Using data science to come out of Covid-19 stronger than the competition

If it isn’t difficult enough emerging from the current crisis and recovering your business to pre-pandemic market trading levels, there is a new threat to your business. Some of the smartest companies, as well as the emergence of some new start-ups, are using data science to help them reduce cost, automate processes, make more intelligent decisions and therefore emerge from this situation stronger and fitter than before. If you don’t accelerate your data science capabilities, there’s a good chance you’ll be left behind post Covid-19. Here are 7 steps to ensuring your data project helps you compete:

1. Define your leadership hypothesis

The first thing you should arrive at is a Leadership Hypothesis — how do the leadership believe the use of data science helps you to steer your business through these troubled waters? The focus for most companies should be on the Four C’s: cash generation, customer retention and cost management, and care of employees.

Whilst not every organisation is in the same situation, these are the areas likely to be most critical to support a business to recover and grow quickly. Will it all be about reducing cost through automation? Or more intelligent pricing decisions? Or increased acquisitions through more impactful marketing? Charging ahead with data science and AI without a consideration of how this might impact the business and support you will likely not succeed.

2. Ensure your organisational culture is setting you up for success

Next, you need to start with the basics, and by this we mean start by changing your organisational culture to be more data savvy, whilst understanding the Leadership Hypothesis. To get the most value out of your data in order to help you compete, your organisation needs a clear leadership structure, defined and led by the person who leads your data initiatives — such as your chief data officer (CDO) — working directly with the board and senior executives.

Boards and execs need to start demanding more data from the business in order to make better decisions, and not just at C level, but throughout the organisation. This is where driving more value from data begins. Let’s face it, if the whole world can switch to working from home in less than a week, then instilling a data culture should be possible within a relatively short timeframe! The CDO should evangelise the opportunities for data science, explaining some of the best examples, and ways that data science can help. Coming out of this should be some knowledgeable business sponsors armed with their challenges.

Who hires the CDO — or does the chief data officer hire themselves?

Sooraj Shah asks who is in the unenviable position of having to hire for a completely new C-level position and whether they’re the right fit. Read here

3. Define your data science process

The next step is to enable business sponsors to know how to engage with data science. The business domain experts from across the business should come armed with their challenges. But to get the most out of data science capabilities, it’s important to have a clearly defined data science process so that the business can understand what help they can expect, where it will come from, and when.

It’s important to find a process that aligns to the culture and aspirations of the company and make sure it’s repeatable, scalable, and builds a common language across analytic and business functions. A good industry-wide process exists — it’s called the CRISP-DM life cycle (Chapman, Clinton, Kerber, et al, 1999). It was first set up for data mining, one aspect of data science, but can be applied to all. In this way, everyone knows the stages of the lifecycle and timescales, and will know how and when to engage in the process. This process should be owned by the head of data science or data science lead.

4. Challenge your challenge

Once a challenge has been identified, it should then be explored from a number of perspectives to ensure it is valuable, solvable and realisable. Many a data science project has failed before a line of code has been written by not asking the right question, whether the data is available, or whether the business is ready for the change an analytic approach will require. Once agreed, data science teams need to work collaboratively with business stakeholders to iteratively develop appropriate solutions rapidly and clearly. This will involve some experimentation against an agreed baseline so we can measure value and progress, and there will need to be rigorous testing and validation by the business.

5. Automate for quick access to data

Key to the success of the data science team is easy access to the data sets they need to perform science on. These need to be built, automated and deployed to an environment where this is possible. The vast majority of companies’ data sources that are valuable for generating value are within their existing structured systems, and the use of automated data warehousing tools will enable new data sets to be created quickly.

Can we automate data quality to support artificial intelligence and machine learning?

Clint Hook, director of Data Governance at Experian, looks at how organisations can automate data quality to support artificial intelligence and machine learning. Read here

6. Get the model outputs into production

Your organisation will then need to address one of the biggest challenges in data science, and the one that will inhibit your organisation getting value from data. That is getting the output from the models into production. Most organisations have failed to deal with this, and now is the time to deal with it head on. IT departments have many years’ experience getting systems through testing and into production. Rarely do you see IT and data science working together well, but this is an area where the CIO should take the lead, albeit that the IT teams will need to work in an agile way to ensure rapid deployment.

7. Privacy and security are key

Finally, we’ll be developing new models rapidly, and as we develop these new data capabilities, we should not forget the vulnerabilities and the unintended consequences of so doing. Protecting our data is not an option, it’s a necessity. Ensuring your data is secure, and that your models are transparent and ethical, matters now more than ever. Everyone, including the chief data officer, head of data science and the business sponsors will need to work closely with the data privacy officer (DPO), who will represent the interests of the customer.

These are challenging times for any organisation, but follow these steps and allow data science to unlock the key to recovery from Covid-19!

Written by Simon Asplen-Taylor, CEO of DataTick, and Rich Pugh, chief data scientist at Mango Solutions

Editor's Choice

Editor's Choice consists of the best articles written by third parties and selected by our editors. You can contact us at timothy.adler at stubbenedge.com