How Dentsu Built 60 Robots in 30 Days (And Saved 125k Hours in the Process)

Dentsu, a global digital marketing communications company, was faced with the seemingly impossible task of extracting and migrating more than 2.8 million records in a matter of weeks.

With deadlines looming on two massive data extraction projects, the team turned to Robotic Process Automation (RPA) to streamline and speed up these time-consuming processes.

Finding a faster, better way to handle data migration 

Migrating data from legacy systems to new platforms is often a tricky and painstaking process, but it’s especially difficult when you’re working on a deadline. And if you don’t have an Application Programming Interface (API) to export that information in bulk, you're likely looking at tens of thousands and in some cases even hundreds of thousands of hours to get the job done.

The latter is the kind of problem that the Automation Center of Excellence (CoE) at Dentsu Aegis Network had to solve in the first half of 2020. Did we mention they were on a deadline?

The CoE led by Max Cheprasov, Chief Automation Officer, is on a mission to elevate human potential by integrating AI and Automation into every business process. An early enthusiast of AI’s potential in business, Cheprasov embarked on his first automation-related project in 2016 by experimenting with natural language processing (NLP) and natural language generation (NLG) technology for data analysis reports and insights.

We can automate a process end-to-end and give that time back to employees to handle more critical tasks that cannot be automated, such as creative, critical, strategic thinking, complex problem solving and more. Our mission is to elevate human potential." 

Max Cheprasov • Chief Automation Officer

With the clock ticking, the duo of premier automation experts in the Automation CoE, Brian Klochkoff (Director of Automation Solutions) and Guilherme Gornati (Automation Product Manager), searched for a practical approach that wouldn’t need to involve hiring hundreds of temporary employees to pitch in and manage a manual copy-and-paste job.

Having been at the forefront of automation technologies since the Automation CoE was set up in 2017, Klochkoff and Gornati have delivered successful AI-powered solutions in the past and knew this was a perfect use case for the UiPath Platform. But this time, the sheer scale of this project, would demand more than just a handful of robots — it would need close to 60 bots that needed to be built in a very short timeframe.

The path forward was clear. The only question was, would they be able to scale up their RPA program’s development capabilities in time?

The two options we had were to either manually copy and paste and extract information—click by click, step by step (for months)—or develop the bots. Doing it manually in the short amount of time that we had would essentially require hiring a small army of people. And, deploying 60 robots in under 30 days was just unheard of in our industry. But we knew that our team had the skills and UiPath had the tech to take the challenge head on!

Max Cheprasov • Chief Automation Officer, Dentsu Aegis Network Americas

Teamwork makes the dream work

The data extraction and migration mandates were separated into two separate workstreams. On one side, the team needed to transfer the organization’s whole collection of client invoices from the past 10 years—roughly 1.8 million documents—from one system to another. On the other side, the team needed to extract all financial data from the legacy platform accounting for just over a million records. Both jobs had to be done quickly (and with 100% precision) without letting a single document, record, or data element slip through the cracks. Dentsu is required to keep a comprehensive archive of every client invoice for compliance purposes and the master data extraction was essential for business continuity in the new platform.

For the invoice extraction, the team estimated that doing this work manually would require 76,500 hours—the equivalent of 41 full-time employees working a full calendar year—to download each invoice and save it to an internal secure and organized storage location. On top of those logistical and cost concerns, there was also the high risk of human error to consider. A single mistake could compromise Dentsu’s compliance standing.

“The data that was being extracted was not only highly sensitive, but also essential for business continuity. It needed to get extracted and migrated perfectly.”                             

Brian Klochkoff, Director of Automation Solutions

You can learn more from Brian in this video where he discusses how Dentsu uses computer vision to scale their global partnerships.

For the master data extraction, the team needed to migrate more than 1 million records accounting for over 32 million data values spread out across 17 different processes and applications within a matter of weeks so that the new platform could be comprehensively tested, go through release cycles, and ultimately transition/launch without any interruption to business continuity. The stakes were especially high given that this data is the foundation of Dentsu’s business, and any error would result in direct impact to client team operations.

Assuming a person could extract one record in three minutes, it would take about 50,000 consecutive hours to complete this phase of the project, not accounting for quality assurance checks and rework. Manually copy-pasting 1 million records accounting for 32 million values creates an incredibly high margin of error which was mitigated by a high caliber project team powered by RPA.

This isn’t a case study in data migration, this is a case study in elevating human potential through technology and partnership. The Professional Services team at UiPath went above and beyond expectations to become an organic extension of our CoE and the larger project team. Each team member contributed and gave us the additional expertise and support needed to ensure an agile yet smooth development process within a tight timeline even as requirements evolved. Together, we enhanced traditional data migration with an incredibly lean team and delivered a solution that enabled our people to do what they do best which exceeded our leadership’s highest expectations!

Brian Klochkoff • Director of Automation Solutions, Dentsu Aegis Network Americas

Rather than spend all of that time and money on error-prone manual processes, Dentsu’s Automation CoE embarked on an ambitious RPA project. By automating over a dozen different workflows that would be executed by 60 unattended robots Dentsu tackled both sides of the project. Having worked with UiPath in the past, Klochkoff was confident that the UiPath Professional Services Group could develop against Klochkoff’s solution design, provide insight to the project as technical obstacles arose, and help to scale up the organization’s RPA program to meet the aggressive timeline.

With the Dentsu and UiPath teams working together, they were able to set up the necessary infrastructure and build the whole array of bots within six weeks. Not only did this approach save on project costs and act as a catalyst for the timeline, but it also enabled teams within Dentsu’s larger transformation initiative to focus on their respective primary aims such as change management, training, testing, process re-engineering, etc. instead of being pulled into ancillary tasks like data QA as an output from data extraction and migration.

Dentsu finishes year long project in 6 weeks

Deploying an army of bots allowed Dentsu to automate what would have taken 67 full-time employees a year to do—than complete the master data extraction piece in the required six weeks it would have required over 300 new hires to get the master data extraction piece done in. Looking at the average handling time per transaction, the team reduced processing time for each transaction by 90% from 3 minutes to just 18 seconds.

All told, the organization saved itself hundreds of thousands of hours by turning to RPA for assistance. That doesn’t even take into account the regulatory compliance headaches and business continuity operations interruptions the team avoided by moving away from error-prone manual processes.

Anyone familiar with RPA has heard about its cost-saving benefits a hundred times before, but the real story here is the scale and speed of this project. Building 60 unattended robots in 30 days and completing a massive data migration initiative that involves millions of records should be impossible. But it’s not, thanks to RPA.

Max Cheprasov sums it up: "the partnership with UiPath allows us to tackle challenges that otherwise would be humanly impossible to achieve."