Accenture’s Penelope Prett on the predictiv e value of data

We have data that goes back to our instantiation as a company. As we’ve moved into a cloud-based position, where we can consume functionality from our vendor partners, we’ve been cleaning up every individual data domain.

But the value does not exist in that data. The value exists in the threads that bind together different data domains — such as finance and HR — to represent how our company operates. Value comes from the complex products that represent the interactions of multiple types of data.

Value also exists in the data’s predictive capability. In today’s world, we operate through spreadsheets, PowerPoints, and people’s opinions. Those people are smart and experienced, but they approach data through their own lens. By lifting the data into a more aggregated, automated mode, we can look further into history and predict further into the future than humans can on their own. Value exists in the predictive power of data aggregation.

What is something you can predict now that you couldn’t in the past?

At Accenture, we used to manually gather global macroeconomic data, apply that external data to our internal data, identify patterns, and make decisions. But today, we can connect into third-party global cloud-based capabilities and obtain that macroeconomic data in an automated fashion. We can very quickly spot global trends that will impact our business, region by region, months earlier and with no human filter. This is truly a revolutionary time for companies to up their performance game by harnessing the power of this kind of data, but they must act fast.

Why do they have to act fast?

The field of data is so vast, that CIOs and chief data officers can anguish about what to spend and where to start. But every day that goes by where you do nothing, a competitor is building value from data. Whether it’s a small startup or a large enterprise, someone is making progress in using the data available in your industry for better predictions and performance. Every day, your competitors’ models become smarter.

It’s a compounding problem. The longer you wait, the further behind you get, until there is a point where can’t catch up, even with today’s leapfrogging technologies.

My advice is just to start. You have reams of data. Start with a very small domain, parse and examine it, and think about how you can combine it with other data. Play with the technology. Get into the fray.

How did Accenture get started on this data value journey?

Accenture’s Chief Operating Officer Manish Sharma commissioned an initiative called the Rapid Data Lab. He challenged a small, focused team to spend six months to select our three most important data processes, such as a monthly review of our global profitability, and create the best flow and visualization from the data. This data had been held in PowerPoint decks and had no predictive element. He then asked the leadership team to conduct our meetings using the data flow, following where the data takes us, not what the team wants to talk about.

After six months, our whole company fell in line, because when you create small capabilities that show the power of data, it is addictive to the population. I guarantee that every company has enough data under the hood to create one of these capabilities.

We called the Rapid Data Lab ‘Velocity One,’ and then Manish asked me to create a plan for ‘Velocity Two,’ a permanent, enduring capability to embed the power of data into our company’s way of life. That’s what I’m doing now.

What is your advice regarding the architecture of a data strategy?

The rate of technology evolution is particularly acute in the data and AI space right now, so selecting technology can be intimidating. But basically, you will need a major platform for your data products layer, which is where you generate, store, and handle the data. Then you need a data mesh to manipulate the data and get it to a state where you can send it to the visualization layer. Wrapped around all of that is a technology toolset that governs the data.

Your choices in each of these four layers — storage, mesh, visualization, and governance — are infinite, but if you understand what you want to do with each layer, the path to the right technology becomes much easier.

How are you democratizing the data?

I seek to push out to my business partners as much data as they are willing to take on. My role is to create a safe sandbox, and their job is to build whatever they want within that sandbox that best serves our business.

Democratization at its best is when everybody can live their best selves with the data that you give them. But you don’t want people to create their own data, so it is important to create ‘shopability’ in the model, so your business customers get what they need from the data you provide and do not generate their own core data. When everyone uses the same core data, the only thing that differs is the interpretation of certain layers of the data. Problems occur when we source data from different places. This is where I am concentrating right now, working in tandem with my business customers to give them the data that they want, and not get in the way of how they use it.

What will be the evolution of driving value from data?

Today, we use data to think about what is next for our businesses. If we apply technology properly, we will soon reach a point where the intelligence of the technology will ask us, ‘Have you thought about this? Have you considered that?’ Data technology will reach beyond what humans bring to the picture. We all operate from our own contextual lens, which will be an additive component at the table. But the data will ask us questions that allow us to maximize what our human brains can do.