoices, accounts receivables, and similar docs’ recognition and processing. The final product was to incorporate the following features:

Automate the process scanned docs received via email and BPM system
Classify and extract the required text, figures, and information items from every processed document
Register documents in the compony’s ERP system
Input recognized items from each doc in the corresponding section of the ERP

In a nutshell, the delivered RPA solution was to cater to various activities, such as inventory, supply chain reporting, returns processing, invoice and contract administration, and store planning management. It must maintain multiple office tasks for the finance department to optimize its activities around regulatory compliance, cashflow management, incentive claims, payables, and receivables.

Development Approach

To develop bots, we used Automation Anywhere platform while we had already had substantial expertise working with its functionality, and it turned out to be perfect for similar challenges. Plus, the low-code capabilities ensured reduced development time.

Our primary focus was on creating the machine learning model to be used in the IQ Bot’s doc scans’ recognition to achieve a satisfactory level of Straight Through Processing. There were five types of documents, meaning we had to create separate learning instances and train five models. In its turn, each model acquired its knowledgebase based on 500+ docs each. However, when we encountered low-quality hard-to-read scans, we had to implement some organizational changes. The first was to establish one more task-performer, the validator, and then teach it to verify IQ Bot’s actions. This way we enforced RPA with amazing AI capabilities, including Computer Vision, Natural Language Processing, and machine learning to analyze, select, and verify the data from processed records and messages.

The final RPA project was designed for front-end integration and did not require any essential alterations to its enduring IT infrastructure. It became a seamless part of the existing architecture and all the employees and subdivisions involved could reach the information at their convenience.


The final solution was fully compliant with the customer’s requirements. As its central part, the bot control hub was fit for management over all the bots in production. That included bot runners to run bots on virtual machines, bot creators to generate new bots, and the IQ Bot aimed at creating and training the ML models for scanned docs’ recognition and extraction of the necessary recognized text and figures.

OTR’s team automated the entire end-to-end process. It combined two separate tasks — automating repetitive operations with RPA and doc scan recognition and processing with the IQ BOT, within a single platform with no additional application integration required.

With automation, we eliminated the need to perform repetitive manual tasks, forming a substantial part of the company’s employees’ daily routines. The RPA solution’s introduction resulted in 31% decreased assistant labor cost while improving product categorization accuracy by an astonishing 98.5%. Other than this, the company managed to redirect its associate staff to address more challenging consumer-centric activities with no need to hire any new workforce.

Want to discuss a similar project? Drop Us a Line!

This website uses cookies to improve site functionality, provide you with a better browsing experience and proper advertising. By using this Site or clicking on “OK”, you consent to our Privacy Policy.